Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2021 Aug 6;16(8):e0254597. doi: 10.1371/journal.pone.0254597

Whole brain and deep gray matter structure segmentation: Quantitative comparison between MPRAGE and MP2RAGE sequences

Amgad Droby 1,2,3,*, Avner Thaler 1,2,3, Nir Giladi 1,2,3, R Matthew Hutchison 4, Anat Mirelman 1,2,3, Dafna Ben Bashat 2,3,5, Moran Artzi 2,3,5
Editor: Joseph Najbauer6
PMCID: PMC8345829  PMID: 34358242

Abstract

Objective

T1-weighted MRI images are commonly used for volumetric assessment of brain structures. Magnetization prepared 2 rapid gradient echo (MP2RAGE) sequence offers superior gray (GM) and white matter (WM) contrast. This study aimed to quantitatively assess the agreement of whole brain tissue and deep GM (DGM) volumes obtained from MP2RAGE compared to the widely used MP-RAGE sequence.

Methods

Twenty-nine healthy participants were included in this study. All subjects underwent a 3T MRI scan acquiring high-resolution 3D MP-RAGE and MP2RAGE images. Twelve participants were re-scanned after one year. The whole brain, as well as DGM segmentation, was performed using CAT12, volBrain, and FSL-FAST automatic segmentation tools based on the acquired images. Finally, contrast-to-noise ratio between WM and GM (CNRWG), the agreement between the obtained tissue volumes, as well as scan-rescan variability of both sequences were explored.

Results

Significantly higher CNRWG was detected in MP2RAGE vs. MP-RAGE (Mean ± SD = 0.97 ± 0.04 vs. 0.8 ± 0.1 respectively; p<0.0001). Significantly higher total brain GM, and lower cerebrospinal fluid‏ volumes were obtained from MP2RAGE vs. MP-RAGE based on all segmentation methods (p<0.05 in all cases). Whole-brain voxel-wise comparisons revealed higher GM tissue probability in the thalamus, putamen, caudate, lingual gyrus, and precentral gyrus based on MP2RAGE compared with MP-RAGE. Moreover, significantly higher WM probability was observed in the cerebellum, corpus callosum, and frontal-and-temporal regions in MP2RAGE vs. MP-RAGE. Finally, MP2RAGE showed a higher mean percentage of change in total brain GM compared to MP-RAGE. On the other hand, MP-RAGE demonstrated a higher overtime percentage of change in WM and DGM volumes compared to MP2RAGE.

Conclusions

Due to its higher CNR, MP2RAGE resulted in reproducible brain tissue segmentation, and thus is a recommended method for volumetric imaging biomarkers for the monitoring of neurological diseases.

Introduction

Magnetic resonance imaging (MRI) is a commonly used method for investigating the neural hallmarks and the monitoring of many neurologic and neuropsychiatric diseases [1]. Assessing the whole brains’ grey matter (GM) and white matter (GM) tissue, as well as deep grey matter (DGM) structures, can be of high importance for the diagnosis and therapy selection in many neurological and neuropsychiatric conditions [2]. This can be achieved mainly by segmenting the brain structures based on T1-weighted, T2-weighted, proton density (PD), magnetic transfer (MT), or multi-contrasts [3,4].

Magnetization Prepared Rapid Gradient Echo (MP-RAGE) [5] is the predominantly used sequence for obtaining 3D T1-weighted MR images of the human brain [3]. MP-RAGE consists of a non-selective (180°) inversion pulse, which inverts the net magnetization (M), allowing M to regrow via T1 relaxation mechanisms over the inversion time interval (TI), during which, the signal is acquired using a spoiled gradient echo (GRE) (Turbo-FLASH) with a low flip angle.

At high (1–3 Tesla) and ultra-high (≥7 Tesla) magnetic fields, the increased in-homogeneities of transmitting and receiving magnetic fields (B1) create intensity variations in the acquired images leading to reduced contrast, affecting segmentation results and accuracy of tissue -classifications [3]. This may even be more pronounced in DGM which have shorter T1 values compared to higher cortical regions, leading to decreased tissue contrast between GM and neighboring white matter (WM) [6,7]. Such classification in-accuracies could result in poorer sample size calculations and decreased power in clinical trials.

MP2RAGE is becoming readily available across several MR vendors. The sequence is a variation of the MP-RAGE, which utilizes two Turbo-FLASH GRE readouts acquiring 2 volumes after each inversion pulse. The first TI (TI1 ≈ 700ms) produces a T1-weighted image with the gray matter nulled at the center of k-space, and the second TI (TI2 ≈ 2500ms), combined with small FAs (α = 4–5°), and long TR (~5000ms), produces a second image with spin-density-weighted contrast. By combining both images from the first and second readouts, T2* and B1 inhomogeneity effects are largely canceled out resulting in a T1-weighted unified (UNI) image with superior GM-to-WM contrast compared to MP-RAGE, allowing a reproducible quantification of DGM structures volumes across subjects [8,9].

Several studies applying MP2RAGE have reported enhanced contrast-to-noise ratio (CNR), as well as a good inter-and-intrasubject agreement between the acquired T1-maps [3,10], yielding larger measured volumes for several DGM structures, as well as more sensitive disease burden markers in neurological conditions, were obtained when assessed based on MP2RAGE images compared to MP-RAGE [9,11]. However, the quantification of tissue volumes does not rely on the sequence acquisition parameters alone. Factors related to the applied algorithm used for segmentation also contribute to the reliability of the obtained volumetric values [12,13]. Therefore, in this study, we aimed to systematically examine the patterns of discrepancies and quantitatively assess the agreement between tissue classification obtained by different widely available, automatic segmentation tools relying on different algorithms that are commonly used for the assessment of whole-brain tissue, as well as DGM volumes based on MP2RAGE compared to MP-RAGE images.

Materials and methods

Study participants

This study was part of the ongoing BEAT-PD study taking place since 2017 at the Tel Aviv Sourasky Medical Center (TASMC). Data from 29 healthy participants (19 females, mean age ± standard deviations (SD) = 52.4 ± 10.25 years), with no history of an outstanding neurological or psychiatric disease, were included in this study. Exclusion criteria were as follows: (i) diagnosed neurological or psychiatric disorder, (ii) a malignancy, (iii) HIV, HBV, or HCV positive, (iv) MRI-related contraindications. The study was approved by the institutional review board (IRB) at the Tel Aviv Sourasky Medical Center (TLVMC). All enrolled participants gave their informed written consent before participation.

Test-retest reliability: Sub-study

Twelve participants (6 females, mean age ± SD = 51.5 ± 6.5 years) were followed up after 1 year and re-scanned using the same MRI protocol and the same MRI system as described below. The acquired MP-RAGE and MP2RAGE images from these participants were segmented by the different automatic segmentation methods, and the whole brain tissue, as well as DGM volumes, were assessed.

MRI acquisition protocol

MR data were acquired using a 3 Tesla Magnetom Prisma® (Siemens, Erlangen, Germany) MR scanner, and a 20-channels phased-array head coil. The MRI protocol included a high-resolution 3D T1-weighted MP-RAGE and 3D MP2RAGE sequences. See Table 1 for detailed sequences parameter and Fig 1A for examples illustrating MRI images acquired using both sequences.

Table 1. Acquisition parameters of magnetization prepared- rapid acquisition gradient echo (MP-RAGE) and MP2RAGE sequences.

MP-RAGE MP2RAGE
Field of view (FoV) 256×256 mm 256×256 mm
Number of slices 192 176
Repetition time (TR) 2200 ms 5000 ms
Echo time (TE) 3.22 ms 3.43 ms
Inversion time (TI)
    • TI1 1100 ms 803 ms
    • TI2 2500 ms
Flip angle (FA)
    • FA 1 9 deg. 4 deg.
    • FA 2 5 deg.
Voxel size 1×1×1 mm 1×1×1 mm
Readout bandwidth 150 Hz/Px 240 Hz/Px
Parallel acquisition GRAPPA, factor 2 GRAPPA, factor 3
Duration 5:06 min 7:07 min
Open in a new tab

Fig 1.

Fig 1

Open in a new tab

(A) Example of the acquired T1 weighted MP-RAGE and MP2RAGE images from one of the study participants. (B) Brain extraction before tissue segmentation. (C) Illustration of the tissue probability maps achieved by the used segmentation tools.

MRI data processing

All described analyses were performed based on the acquired MP-RAGE and the MP2RAGE derived T1-weighed (UNI) images of all participants. These included CNR between GM and WM tissues (CNRWG), whole-brain tissue, and DGM segmentation.

Whole brain tissue segmentation

Brain tissue segmentation was performed using free and widely used automatic segmentation tools that rely on different segmentation algorithms including: (1) the computational analysis toolbox (CAT12, http://www.neuro.uni-jena.de/cat/) extension for statistical parametric mapping (SPM12, https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) that uses a-priori tissue probability maps, to perform an initial skull-stripping, and to initialize the segmentation. Then, it uses an adaptive maximum a posteriori segmentation approach (hypothesis-free approach) [14]. (2) volBrain processing pipeline (https://volbrain.upv.es/instructions.php) that performs non-local intracranial cavity extraction based on a library of pre-labeled brain images allowing to capture the large variability of brain shapes [15,16]. After brain extraction, mean values of the different tissue types are estimates by trimmed mean segmentation (TMS), excluding partial volume voxels from the estimation jointly with the use of an unbiased robust mean estimator [16]. (3) FMRIB’s Automated Segmentation Tool (FSL-FAST, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FAST), which is based on a hidden Markov random field model and an associated Expectation-Maximization algorithm for tissue segmentation [17] before image segmentation, bias field correction embedded in the different used pipelines and brain extraction were performed on all images. These segmentation processes yielded GM, WM, and cerebrospinal fluid (CSF) probability maps (See Fig 1B & 1C). The segmented probability maps achieved from the used segmentation tool were then registered to standard space (Montreal Neurological Institute, MNI) using linear image registration tool and affine transformation (12 parameters model) to achieve spatial overlap between subjects and further voxel-based morphometric comparisons (VBM).

CNRWG

CNR was calculated on the tissue WM and GM (CNRWG) probability maps extracted from the acquired raw T1 images. The extracted WM and GM probability maps were thresholded at 0.9, thus reflecting only intersect areas between segmentation tools, as well as minimizing the partial volume effect. CNR between the WM and DGM (CNRWdG) structures was defined based on the extracted WM and DGM mask FIRST (see deep grey matter segmentation). Moreover, CNRWG differences between both sequences were investigated sequences per lobes as well. The CNRWG and CNRWdG values were calculated as the difference between the mean values of the two tissues normalized for scan time as described in Haast et al., 2016 [18] (calculated separately for the MP-RAGE and MP2RAGE).

Deep grey matter (DGM) segmentation

DGM segmentation was performed using: (1) FSL FIRST segmentation tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FIRST), which is based on Bayesian statistical models of shape and appearance for subcortical brain segmentation [19] (2) FreeSurfer (https://surfer.nmr.mgh.harvard.edu/) that segments gray-white matter into different cortical regions based on sulci and gyri, and also segments major subcortical regions in the brain. FreeSurfer provides a full processing stream for structural MRI data, including Skull stripping, B1 bias field correction, and gray-white matter segmentation [20], and (3) volBrain labeling processing pipeline that applies novel patch-based method relying on expert manual segmentations as priors [21], producing a total of 14 DGM segments: the left and the right thalamus, caudate, putamen, pallidum, hippocampus, amygdala, and accumbens.

Statistical analysis

VBM comparisons were performed using SPM software (SPM12; https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). For the group-level voxel-wise comparisons, paired-samples t-tests were used for inter-modality comparisons (i.e, MP-RAGE vs. MP2RAGE) of whole-brain non-binary masked GM and WM tissue probability maps obtained from the different segmentation tools at baseline. Similarly, paired-samples t-tests were used to investigate the overtime changes in both tissue probability maps obtained from the different segmentation tools in the scan-rescan sub-study after 1 year. P < 0.05 FWE was adopted for all voxel-wise comparisons. Using SPSS® v22 (IBM, Chicago, USA), paired-samples T-tests were used to test differences between the whole-brain tissue volumes, as well as volumes of DGM structures inter-modality comparisons. Due to the smaller sample size in the scan re-scan sub-group who underwent MRI assessment after one year, Wilcoxon’s-rank test was used to test for significant differences in the whole-brain tissue volumes, as well as volumes of DGM structures. Finally, as measures for reliability, the mean percentage change, Cronbach’s Alpha, as well as paired-samples T-test were calculated to explore the difference in tissue volumes assessed by the various methods based on both MRI sequences in the test-retest sub-study.

Results

CNRWG

Significantly higher mean CNRWG was detected for the MP2RAGE vs. MP-RAGE for (Mean±SD = 0.97 ± 0.04 and 0.8 ± 0.1 respectively; paired-samples T-test, t(df = 28) = 15.5, p<0.0001). Moreover, MP2RAGE showed significantly higher CNRWdG compared to MP-RAGE (Mean±SD = 0.3 ± 0.04 and 0.05 ± 0.34 respectively; paired-samples T-test, t(df = 28) = -4, p<0.0001). The calculated CNR values per lobe based on MP-RAGE showed that the range of variation of CNR relative to the group mean in MPRAGE was found to be higher (9–13%) compared to MP2RAGE (1.5–1.7%).

Whole-brain volumes

Significantly higher total brain GM volumes were obtained in MP2RAGE using the three segmentation tools (paired-samples T-test, t(df = 28)<-11, p<0.001 in all cases). MP2RAGE yielded significantly higher WM total brain tissue volume in comparison to MP-RAGE when segmented using CAT-12 (paired-samples T-test, t(df = 28)>6.4, p<0.001). However, significantly lower whole-brain WM volumes were obtained from MP2RAGE using FAST and volBrain segmentation compared to MP-RAGE (paired-samples T-test, t(df = 28) = 4.9, p<0.001 in both cases). Finally, in all three segmentation tools, significantly lower CSF volumes were obtained from MP2RAGE in comparison to MP-RAGE (paired-samples T-test, t(df = 28)>3.55, p<0.001 in all cases) (Table 2 & Fig 2).

Table 2. Whole-brain GM, WM, and CSF volumes obtained from the segmentation of MP-RAGE and MP2RAGE images using the different automatic segmentation tools.

MP-RAGE MP2RAGE
CAT-12 GM** 595.69 ± 51.2 622 ± 58.59
WM** 509 ± 45.25 509 ± 46.82
CSF** 314.4 ± 61.93 299.2 ± 54.44
FSL-FAST GM** 585.95 ± 57.18 699.39 ± 67.46
WM** 513.06 ± 50 487.53 ± 45.67
CSF** 408.38 ± 62.67 336.27 ± 51.62
volBrain GM** 658.9 ± 53.29 715.96 ± 56.98
WM** 524.41 ± 50.32 497.55 ± 52.97
CSF** 205 ± 50.7 183.22 ± 43.03
Open in a new tab

All values are reported in mean ± SD.

** Paired-samples T-test, p < 0.001. Abbreviations; GM: Grey matter, WM: White matter, CSF: Cerebrospinal fluid.

Fig 2.

Fig 2

Open in a new tab

Scatter plots demonstrating GM (blue), WM (red), and CSF (green) volumes measured by the different segmentation tools based on MP-RAGE and MP2RAGE sequences. Points above the identity line (dashed) indicates higher volumes were calculated in MP-RAGE vs. MP2RAGE, while points below the line indicate higher values based on. MP2RAGE vs. MPRAGE.

Inter-modality voxel-wise differences

Fig 3A demonstrates the voxel-wise and whole-brain differences in tissue class classification between MP2RAGE and MP-RAGE obtained from the three segmentation tools. MP2RAGE showed significantly higher GM tissue probability in several regions compared to MP-RAGE including the bilateral thalamus, putamen, and caudate nucleus, the lingual gyrus, the inferior fronto-orbital gyri, and bilateral precentral gyrus when assessed by both CAT12 and volBrain (pFWE<0.05). In WM, MP2RAGE showed significantly higher tissue probability in the cerebellum, frontal, temporal regions, as well as in the genu and splenium of the corpus callosum (CC) in comparison to MP-RAGE. On the other hand, MPRAGE showed higher WM densities in the bilateral thalamus, putamen and precentral gyrus (PCG) compared to MP2RAGE using CAT12 (pFWE < 0.05, in all cases). In FAST, MP2RAGE showed an overall significantly higher GM and WM probabilities compared to MP-RAGE. Finally, MP-RAGE showed significantly higher WM within the anterior cerebellar lobe, and bilateral putamen (pFWE < 0.05). No significant differences between both sequences were detected in WM when assessed by volBrain.

Fig 3.

Fig 3

Open in a new tab

(A) Voxel-wise grey matter (GM) and white matter (WM) tissue probabilities differences when comparing MP-RAGE vs. MP2RAGE sequences based on tissue probability maps obtained by CAT12, FSL-FAST, and volBrain (pFWE < 0.05).

In the-sub study, significantly higher GM probability was found in the left thalamus in MP2RAGE vs. MP-RAGE when assessed using CAT12. In WM, MP2RAGE showed significantly higher probability in the splenium compared to MP-RAGE (pFWE<0.05 in all cases). However, MP-RAGE showed a significantly higher WM probability in the red nucleus. No significant inter-modality differences were detected between both sequences when assessed by FAST and volBrain.

Deep grey matter (DGM) volumes

In the three segmentation tools, MP2RAGE yielded higher mean volumes of the segmented DGM structures compared to MP-RAGE. Specifically, using FIRST, significantly higher volumes of the left putamen, left hippocampus, bilateral amygdala, accumbnes, right caudate, and right pallidum was obtained in MP2RAGE compared to MP-RAGE (paired-samples T-test, t(df = 28)<-1.65, p<0.05 in all cases) (Table 3A). In volBrain, significantly higher volumes were obtained for bilateral thalamus, putamen, pallidum, the left hippocampus, and right accumbens in MP2RAGE vs. MP-RAGE (paired-samples T-test, t(df = 28)<-2.82, p<0.05, in all cases). However, MP-RAGE yielded a significantly higher volume of the left caudate (paired-samples T-test, t(df = 28) = 3.22, p = 0.003) compared to MP2RAGE (Table 3B). Finally, using Freesurfer, significantly higher volumes of bilateral putamen, right pallidum, and left accumbnes were obtained from MP2RAGE compared to MP-RAGE (paired-samples T-test, t(df = 28)>4.09, p<0.05, in all cases). Moreover, based on MP-RAGE, Freesurfer yielded significantly higher volumes of bilateral thalamus, hippocampus, amygdala, left caudate, and left pallidum were obtained by Freesurfer compared to MP2RAGE (paired-samples T-test, t(df = 28)< -0.61, p<0.05, in all cases) (Table 3C).

Table 3. Measured deep grey matter (DGM) structures volumes using the different segmentation tools based on MP-RAGE and MP2RAGE images.

Method DGM structure MP-RAGE MP2RAGE p- value
A. FIRST L Thalamus 7.87 ± 0.66 7.9 ± 0.64 n.s
L Caudate 3.40 ± 0.4 3.38 ± 0.4 n.s
L Putamen 4.84 ± 0.5 5.0 ± 0.54 ≤ 0.001
L Pallidum 1.76 ± 0.26 1.8 ± 0.24 n.s
L Hippocampus 3.66 ± 0.46 3.9 ± 0.55 ≤ 0.001
L Amygdala 1.21 ± 0.27 1.3 ± 0.27 ≤ 0.01
L Accumbens 0.5 ± 0.12 0.55 ± 0.1 ≤ 0.001
R Thalamus 7.72 ± 0.71 7.7 ± 0.74 n.s
R Caudate 3.41 ± 0.4 3.56 ± 0.41 ≤ 0.05
R Putamen 5.0 ± 0.55 4.94 ± 0.59 n.s
R Pallidum 1.82 ± 0.23 1.87 ± 0.23 ≤ 0.05
R Hippocampus 3.72 ± 0.49 3.96 ± 0.55 n.s
R Amygdala 1.28 ± 0.22 1.51 ± 0.22 ≤ 0.001
R Accumbens 0.41 ± 0.1 0.45 ± 0.08 ≤ 0.01
B. volBrain L Thalamus 5.58 ± 0.47 5.68 ± 0.39 ≤ 0.001
L Caudate 3.41 ± 0.41 3.4 ± 0.42 ≤ 0.01
L Putamen 4.12 ± 0.43 4.17 ± 3.6 ≤ 0.001
L Pallidum 1.16 ± 0.13 1.26 ± 0.14 ≤ 0.001
L Hippocampus 3.76 ± 0.45 3.73 ± 0.4 ≤ 0.01
L Amygdala 0.8 ± 0.11 0.8 ± 0.12 n.s
L Accumbens 0.33 ± 0.06 0.33 ± 0.05 n.s
R Thalamus 5.56 ± 0.5 5.61 ± 0.42 ≤ 0.001
R Caudate 3.41 ± 0.41 4.44 ± 5.26 n.s
R Putamen 4.1 ± 0.45 4.1 ± 0.34 ≤ 0.001
R Pallidum 1.18 ± 0.13 1.27 ± 0.14 ≤ 0.001
R Hippocampus 3.73 ± 0.66 3.71 ± 0.77 n.s
R Amygdala 0.81 ± 0.12 0.81 ± 0.12 n.s
R Accumbens 0.3 ± 0.06 0.31 ± 0.05 ≤ 0.01
C. FreeSurfer L Thalamus 7.09 ± 0.7 6.73 ± 0.67 ≤ 0.001
L Caudate 3.52 ± 0.44 3.34 ± 0.4 ≤ 0.001
L Putamen 5.29 ± 0.53 5.43 ± 0.66 ≤ 0.01
L Pallidum 1.66 ± 0.17 1.8 ± 0.22 ≤ 0.001
L Hippocampus 4.17 ± 0.44 3.84 ± 0.44 ≤ 0.001
L Amygdala 1.57 ± 0.22 1.43 ± 0.24 ≤ 0.001
L Accumbens 0.51 ± 0.1 0.55 ± 0.1 ≤ 0.05
R Thalamus 7.24 ± 0.74 6.68± 0.64 ≤ 0.001
R Caudate 3.43 ± 0.44 3.44 ± 0.42 n.s
R Putamen 5.0 ± 0.55 5.21 ± 0.58 ≤ 0.001
R Pallidum 1.49 ± 0.16 1.64 ± 0.16 ≤ 0.001
R Hippocampus 4.31 ± 0.42 4.02 ± 0.41 ≤ 0.001
R Amygdala 1.6 ± 0.2 1.5 ± .21 ≤ 0.001
R Accumbens 0.58 ± 0.09 0.59 ± 0.08 n.s
Open in a new tab

All values reported in mean ± SD. Abbreviations: L: Left, R: Right.

Test-retest reliability: Sub-study

At re-scan, MP2RAGE demonstrated higher whole-brain GM, lower WM, and CSF volumes compared to MP-RAGE in all three used tools. In the used automatic segmentation methods a higher mean percentage of change was observed in total brain GM assessed based on MP2RAGE compared to MP-RAGE. On the other hand, MP-RAGE demonstrated a higher overtime percentage of change in WM and DGM volumes compared to MP2RAGE. Across all used segmentation methods, high scan-rescan reliability was obtained in both sequences as reflected by Cronbach’s α coefficients (Table 4). No significant differences were observed in the obtained whole-brain and DGM volumes based on both sequences from all used tools (p>0.05; in all cases). In VBM, significantly higher GM probability was found in the left thalamus in MP2RAGE vs. MP-RAGE when assessed using CAT12). In WM, MP2RAGE showed a significantly higher probability in the splenium compared to MP-RAGE (pFWE<0.05). Additionally, MP-RAGE showed significantly higher WM probability in the red nucleus. No significant inter-modality differences were detected between both sequences when assessed by FAST and volBrain (Fig 4).

Table 4. Test-re-test of the mean percentage of overtime change and Cronbach’s α in whole-brain GM and WM, as well as deep grey matter (DGM) volumes assessed using the different automatic tools based on MP-RAGE and MP2RAGE.

MP-RAGE MP2RAGE
Whole brain CAT-12 GM 1.17±0.33/0.99 2.3±0.76/0.98
WM 0.78±0.18/0.99 1.14±0.30/0.99
FSL-FAST GM 4.02±1.18/0.92 3.56±0.91/0.97
WM 6.07±1.92/0.90 7.14±1.96/0.90
volBrain GM 1.51±0.38/0.99 2.1±0.65/0.94
WM 1.7±0.29/0.99 6.18±1.92/0.92
DGM FSL-FIRST 5.08±2.30/0.96 3.32±1.76/0.96
Freesurfer 4.44±2.28/0.94 4.13±1.90/0.96
volBrain 1.03±2.53/0.94 3.20±3.76/0.93
Open in a new tab

All values are reported in mean ± SE/ Cronbach’s α.

Fig 4. Sub-study VBM results of inter-modality differences in grey matter (GM) and white matter (WM) tissue probabilities differences based on tissue probability maps obtained by CAT12, FSL-FAST, and volBrain.

Fig 4

Open in a new tab

Discussion

We compared the results of brain tissue segmentation performed on two MRI sequences (MP2RAGE vs. MP-RAGE) using different automatic segmentation tools. Based on the obtained results, MP2RAGE yielded higher CNRWG CNRWdG, and consequently, resulted in higher whole-brain GM and DGM, as well as lower CSF volumes when compared to MP-RAGE.

Compared to MP-RAGE, significantly higher CNR in whole-brain and DGM (80% & 16% respectively), as well as lower within-group variance were calculated in MP2RAGE images based on the segmented tissue probability maps. This can be explained by the differences between the two sequences, and the fact that increased rates of bias-field effects are inherent in MP-RAGE images at high and ultra-high magnetic fields (≥3 Tesla), whereas, MP2RAGE is less sensitive to B1 bias, enabling the acquisition of images with enhanced CNR [3,9,14]. Additionally, MP2RAGE uses a higher bandwidth compared to MPRAGE (240 Hz/Px vs. 150 Hz/Px). This use of high bandwidth is expected to reduce susceptibility effects including eddy currents associated with metal, and thus improve image SNR [3].

More accurate classification of tissue boundaries especially in GM and DGM due to the enhanced image quality of MP2RAGE images was observed, especially using tools that rely on edge detection such as CAT-12 and volBrain. This was evident from the voxel-wise comparisons where higher GM probabilities in MP2RAGE were observed mainly in several ventral brain regions, including the thalamus and putamen, as well as cortical sulci, all of which suffer low-degree of tissue boundary detection accuracy [12,22,23]. This finding was further validated by the calculated whole-brain volumes, where significantly higher whole-brain GM and lower CSF volumes were obtained based on MP2RAGE images compared to MP-RAGE. Image quality on an image plays a central role in the segmentation and quantitative assessment of the brains’ tissue [3,24]. Yet, unified MP2RAGE images contain a noisy background, which might affect the accuracy of tissue segmentation [25,26]. This can be overcome using brain extraction and skull stripping algorithms that can be applied prior to segmentation [8,27].

CAT12 and volBrain yielded comparable results in the voxel-based as well as in whole-brain tissue volume comparisons in the cross-sectional VBM analysis. Both of these approaches apply prior spatial information for each tissue class based on the standard MNI template [21,28,29]. Whereas by relying on signal intensity within each voxel, FSL-FAST applies a discrete Markov random field (MRF) to obtain prior spatial information for each voxel, and then iteratively updates its assignment to the correct tissue class [17]. This explains the obtained inter-modality voxel-wise differences using FAST segmentation, where due to its’ higher CNR, higher GM probabilities were observed in MP2RAGE, suggesting that FAST might be advantageous for segmenting low-quality images, as it is designed to be robust against lower signal-to-noise ratio (SNRs) [14,17]. Meanwhile, due to the enhanced CNR of MP2RAGE, automatic segmentation tools that rely on boundary detection as the basis for tissue classification would yield a more accurate estimation of volumetric and thickness measures [30,31]. This is further implicated by the findings of the sub-study inter-modality VBM analysis, in which significant GM and WM differences between both sequences were observed only in CAT12, illustrating the robustness of the segmentation algorithms implemented in FAST and volBrain compared to CAT12.

Automatic DGM segmentation tools often yield inconsistent results as these brain structures have shorter T1 values and decreased contrast between subcortical GM and the neighboring WM structures [32]. Considering that the used segmentation methods were developed and applied mainly for the processing of T1-weighted images (MP-RAGE, SPGR), comparable DGM volumes were obtained from MP-RAGE using the three tools. However, by faster evaluation of T1 values [3,33], MP2RAGE overcomes T1 decay and consequently resulting in better CNR in these regions [9]. Indeed, higher tissue probabilities of the putamen, caudate, and thalamus boundaries were observed using CAT12 and volBrain based on the voxel-wise comparisons. Moreover, significantly higher volumes were obtained based on MP2RAGE compared to MP-RAGE, mainly when using FIRST and volBrain. This is in line with previously reported findings by Okubo et al., where higher DGM probabilities were detected in MP2RAGE vs. MP-RAGE in several structures including the putamen, caudate, thalamus, and substantia nigra [9]. Conversely, using Freesurfer, higher volumes in several structures such as the hippocampus and the amygdala were obtained with MP-RAGE compared to MP2RAGE. This could be attributed to the poor delineation of these subcortical structures due to its basal location in the brain and the reduced SNR in MP-RAGE as a result [12]. Such observed discrepancies further emphasize that pooling together and interpreting the findings of the various studies, where different segmentation tools have been implemented can be limited and should be taken with caution [32].

Due to the lack of a ’’gold standard’, we performed a test-retest analysis based on a sub-group of participants who were scanned after one year. Typically, manual segmentation is performed by an experienced operator. However, applying this method for both whole-brain and DGM segmentation is labor-intensive and is prone to tracer bias [23]. Based on the sub-study, varying mean percentage of change and within-group variance in the measured whole brain, as well as DGM structures, were obtained for MP2RAGE vs. MP-RAGE. MP2RAGE demonstrated a higher mean percentage of change in whole brain volume, and lower in DGM volumes compared to MP-RAGE. The observed higher percentage of annual whole-brain volume change based on MP2RAGE falls in the range of normative values reported by a longitudinal study in normal aging [34]. Since our studied subjects were above the age of 50, these findings strongly suggest that overcoming partial volume effects that are inherent in weighted images [35], MP2RAGE might be more sensitive to subtle overtime changes in tissue volumes with age. In clinical trials, this would lead to enhanced statistical power, thus allowing for the detection of possible effects of potential drugs or treatments.

The limitations of the study are: firstly, this was a single-site study. Further studies enrolling larger sample sizes in multi-centric settings would contribute to better establish the reproducibility of volumetric measurements based on MP2RAGE. Additionally, we systematically demonstrate differences in whole-brain GM, WM, and DGM assessed by the different segmentation tools. Further e inter-and-intra modality investigations are needed to determine differences in cortical thickness measures and to establish whether the net gain of MP2RAGE would out-weight the cost of the additional acquisition time.

In conclusion, the obtained results indicate that MP2RAGE offers enhanced CNR compared to MP-RAGE. Using the same segmentation tools for multi-center or longitudinal studies is essential for more reproducible tissue classification and optimal segmentation results. Therefore, it is a strong candidate to be used for whole-brain and DGM segmentation in future basic and clinical trials, allowing more sensitive structural metrics for the monitoring of the progression of many diseases and outcomes of the different treatment strategies.

Data Availability

De-identified data set supporting this work are available from the upon institutional review board (IRB) at Tel Aviv Sourasky Medical Center (Tel: +972-3-6974003; FAX: +972-3-6973974; Email: anatmi@tlvmc.gov.il) upon reasonable request by researchers who meet the criteria for access to confidential data.

Funding Statement

This study was funded by Biogen Inc., who also provided support for this study in the form of salary for RMH. The specific roles of these authors are articulated in the ‘author contributions’ section. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.A G, N DS. Clinical use of brain volumetry. Journal of magnetic resonance imaging : J Mag Reson Imaging. 2013;37(1):1–14. doi: 10.1002/jmri.23671 [DOI] [PubMed] [Google Scholar]
  • 2.Pagnozzi AM, Fripp J, Rose SE. Quantifying deep grey matter atrophy using automated segmentation approaches: A systematic review of structural MRI studies. Neuroimage. 2019;201:116018. doi: 10.1016/j.neuroimage.2019.116018 [DOI] [PubMed] [Google Scholar]
  • 3.Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele P-F, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage. 2010;49(2):1271–1281. doi: 10.1016/j.neuroimage.2009.10.002 [DOI] [PubMed] [Google Scholar]
  • 4.Choi U-S, Kawaguchi H, Matsuoka Y, Kober T, Kida I. Brain tissue segmentation based on MP2RAGE multi-contrast images in 7 T MRI. PLoS ONE. 2019;14(2):e0210803. doi: 10.1371/journal.pone.0210803 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Mugler JP, Brookeman JR. Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE). Magn Reson Med. 1990;15(1):152–157. doi: 10.1002/mrm.1910150117 [DOI] [PubMed] [Google Scholar]
  • 6.Deoni SCL. High-resolution T1 mapping of the brain at 3T with driven equilibrium single pulse observation of T1 with high-speed incorporation of RF field inhomogeneities (DESPOT1-HIFI). J Magn Reson Imaging. 2007;26(4):1106–1111. doi: 10.1002/jmri.21130 [DOI] [PubMed] [Google Scholar]
  • 7.Zhu DC, Penn RD. Full-brain T1 mapping through inversion recovery fast spin echo imaging with time-efficient slice ordering. Magn Reson Med. 2005;54(3):725–731. doi: 10.1002/mrm.20602 [DOI] [PubMed] [Google Scholar]
  • 8.Van de Moortele P-F, Auerbach EJ, Olman C, Yacoub E, Uğurbil K, Moeller S. T1 weighted brain images at 7 Tesla unbiased for Proton Density, T2* contrast and RF coil receive B1 sensitivity with simultaneous vessel visualization. Neuroimage. 2009;46(2):432–446. doi: 10.1016/j.neuroimage.2009.02.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Okubo G, Okada T, Yamamoto A, et al. MP2RAGE for deep gray matter measurement of the brain: A comparative study with MPRAGE. J Magn Reson Imaging. 2016;43(1):55–62. doi: 10.1002/jmri.24960 [DOI] [PubMed] [Google Scholar]
  • 10.W S, T S, Sh O, Mj L. CNR improvement of MP2RAGE from slice encoding directional acceleration. Mag Reson Imaging. 2016;34(6):779–784. doi: 10.1016/j.mri.2016.03.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kober T, Granziera C, Ribes D, et al. MP2RAGE multiple sclerosis magnetic resonance imaging at 3 T. Invest Radiol. 2012;47(6):346–352. doi: 10.1097/RLI.0b013e31824600e9 [DOI] [PubMed] [Google Scholar]
  • 12.Jovicich J, Czanner S, Han X, et al. MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: Reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths. Neuroimage. 2009;46(1):177–192. doi: 10.1016/j.neuroimage.2009.02.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Derakhshan M, Caramanos Z, Giacomini PS, et al. Evaluation of automated techniques for the quantification of grey matter atrophy in patients with multiple sclerosis. Neuroimage. 2010;52(4):1261–1267. doi: 10.1016/j.neuroimage.2010.05.029 [DOI] [PubMed] [Google Scholar]
  • 14.Rajapakse JC, Giedd JN, Rapoport JL. Statistical approach to segmentation of single-channel cerebral MR images. IEEE Trans Med Imaging. 1997;16(2):176–186. doi: 10.1109/42.563663 [DOI] [PubMed] [Google Scholar]
  • 15.Manjón JV, Eskildsen SF, Coupé P, Romero JE, Collins DL, Robles M. Nonlocal intracranial cavity extraction. Int J Biomed Imaging. 2014;2014:820205. doi: 10.1155/2014/820205 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Manjón JV, Coupé P. volBrain: An Online MRI Brain Volumetry System. Front Neuroinform. 2016;10. doi: 10.3389/fninf.2016.00030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zhang Y, Brady M, Smith S. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans Med Imaging. 2001;20(1):45–57. doi: 10.1109/42.906424 [DOI] [PubMed] [Google Scholar]
  • 18.Haast RAM, Ivanov D, Formisano E, Uludaǧ K. Reproducibility and Reliability of Quantitative and Weighted T1 and T2* Mapping for Myelin-Based Cortical Parcellation at 7 Tesla. Front Neuroanat. 2016;10. doi: 10.3389/fnana.2016.00112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Patenaude B, Smith SM, Kennedy DN, Jenkinson M. A Bayesian model of shape and appearance for subcortical brain segmentation. Neuroimage. 2011;56(3):907–922. doi: 10.1016/j.neuroimage.2011.02.046 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Fischl B, Salat DH, Busa E, et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron. 2002;33(3):341–355. https://surfer.nmr.mgh.harvard.edu/ftp/articles/fischl02-labeling.pdf. doi: 10.1016/s0896-6273(02)00569-x [DOI] [PubMed] [Google Scholar]
  • 21.Coupé P, Manjón JV, Fonov V, Pruessner J, Robles M, Collins DL. Patch-based segmentation using expert priors: application to hippocampus and ventricle segmentation. Neuroimage. 2011;54(2):940–954. doi: 10.1016/j.neuroimage.2010.09.018 [DOI] [PubMed] [Google Scholar]
  • 22.Helms G. Segmentation of human brain using structural MRI. MAGMA. 2016;29(2):111–124. doi: 10.1007/s10334-015-0518-z [DOI] [PubMed] [Google Scholar]
  • 23.Morey RA, Selgrade ES, Wagner HR, Huettel SA, Wang L, McCarthy G. Scan–rescan reliability of subcortical brain volumes derived from automated segmentation. Human Brain Mapping. 2010;31(11):1751–1762. doi: 10.1002/hbm.20973 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Despotović I, Goossens B, Philips W. MRI Segmentation of the Human Brain: Challenges, Methods, and Applications. Comput Math Methods Med. 2015;2015. doi: 10.1155/2015/450341 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Smith SM. Fast robust automated brain extraction. Hum Brain Mapp. 2002;17(3):143–155. doi: 10.1002/hbm.10062 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.O’Brien KR, Kober T, Hagmann P, et al. Robust T1-weighted structural brain imaging and morphometry at 7T using MP2RAGE. PLoS One. 2014;9(6):e99676. doi: 10.1371/journal.pone.0099676 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Tustison NJ, Avants BB, Cook PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010;29(6):1310–1320. doi: 10.1109/TMI.2010.2046908 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Evans AC, Kamber M, Collins DL, MacDonald D. An MRI-Based Probabilistic Atlas of Neuroanatomy. In: Shorvon SD, Fish DR, Andermann F, Bydder GM, Stefan H, eds. Magnetic Resonance Scanning and Epilepsy. NATO ASI Series. Springer; US; 1994:263–274. doi: 10.1007/978-1-4615-2546-2_48. [DOI] [Google Scholar]
  • 29.Ashburner J, Friston K. Multimodal image coregistration and partitioning—a unified framework. Neuroimage. 1997;6(3):209–217. doi: 10.1006/nimg.1997.0290 [DOI] [PubMed] [Google Scholar]
  • 30.Reuter M, Tisdall MD, Qureshi A, Buckner RL, van der Kouwe AJW, Fischl B. Head motion during MRI acquisition reduces gray matter volume and thickness estimates. Neuroimage. 2015;107:107–115. doi: 10.1016/j.neuroimage.2014.12.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Rosen AFG, Roalf DR, Ruparel K, et al. Quantitative assessment of structural image quality. NeuroImage. 2018;169:407–418. doi: 10.1016/j.neuroimage.2017.12.059 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Meijerman A, Amiri H, Steenwijk MD, et al. Reproducibility of Deep Gray Matter Atrophy Rate Measurement in a Large Multicenter Dataset. AJNR Am J Neuroradiol. 2018;39(1):46–53. doi: 10.3174/ajnr.A5459 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Brown EM, Pierce ME, Clark DC, et al. Test-retest reliability of FreeSurfer automated hippocampal subfield segmentation within and across scanners. NeuroImage. 2020;210:116563. doi: 10.1016/j.neuroimage.2020.116563 [DOI] [PubMed] [Google Scholar]
  • 34.Takao H, Hayashi N, Ohtomo K. A longitudinal study of brain volume changes in normal aging. European Journal of Radiology. 2012;81(10):2801–2804. doi: 10.1016/j.ejrad.2011.10.011 [DOI] [PubMed] [Google Scholar]
  • 35.Ruan S, Jaggi C, Xue J, Fadili J, Bloyet D. Brain tissue classification of magnetic resonance images using partial volume modeling. IEEE Transactions on Medical Imaging. 2000;19(12):1179–1187. doi: 10.1109/42.897810 [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Joseph Najbauer

3 Feb 2021

Pécs, Hungary

February 2, 2021

PONE-D-20-40185

Whole brain and deep gray matter structure segmentation: quantitative comparison between MPRAGE and MP2RAGE sequences

PLOS ONE

Dear Dr. Droby,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised by the Reviewers, listed below.

Please submit your revised manuscript by Mar 20 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Joseph Najbauer, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Competing Interest section:

"NG: serves as a member of the Editorial Board for the Journal of Parkinson's Disease. He serves as consultant to Sionara, Accelmed, Teva, NeuroDerm, Intec Pharma, Pharma2B, Denali and Abbvie. He receives royalties from Lysosomal Therapeutics (LTI) and payment for lectures at Teva, UCB, Abbvie, Sanofi-Genzyme, Bial and Movement Disorder Society. He received research support from the Michael J Fox Foundation, the National Parkinson Foundation, the European Union 7th Framework Program and the Israel Science Foundation as well as from Teva NNE program, Biogen, LTI, and Pfizer. RMH is an employee of and owns stock in Biogen. JC: is a former employee of Biogen. "

We note that one or more of the authors have an affiliation to the commercial funders of this research study : Biogen Inc.

2.1. Please provide an amended Funding Statement declaring this commercial affiliation, as well as a statement regarding the Role of Funders in your study. If the funding organization did not play a role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript and only provided financial support in the form of authors' salaries and/or research materials, please review your statements relating to the author contributions, and ensure you have specifically and accurately indicated the role(s) that these authors had in your study. You can update author roles in the Author Contributions section of the online submission form.

Please also include the following statement within your amended Funding Statement.

“The funder provided support in the form of salaries for authors [insert relevant initials], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section.”

If your commercial affiliation did play a role in your study, please state and explain this role within your updated Funding Statement.

2.2. Please also provide an updated Competing Interests Statement declaring this commercial affiliation along with any other relevant declarations relating to employment, consultancy, patents, products in development, or marketed products, etc.  

Within your Competing Interests Statement, please confirm that this commercial affiliation does not alter your adherence to all PLOS ONE policies on sharing data and materials by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If this adherence statement is not accurate and  there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include both an updated Funding Statement and Competing Interests Statement in your cover letter. We will change the online submission form on your behalf.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

3.  We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

4. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. Please see the following video for instructions on linking an ORCID iD to your Editorial Manager account: https://www.youtube.com/watch?v=_xcclfuvtxQ

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors investigated the agreement of brain tissues (white matter, grey matter and deep grey matter) volumes obtained from MP2RAGE compared to MPRAGE. The authors compared the volumes obtained from different segmentation methods at baseline and 1 year follow-up to assess the accuracy and the reproducibility of volumetric measures.

The authors showed the relevance of using MP2RAGE to estimate brain volumes.

I have few comments and questions to the authors:

Material and Methods

-Please provide more detailed description of the used segmentation methods.

-Please provide demographical information for the longitudinal group.

-Please explain why you did not used a removing background noise method on the MP2RAGE and N3/N4 bias field correction on the MPRAGE before tissues classification (O’Brian et al., 2014) ?

Results

-Please provide the CNR difference between DGM and WM.

-Please provide a Figure for each tissue segmentation estimated by the used methods.

-Please provide the results of the intra-modality voxel-wise differences between baseline and follow-up as well as the inter-modality voxel-wise differences at follow-up.

-What about inter and intra-modality cortical thickness differences at baseline and follow-up?

-Table 4: please provide mean percentage change overtime for WM volume.

Discussion

-Please discuss the potential effect of background noise on tissues classification.

-Please update your discussion according to my comments/questions.

Reviewer #2: In this work, Droby et al. compared whole brain, cortical and subcortical gray, and white matter tissue segmentations between results based on MP2RAGE and MPRAGE data and different segmentation software, as well as between data acquired at different time points (1 yr apart). Across a total of 29 subjects, they found that MP2RAGE data is characterized by significant higher CNR leading to differences in subcortical and white and gray matter tissue probabilities, depending on the segmentation tool used. Comparative studies such as this are important to validate popular sequences, in this case MPRAGE and MP2RAGE, for anatomical imaging. Especially clinical researchers that rely on accurate segmentation of brain structures will benefit from this work and/or enable them to better design acquisition and analysis protocols. Therefore, I see the value of this manuscript but think it must take care of the following major issues to consider publication:

First, with regards to the impact of data quality and CNR analysis:

1. The authors define the sequence-specific CNR as the difference between the mean values of GM and WM, divided by the sum of squares of their standard deviation. However, ideally, CNR values need to be corrected for differences in total acquisition time (T) as longer acquisition (such as for the MP2RAGE sequence) positively affect achievable CNR. As such, current CNR needs to be converted to CNR per unit time, by normalizing it to the square root of T (in minutes) as for example done in https://doi.org/10.3389/fnana.2016.00112. This, as well as the fact that the MP2RAGE T1w image is a combination of two separate inversion images, and thus based on more data, needs to be further discussed to put the results in perspective.

2. Considering that the quality of the data might vary across the brain due to B1 transmit and receive homogeneity differences, as mentioned in the introduction by the authors too, I think it would be of interest to see whether difference in CNR between MPRAGE and MP2RAGE varies across the brain, and whether this spatial variation is more prominent for MPRAGE or MP2RAGE. This could be tested by averaging per lobe for example and will also ease interpretation of the observed differences between tissue probability maps.

3. In line my previous comment and with the authors’ hypothesis stated in the discussion that the observed differences in CNR “can be explained by the differences between the two sequences, and the fact that increased rates of bias-field effects are inherent in MP-RAGE images at high and ultra-high magnetic fields (≥3 Tesla), whereas, MP2RAGE is less sensitive to B1 bias, thus enabling the acquisition of images with enhanced” I would propose to calculate subject-specific difference maps between the MP-RAGE and MP2RAGE images (e.g., after normalization to the average CSF value, for example). This will provide a surrogate bias map and might provide additional information with regards to origin of the observed differences in terms of tissue segmentation. See also https://doi.org/10.3389/fnana.2016.00112 and https://doi.org/10.1016/j.neuroimage.2020.117373 for relevant demonstrations and discussions on the effect of B1 biases on cortical and subcortical segmentations using MP2RAGE data.

With regards to the tissue probability map comparisons:

4. Import details are lacking with regards to the voxel-based analyses of the tissue probability maps. The authors should elaborate more on the SPM analyses as it is impossible now for the reader to replicate the current analyses. For example, which registration tool was used? And why was chosen for a linear registration instead of non-linear transformation? The latter will ensure larger overlap across subjects for statistical analyses and more power. Moreover, did the authors coregister each MP-RAGE and MP2RAGE image individually to the template space or did they use the same transformation matrix for both types of images? This coregistration process should be clarified to rule out potential sources of biases in the comparison.

5. Finally, it is confusing why the WM results using volBrain are not shown in Figure 2 as for CAT-12 and FSL-FAST and as suggested in the caption.

In general:

6. Especially in the introduction the authors should better emphasize the novelty of the current work (e.g., compared to https://doi.org/10.1002/jmri.24960).

7. I think the paper would benefit from an extra table/figure that summarizes all findings. This will make it easier for the reader to identify the key message of the manuscript and follow the narrative of the discussion.

Minor points

Introduction:

8. Define ‘high’ and ‘ultra-high’ magnetic field strengths

9. Last paragraph, ‘doesn’t’ to ‘does not’

Test-retest reliability (methods):

10. Add characteristics for the test-retest subjects

11. First sentence: ‘described above’ to ‘described below’ or move paragraph

12. Second sentence:

a. Change ‘T1-MPRAGE’ to MP-RAGE to follow previous notations

b. Change ‘patients’ to ‘participants’

MRI acquisition protocol (methods):

13. No need to define MP-RAGE again.

14. Remove sequence parameters from text as this is redundant with Table 1. Also add readout bandwidth to better estimate SNR and interpret the CNR differences.

15. Last sentence: would be informative for the reader to add a bit more detail what the authors are referring to with ‘Table 1 & Figure 1’

MRI data processing (methods):

16. As for MP-RAGE, no need to define CNR again.

Whole brain tissue segmentation (methods):

17. First sentence: change ‘free-widely’ to ‘free and widely’

18. Last sentence:

c. ‘Montreal neurological institute’ to ‘Montreal Neurological Institute’

Contrast-to-noise (methods):

19. Subscript ‘WG’ in ‘CNRWG’ as used in the equation and later in the results

DGM segmentation (methods):

20. I do not understand why there is a difference in the tools used for whole-brain and DGM segmentation. I would suggest using the same set of tools across both analyses or motivate why it was chosen not to.

Statistical analysis (methods)s:

21. Is there are specific reason why the authors did not use a multivariate (i.e., across multiple regions of interest) ANOVA test for the statistical differences? This will also allow to test for the effect of segmentation tools, region of interest and/or potential interactions.

Whole-brain volumes (results):

22. I think it would be helpful for the reader to see example tissue segmentation results for a single subject using both types of data. This could for example be combined with Figure 1.

DGM segmentation (results):

23. Second sentence: ‘accumbnes’ to ‘accumbens’

Test-retest reliability (results):

24. First sentence:

d. ‘MP2RGAE’ to ‘MP2RAGE’

e. Higher what? Volume? Please specify

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Aug 6;16(8):e0254597. doi: 10.1371/journal.pone.0254597.r002

Author response to Decision Letter 0

6 Apr 2021

Response to reviewers' comments

We would like to thank the esteemed reviewers deeply for their positive evaluation of our manuscript, and their valuable feedback. Below, are our point-by-point response to the various comments.

Reviewer #1

The authors investigated the agreement of brain tissues (white matter, grey matter and deep grey matter) volumes obtained from MP2RAGE compared to MPRAGE. The authors compared the volumes obtained from different segmentation methods at baseline and 1 year follow-up to assess the accuracy and the reproducibility of volumetric measures.

The authors showed the relevance of using MP2RAGE to estimate brain volumes.

I have few comments and questions to the authors:

Material and Methods

1. Please provide more detailed description of the used segmentation methods.

Response: We agree with the reviewer that a detailed description of the applied segmentation tools would be helpful for the readers. We have modified the corresponding sections in the methods adding the following information:

1. Brain tissue segmentation: was performed using widely used automatic segmentation tools that rely on different segmentation algorithms including: (1) the computational analysis toolbox (CAT12, http://www.neuro.uni-jena.de/cat/) extension for statistical parametric mapping (SPM12, https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) that uses a-priori tissue probability maps, to perform an initial skull-stripping, and to initialize the segmentation. Then, it uses an adaptive maximum a posteriori segmentation approach (hypothesis-free approach) [Rajapakse et al., 1997]. (2) volBrain processing pipeline (https://volbrain.upv.es/instructions.php) that performs non-local intracranial cavity extraction based on a library of prelabeled brain images allowing to capture the large variability of brain shapes [Manjon et al., 2014; Majon et al., 2016]. After brain extraction, mean values of the different tissues types are estimates by trimmed mean segmentation (TMS), excluding partial volume voxels from the estimation jointly with the use of an unbiased robust mean estimator [Manjón et al., 2008; Manjon et al., 2016]. (3) FMRIB's Automated Segmentation Tool (FSL-FAST, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FAST), which is based on a hidden Markov random field model and an associated Expectation-Maximization algorithm for tissue segmentation [Zhang et al., 2001] Brain extraction (BET) was performed prior to image segmentation.

2. Deep grey matter segmentation: the following tools were used: DGM segmentation was performed using: (1) FSL FIRST segmentation tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FIRST), which is based on bayesian statistical models of shape and appearance for subcortical brain segmentation [Patenuade et al., 2011] (2) FreeSurfer (https://surfer.nmr.mgh.harvard.edu/) that segments gray-white matter into different cortical regions based on sulci and gyri, and also segments major subcortical regions in the brain. FreeSurfer provides a full processing stream for structural MRI data, including Skull stripping, B1 bias field correction, and gray-white matter segmentation [Fischl et al., 2002], and (3) volBrain labeling processing pipeline that applies novel patch-based method relying on expert manual segmentations as priors [Coupe et al., 2011].

2. Please provide demographical information for the longitudinal group.

Response: In the longitudinal group, N=12 participants (6 females, mean age ± SD = 51.5 ± 6.5 years) were included, and re-scanned after 1 year using the same MRI system and protocol. This information was now added to the corresponding section in the Methods.

3. Please explain why you did not used a removing background noise method on the MP2RAGE and N3/N4 bias field correction on the MPRAGE before tissues classification (O’Brian et al., 2014) ?

Response: We apologize that we failed to provide this information in the methods section. In fact, all images underwent brain extraction prior to the segmentation process results including bias field correction, skull, and background noise removal using the pipelines embedded within the used pipelines (i.e, CAT12, FAST, FIRST, Freesurfer, and volBrain). We added this information in the corresponding section of the methods.

Results

4. Please provide the CNR difference between DGM and WM.

Response: CNR between the WM and DGM (CNRWdG) structures were calculated based on the extracted WM DGM masks obtained by FIRST. Significantly higher mean CNRWdG was detected for the MP2RAGE vs. MP-RAGE (Mean±SD = 0.98±0.13 and 0.16±1.00 respectively; p<0.0001). This information was added to the corresponding sections in the Methods and Results.

5. Please provide a Figure for each tissue segmentation estimated by the used methods.

Response: Thank you for this suggestion. We have included a figure (Figure 1B) illustrating the tissue probability maps achieved by the different segmentation tools.

6. Please provide the results of the intra-modality voxel-wise differences between baseline and follow-up as well as the inter-modality voxel-wise differences at follow-up.

Response: At pFWE<0.05, significant higher GM probability were found in the left thalamus in MP2RAGE vs. MP-RAGE when assessed using CAT12. In WM, MP2RAGE showed significantly higher probability in the splenium compared to MP-RAGE. However, MP-RAGE showed significantly higher WM probability in the red nucleus. No significant inter-modality differences were detected between both sequences when assessed by FAST and volBrain. This finding might be indicative of robustness of the segmentation algorithms implemented in FAST and volBrain compared to CAT12, an in line with the values in Table 4, where CAT12 showed the highest percentage of change over time compared to FAST and volBrain. Moreover, in the sub-study of 12 participants, no significant intra-modality differences were observed between both time points in all three used automatic tools. We believe that these differences are possibly due to the relatively small group, and hence low statistical power. We added these findings to the corresponding section of the results and Figure 4.

7. What about inter and intra-modality cortical thickness differences at baseline and follow-up?

Response: We thank the reviewer for this comment. We opted to not include inter-and intra-modality differences in regards to cortical thickness measures manuscript due to the following considerations. Firstly, among the investigated and used segmentation pipelines, only Freesurfer and CAT12 offer cortical thickness measures, while no such measures can be obtained by the other tested pipelines (i.e, volBrain, and FAST). More importantly, both CAT12 and Freesurfer rely on different atlases for cortical parcellation and regional labeling (CAT 12 uses Neuromorphometrics [http://neuromorphometrics.com], while Fressurfer relies on Desikan-Killiany Atlas/ Destrieux Atlas]. In this case, it would be inaccurate to infer about inter-and- intra modality comparison of the thickness assessed using both of these tools at baseline [See Tustison et al., 2014]. Therefore, we saw that it would be more meaningful to focus on global volumetric measures in order maintain coherence throughout the manuscript, especially since the accuracy of thickness estimation heavily depends on the results of tissue segmentation step [Clarkson et al., 2011; Neuroimage]. To this end, we have highlighted this as a limitation of this study the current study. We do nevertheless agree with the reviewer that this issue would be of high relevance for future follow-up studies that could address the inter-and-intra modality differences of cortical thickness measures achieved from various tools in a more systematic and coherent fashion.

8. Table 4: please provide mean percentage change overtime for WM volume.

Response: Table 4 was now updated. We inserted mean percentage of overtime change for both GM and WM.

Discussion

9. Please discuss the potential effect of background noise on tissues classification.

Response: Background noise in the image can affect the accuracy of tissue classification, and consequently, on the measured volume. However and brain extraction tools show to overcome this and improve tissue segmentation [Smith, 2002; Tustison et al., 2010; O'Brien et al., 2014]. This is of importance in case of MP2RAGE, where the performance of the different automatic segmentation tools maybe challenged due to the noisy background of the UNI images. As now detailed in the methods section, in order to overcome this problem, brain extraction step was carried out on all segmented images using the different applied tools. Moreover, we added an illustration of the step in Figure 1B. As suggested by the reviewer, this point was also addressed in the discussion as well.

10. Please update your discussion according to my comments/questions.

Response: The discussion was updated based on the new additions based on comments 4, 6-9.

Reviewer #2

In this work, Droby et al. compared whole brain, cortical and subcortical gray, and white matter tissue segmentations between results based on MP2RAGE and MPRAGE data and different segmentation software, as well as between data acquired at different time points (1 yr apart). Across a total of 29 subjects, they found that MP2RAGE data is characterized by significant higher CNR leading to differences in subcortical and white and gray matter tissue probabilities, depending on the segmentation tool used. Comparative studies such as this are important to validate popular sequences, in this case MPRAGE and MP2RAGE, for anatomical imaging. Especially clinical researchers that rely on accurate segmentation of brain structures will benefit from this work and/or enable them to better design acquisition and analysis protocols. Therefore, I see the value of this manuscript but think it must take care of the following major issues to consider publication:

First, with regards to the impact of data quality and CNR analysis:

1. The authors define the sequence-specific CNR as the difference between the mean values of GM and WM, divided by the sum of squares of their standard deviation. However, ideally, CNR values need to be corrected for differences in total acquisition time (T) as longer acquisition (such as for the MP2RAGE sequence) positively affect achievable CNR. As such, current CNR needs to be converted to CNR per unit time, by normalizing it to the square root of T (in minutes) as for example done in https://doi.org/10.3389/fnana.2016.00112. This, as well as the fact that the MP2RAGE T1w image is a combination of two separate inversion images, and thus based on more data, needs to be further discussed to put the results in perspective.

Response: Normalized CNR values - corrected per unit time similar to Haast et al. are given in the table below for both whole brain W/GM as well as DGM:

CNRWdG CNRWG

Uncorrected Corrected* Uncorrected Corrected*

MPRAGE 0.16±1.00 0.05±0.34 2.52±0.33 0.80±0.10

MP2RAGE 0.98±0.13 0.30±0.04 3.53±0.12 0.97±0.04

* Corrected = corrected per unit time similar to Haast et al.

As it can be observed from the Table, the general trend (higher CNR for the MP2RAGE compared to MPRAGE) was maintained. We have amended these normalized CNR value to the revised version of the manuscript in the corresponding sections of the Methods and results.

2. Considering that the quality of the data might vary across the brain due to B1 transmit and receive homogeneity differences, as mentioned in the introduction by the authors too, I think it would be of interest to see whether difference in CNR between MPRAGE and MP2RAGE varies across the brain, and whether this spatial variation is more prominent for MPRAGE or MP2RAGE. This could be tested by averaging per lobe for example and will also ease interpretation of the observed differences between tissue probability maps.

Response: At 3T, increased rates of transmit and receive B1 in-homogeneities can create intensity variations in the acquired images leading to reduced contrast bias field. In MP2RAGE, by combining the images from the first and second readouts, T2* and B1 inhomogeneity effects can be largely canceled out, resulting in a strongly T1-weighted image with superior GM-to-WM contrast [Van de Moortele et al., 2009; Okubo et al., 2016]. Following the reviewers' request, we calculated the CNR differences between both acquired sequences within the different lobes. For MPRAGE; the calculated CNR values per lobe were [a.u]: 354.06±46.85, 398±48.46, 413.35±36.01, and 424.65±45.51 for the frontal, parietal, occipital, and temporal lobes respectively. In MP2RAGE, the calculated CNR per lobe were: 3053.64±52.24; 3132.97±49.32, 3133.63±49.96, and 3145.05±49.29 respectively. As it can be seen based on these values, in each modality, slight variations in CNR among the different lobes can be seen, however the range of variation of CNR relative to the group mean in MPRAGE (9-13%) is higher compared to MP2RAGE (1.5-1.7%). Thus, further strengthening the conclusion that despite local variation in CNR, T1-weighted image resulting from MP2RAGE, with superior GM-to-WM. In case the reviewer sees that this data would further strengthen our here reported results, we would add this gladly to the manuscript.

3. In line my previous comment and with the authors’ hypothesis stated in the

discussion that the observed differences in CNR “can be explained by the differences between the two sequences, and the fact that increased rates of bias-field effects are inherent in MP-RAGE images at high and ultra-high magnetic fields (≥3 Tesla), whereas, MP2RAGE is less sensitive to B1 bias, thus enabling the acquisition of images with enhanced” I would propose to calculate subject-specific difference maps between the MP-RAGE and MP2RAGE images (e.g., after normalization to the average CSF value, for example). This will provide a surrogate bias map and might provide additional information with regards to origin of the observed differences in terms of tissue segmentation. See also https://doi.org/10.3389/fnana.2016.00112 and https://doi.org/10.1016/j.neuroimage.2020.117373 for relevant demonstrations and discussions on the effect of B1 biases on cortical and subcortical segmentations using MP2RAGE data.

Response: We thank the reviewer for this suggestion. In the calculated intra-subject differences maps (see example below), we observe higher GM tissue probability in MP2RAGE compared to MP-RAGE, mainly in deep grey matter. This is consistent with the reported group-level VBM results in Figure 2. Moreover, for each subject, we calculated the dice similarity coefficient (DSC) in order to assess the similarity between GM and WM segmented maps achieved from both investigated sequences. The group mean DCS coefficient were as follows: DSCGM= 0.85±0.04; DSCWM= 0.94±0.01. These findings further confirm the values presented in Table 2 and Fig3, in which MP-RAGE resulted in higher whole-brain GM and DGM compared to MP-RAGE. Based on these findings, we strongly believe that the enhanced CNR of MP2RAGE enhances edge-detection during segmentation, leading to more accurate classification of the brain tissue types.

Illustration of calculated difference tissue probability maps. Grey voxels represent voxels that were classified as GM by MP2RAGE only. Yellow voxels are voxels classified as GM in both sequences.

With regards to the tissue probability map comparisons:

4. Import details are lacking with regards to the voxel-based analyses of the tissue probability maps. The authors should elaborate more on the SPM analyses as it is impossible now for the reader to replicate the current analyses. For example, which registration tool was used? And why was chosen for a linear registration instead of non-linear transformation? The latter will ensure larger overlap across subjects for statistical analyses and more power. Moreover, did the authors co-register each MP-RAGE and MP2RAGE image individually to the template space or did they use the same transformation matrix for both types of images? This co-registration process should be clarified to rule out potential sources of biases in the comparison.

Response: For the VBM analyses, the obtained GM and WM probability maps from MP2RAGE and MP-RAGE using the different segmentation pipelines were registered to MNI using linear registration in SPM in order to achieve spatial overlap between subjects. Furthermore, VBM comparisons were performed using SPM software (SPM12; https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). For the group-level voxel-wise comparisons, paired-samples t-tests were used for inter-modality comparisons (i.e, MP-RAGE vs. MP2RAGE) of whole-brain tissue probability maps obtained from the different segmentation tools. We stated this detail in the corresponding section of the Methods.

Since the investigated participants in this study were all healthy without any history of any neurological disorders as well as no evidence of regional structural abnormalities and/or atrophy, we opted to use linear instead of non-linear registration. The use of non-linear registration might be advantageous in cases where there is large variance between the brain images from different subjects [Ying et al., 2017]. Moreover, since both images were acquired in the same MRI scan session, and using the same slice orientation and coverage for the MP-RAGE and MP2RAGE images were co-registered individually to MNI, in order to avoid un-necessary interpolations.

5. Finally, it is confusing why the WM results using volBrain are not shown in Figure 2 as for CAT-12 and FSL-FAST and as suggested in the caption.

Response: We agree with the reviewer. In order to maintain coherence in the displayed results in the figure, we modified Figure 2, adding the WM voxel-wise results. As mentioned in the results, no significant differences were found in WM using volBrain between both sequences.

In general:

6. Especially in the introduction the authors should better emphasize the novelty of the current work (e.g., compared to https://doi.org/10.1002/jmri.24960).

Response: We thank the reviewer for this remark. In the referenced study by Okubo et al., 2016, the authors showed that compared to MP-RAGE, MP2RAGE showed higher reproducibility and CNR between tissue classes compared to MP-RAGE. In their study, the authors applied CAT12 (http://www.neuro.uni-jena.de/cat/) which relies on the segmentation algorithm embedded in SPM. In our study, we systematically investigated both the reproducibility and patterns of tissue classification differences between both sequences using other segmentation tools including CAT12, FSL-FAST, volBrain, and Freesurfer, which are widely used and available environments in use. Thus, enabling us to demonstrate whether a tool-sequences interaction can be detected.

In order to better emphasize the novelty of this study, we modified the aims of the study paragraph as well as follows: '' in this study, we aimed to systematically examine the patterns of discrepancies and quantitatively assess the agreement between tissue classification obtained by different widely available automatic segmentation tools relying on different algorithms in the assessment of whole-brain tissue, as well as DGM volumes based on MP2RAGE compared to MP-RAGE images.''

7. I think the paper would benefit from an extra table/figure that summarizes all findings. This will make it easier for the reader to identify the key message of the manuscript and follow the narrative of the discussion.

Response: We agree with this suggestion of the esteemed reviewer. We have accordingly modified Figure 1 illustrating the steps conducted by the different used tools based on MP-RAGE and MP2RAGE sequences. Moreover, we added Figure 3; a scatter plots demonstrating the differences in tissue volumes measured by the different segmentation tools based on MP-RAGE and MP2RAGE sequences.

Minor points

Introduction:

8. Define ‘high’ and ‘ultra-high’ magnetic field strengths

Response: it is commonly accepted that field strengths 1T-3T are defined as high-field, whereas field strengths of ≥7T are 'ultra-high'. We added this clarification to the corresponding part in the introduction as advised.

9. Last paragraph, ‘doesn’t’ to ‘does not’

Response: This was corrected as suggested.

Test-retest reliability (methods):

10. Add characteristics for the test-retest subjects

Response: In the longitudinal group, N=12 participants (6 females, mean age ± SD = 51.5 ± 6.5 years) were included, and re-scanned after 1 year using the same MRI protocol. This information was now added to the corresponding section in the Methods.

11. First sentence: ‘described above’ to ‘described below’ or move paragraph

Response: This was corrected accordingly.

12. Second sentence:

a. Change ‘T1-MPRAGE’ to MP-RAGE to follow previous notations

b. Change ‘patients’ to ‘participants’

Response: Thank you for pointing out these points. These were changed accordingly.

MRI acquisition protocol (methods):

13. No need to define MP-RAGE again.

Response: This was adjusted as advised.

14. Remove sequence parameters from text as this is redundant with Table 1. Also add readout bandwidth to better estimate SNR and interpret the CNR differences.

Response: We now added the readout bandwidth for both MP-RAGE (150 Hz/Px), and for MP2RAGE (240 Hz/Px) to Table 1.

15. Last sentence: would be informative for the reader to add a bit more detail what the authors are referring to with ‘Table 1 & Figure 1’

Response: Thank you for this suggestion. This sentence now reads as follows: '' See Table 1 for detailed sequences parameter and Figure 1 for examples illustrating MRI images acquired using both sequences'' allowing more details on what is depicted in both Table and Figure referred to.

MRI data processing (methods):

16. As for MP-RAGE, no need to define CNR again.

Response: This was adjusted throughout accordingly.

Whole brain tissue segmentation (methods):

17. First sentence: change ‘free-widely’ to ‘free and widely’

18. Last sentence:

c. ‘Montreal neurological institute’ to ‘Montreal Neurological Institute’

Response: Both points pointed to above were adjusted accordingly in the corresponding section in the Methods.

Contrast-to-noise (methods):

19. Subscript ‘WG’ in ‘CNRWG’ as used in the equation and later in the results

Response: This was adjusted as suggested.

DGM segmentation (methods):

20. I do not understand why there is a difference in the tools used for whole-brain and DGM segmentation. I would suggest using the same set of tools across both analyses or motivate why it was chosen not to.

Response: In this study we aimed to systematically assess the differences in volumetric measures for whole brain and DGM between MP-RAGR and MP2RAGE using the different commonly-used automatic tools. Different tools and segmentation pipelines tools chosen by the users depending mainly on familiarity and prior experience. The main reason behind the discrepancy between the used tools across the different analyses, is to offer the readership a wide-and- comprehensive overview about the volumetric differences that can be observed between both modalities when assessed by the different tools, we opted to use open source tools operating on Unix-FSL, Matlab-SPM, as well as volBrain; is an open-access free of charge platform. As such, we included FSL FAST and FIRST, as well as volBrain for whole brain and DGM measures. In SPM, CAT-12 is mainly used for whole brain volumetric assessments and VBM analysis, however it is less used for DGM volumetric assessment. Moreover, it DGM volumetric assessment are often reported relying on Fressurfer as well. Therefore, in order to keep consistency in the number of used tools for whole-brain and DGM, we decided to include Freesurfer as third tool for DGM segmentation. In case the reviewer sees that this might make the paper more coherent and justified the Freesurfer DGM results can be omitted from the manuscript. We modified the aims paragraph in order to make this point more clearer as follows:'' we aimed to systematically examine the patterns of discrepancies and quantitatively assess the agreement between tissue classification obtained by different widely available, open-source automatic segmentation tools relying on different algorithms that are commonly used for the assessment of whole-brain tissue, as well as DGM volumes based on MP2RAGE compared to MP-RAGE images''.

Statistical analysis (methods):

21. Is there are specific reason why the authors did not use a multivariate (i.e., across multiple regions of interest) ANOVA test for the statistical differences? This will also allow to test for the effect of segmentation tools, region of interest and/or potential interactions.

Response: We thank the reviewer for this remark. In this study, our main aims were to investigate the patterns of volumetric differences between MP-RAGE and MP2RAGE sequences using widely available methods, as well as examine the scan-rescan stability of both sequences. By adjusting for subject-related inherent characteristics, the used statistical test (Paired-samples t-test), both in the voxel-wise and volumetric comparisons, allows us to infer on the questions at hand.

We do strongly agree with the reviewer that an ANOVA model or multivariate analyses would allow to infer about possible effects of segmentation tool, or possible tool × sequence interactions. However, we refrained from using multivariate models in the current analysis due to two main reasons: Firstly, the relatively small sample size of the cross-sectional comparisons (N=29), and N=12 in the scan-rescan sub-study limits our statistical power, and the number of main effects and interactions we can explore. We highlighted this point as a limitation of the study. Secondly, since our overriding aim in this paper is to systematically demonstrate that due its' higher CNR and lower susceptibility to B1 inhomogeneity, MP2RAGE-derived measures are more sensitive and reproducible, inference regarding possible interactions were out of scope of the current study aims. We fear that drawing such conclusions would sound less impartial, and would possibly lead readers toward using a certain segmentation method, which we hope to avoid. Yet, this raised point remains very relevant to pursue in future follow-up studies, including larger numbers of participants.

Whole-brain volumes (results):

22. I think it would be helpful for the reader to see example tissue segmentation results for a single subject using both types of data. This could for example be combined with Figure 1.

Response: We have included a figure (Figure 1B) illustrating the tissue probability maps achieved by the different segmentation tools.

DGM segmentation (results):

23. Second sentence: ‘accumbnes’ to ‘accumbens’

Response: This was corrected.

Test-retest reliability (results):

24. First sentence:

d. ‘MP2RGAE’ to ‘MP2RAGE’

e. Higher what? Volume? Please specify

Response: Thank you for pointing out these points. These were corrected. This sentence now reads as follows: '' At re-scan, MP2RAGE demonstrated higher whole-brain GM, lower WM, and CSF volumes compared to MP-RAGE in all three used tools.''

References

1. Coupé P, Manjón JV, Fonov V, Pruessner J, Robles M, Collins DL. Patch-based segmentation using expert priors: application to hippocampus and ventricle segmentation. Neuroimage. 2011;54(2):940-954. doi:10.1016/j.neuroimage.2010.09.018

2. Fischl B, Salat DH, Busa E, et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron. 2002;33(3):341-355. https://surfer.nmr.mgh.harvard.edu/ftp/articles/fischl02-labeling.pdf

3. Haast RAM, Ivanov D, Formisano E, Uludaǧ K. Reproducibility and Reliability of Quantitative and Weighted T1 and T2∗ Mapping for Myelin-Based Cortical Parcellation at 7 Tesla. Front Neuroanat. 2016;10. doi:10.3389/fnana.2016.00112

4. Manjón JV, Tohka J, García-Martí G, et al. Robust MRI brain tissue parameter estimation by multistage outlier rejection. Magn Reson Med. 2008;59(4):866-873. doi:10.1002/mrm.21521

5. Manjón JV, Eskildsen SF, Coupé P, Romero JE, Collins DL, Robles M. Nonlocal intracranial cavity extraction. Int J Biomed Imaging. 2014;2014:820205. doi:10.1155/2014/820205

6. Manjón JV, Coupé P. volBrain: An Online MRI Brain Volumetry System. Front Neuroinform. 2016;10. doi:10.3389/fninf.2016.00030

7. O’Brien KR, Kober T, Hagmann P, et al. Robust T1-weighted structural brain imaging and morphometry at 7T using MP2RAGE. PLoS One. 2014;9(6):e99676. doi:10.1371/journal.pone.0099676

8. Okubo G, Okada T, Yamamoto A, et al. MP2RAGE for deep gray matter measurement of the brain: A comparative study with MPRAGE. J Magn Reson Imaging. 2016;43(1):55-62. doi:10.1002/jmri.24960

9. Patenaude B, Smith SM, Kennedy DN, Jenkinson M. A Bayesian model of shape and appearance for subcortical brain segmentation. Neuroimage. 2011;56(3):907-922. doi:10.1016/j.neuroimage.2011.02.046

10. Rajapakse JC, Giedd JN, Rapoport JL. Statistical approach to segmentation of single-channel cerebral MR images. IEEE Trans Med Imaging. 1997;16(2):176-186. doi:10.1109/42.563663

11. Smith SM. Fast robust automated brain extraction. Hum Brain Mapp. 2002;17(3):143-155. doi:10.1002/hbm.10062

12. Tustison NJ, Avants BB, Cook PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010;29(6):1310-1320. doi:10.1109/TMI.2010.2046908

13. Van de Moortele P-F, Auerbach EJ, Olman C, Yacoub E, Uğurbil K, Moeller S. T1 weighted brain images at 7 Tesla unbiased for Proton Density, T2* contrast and RF coil receive B1 sensitivity with simultaneous vessel visualization. Neuroimage. 2009;46(2):432-446. doi:10.1016/j.neuroimage.2009.02.009

14. Ying S, Li D, Xiao B, Peng Y, Du S, Xu M. Nonlinear image registration with bidirectional metric and reciprocal regularization. PLOS ONE. 2017;12(2):e0172432. doi:10.1371/journal.pone.0172432

Attachment

Submitted filename: Reply_to_Reviewers_comments.docx

Click here for additional data file. (39.6KB, docx)

Decision Letter 1

Joseph Najbauer

13 May 2021

Pécs, Hungary

May 13, 2021

PONE-D-20-40185R1

Whole brain and deep gray matter structure segmentation: quantitative comparison between MPRAGE and MP2RAGE sequences

PLOS ONE

Dear Dr. Droby,

Thank you for submitting your manuscript (R1 version) to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised by Reviewer #2.

Please submit your revised manuscript by Jun 27 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Joseph Najbauer, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I would like to thank the authors for addressing all my comments.

I have no additional comments regarding the revised manuscript.

Reviewer #2: I appreciate the authors' efforts to accommodate most of my concerns, which has improved the manuscript. I like the new Figure 3, which nicely shows the differences for the GM, WM and CSF segmentations between acquisitions for each method. WM and CSF are generally larger based on MPRAGE data using FSL-FAST (especially) and volBrain, while more similar using CAT12. The opposite is true for GM. In addition, I thank the authors for calculating CNR per lobe and I think this is informative and should be added to the main text.

Finally, I recommend further clarification of a couple of remaining issues:

- The authors state that 'MP2RAGE resulted in more accurate brain tissue segmentation' (abstract). This needs to be rephrased as the 'ground truth' is not known so it is difficult to define 'more accurate', or, it should be clarified what they mean with 'more accurate'.

- I think the difference between the MPRAGE and MP2RAGE sequences need to be emphasized more. Especially in terms of how the T1w-like image is obtained (i.e., synthesized) for the MP2RAGE sequence.

- In their response to my comment #3, the authors provided a intra-subject difference map based on the segmentation results. Although informative as well, what I was suggesting was to calculate the difference between the raw images. This allows characterization of any spatial bias in the difference between MPRAGE and MP2RAGE data. For example, are differences larger towards the frontal lobe?

- Please verify that CNR values were calculated using the raw T1w images, and elaborate on what the potential impact of readout bandwidth differences between MPRAGE and MP2RAGE acquisitions on CNR could be.

- I would advise to add the T-statistical value and degrees of freedom to the statistical results. Also, specify that the non-binary probability masks were used and if statistical testing was restricted to GM and/or WM (i.e., masked), or not.

Figures:

- Figure 1, if possible, I would recommend to use data from the same subject for all sub-panels.

- Figure 4, it is very difficult to see the significant voxels. I would advise to zoom in on the interesting part and only show that in the figure.

Some minor textual comments I noticed:

- Abstract: 'In a sub-study, twelve participants were scanned after one year.' change to ' rescanned'

- Abstract: 'Mean ± SD = 0.97 ± 0.04 and 0.8 ± 0.1 respectively; p<0.0001' change 'and' to 'vs'

- Introduction, last paragraph: 'different widely available, open-source automatic segmentation tools' remove open-source. I was not able to find the source code for volBrain

- Methods, CNRwg: 'corrected per scan time' change to 'normalized for scan time'

- Please go through the manuscript to correct small grammatical errors.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Roy Haast

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Aug 6;16(8):e0254597. doi: 10.1371/journal.pone.0254597.r004

Author response to Decision Letter 1

7 Jun 2021

Reply to Reviewer's comments

Reviewer #2: I appreciate the authors' efforts to accommodate most of my concerns, which has improved the manuscript. I like the new Figure 3, which nicely shows the differences for the GM, WM and CSF segmentations between acquisitions for each method. WM and CSF are generally larger based on MPRAGE data using FSL-FAST (especially) and volBrain, while more similar using CAT12. The opposite is true for GM. In addition, I thank the authors for calculating CNR per lobe and I think this is informative and should be added to the main text.

We thank the reviewer for his valuable efforts and constructive feedback and suggestions. We added the CNR per lobe to the text. Furthermore, we added the range of variation of CNR per lobe as suggested to the results section.

Finally, I recommend further clarification of a couple of remaining issues:

1. The authors state that 'MP2RAGE resulted in more accurate brain tissue segmentation' (abstract). This needs to be rephrased as the 'ground truth' is not known so it is difficult to define 'more accurate', or, it should be clarified what they mean with 'more accurate'.

Response: We thank the reviewers for this comment. Indeed, the use of the phrase "more accurate" here is not accurate. Therefore, we replaced it with "reproducible", as supported by the results of test- retest study. We also changed this in the conclusions paragraph in the discussion.

2. I think the difference between the MPRAGE and MP2RAGE sequences need to be emphasized more. Especially in terms of how the T1w-like image is obtained (i.e., synthesized) for the MP2RAGE sequence.

Response: We have modified the third paragraph describing MP2RAGE in the introduction, further clarifying how the MP2RAGE image is acquired and synthesized.

3. In their response to my comment #3, the authors provided an intra-subject difference map based on the segmentation results. Although informative as well, what I was suggesting was to calculate the difference between the raw images. This allows characterization of any spatial bias in the difference between MPRAGE and MP2RAGE data. For example, are differences larger towards the frontal lobe?

Response: We apologize for the misunderstanding of this point in the previous round. Since MPRAGE is not a quantitative sequence, and due to the difference in dynamic range between both sequences comparison of the raw images might not be meaningful (or accurate). Nevertheless, to perform the spatial comparison as suggested we first normalized both images relative to the white matter signal, and then we calculated the differences between the normalized images (See Figure Rev1. below). As it can be seen in the figure, following normalization only minor differences were demonstrated between the sequences for the WM and GM tissues (<±0.25, normalized values).

Figure Rev1:

4. Please verify that CNR values were calculated using the raw T1w images, and elaborate on what the potential impact of readout bandwidth differences between MPRAGE and MP2RAGE acquisitions on CNR could be.

Response: We confirm that CNR was calculated on the acquired raw T1 images. We clarified this point in the corresponding section in the methods. MP2RAGE has higher bandwidth compared to MPRAGE (240 Hz/Px vs. 150 Hz/Px). The use of high bandwidth is expected to reduce susceptibility effects including eddy currents associated with metal, and thus to improve image SNR [See: Marquess et al., 2010, Neuroimage]. We added this information to the discussion.

5. I would advise to add the T-statistical value and degrees of freedom to the statistical results. Also, specify that the non-binary probability masks were used and if statistical testing was restricted to GM and/or WM (i.e., masked), or not.

Response: We now added the T statistical values in the corresponding sections in the results as suggested. Also, in Statistical analysis sub-section of the Methods, we section further clarified that the statistical testing was performed on the masked GM and WM probability maps achieved by the different used segmentation tools.

Figures:

6. Figure 1, if possible, I would recommend to use data from the same subject for all sub-panels.

Response: All the images in the sub-panels in Figure are actually from the same subject. As noted by the reviewer, slight variations in the slices were found however, we now rectified this and made sure that all slices match.

7. Figure 4, it is very difficult to see the significant voxels. I would advise to zoom in on the interesting part and only show that in the figure.

Response: We agree with this comment. We have reduced the number of slices shown in the left panel to better depict the significant voxels in GM and WM.

Some minor textual comments I noticed:

8. Abstract: 'In a sub-study, twelve participants were scanned after one year.' change to ' rescanned'

Response: This was changed as suggested.

9. Abstract: 'Mean ± SD = 0.97 ± 0.04 and 0.8 ± 0.1 respectively; p<0.0001' change 'and' to 'vs'

Response: The word and was changes to VS.

10. Introduction, last paragraph: 'different widely available, open-source automatic segmentation tools' remove open-source. I was not able to find the source code for volBrain

Response: We omitted the phrase "open-source" as advised.

11. Methods, CNRwg: 'corrected per scan time' change to 'normalized for scan time'

Response: This was changed accordingly.

12. Please go through the manuscript to correct small grammatical errors.

Response: We went over the manuscript thoroughly once again and corrected some minor mistakes. Hopefully we got them all!

Attachment

Submitted filename: Reviewers_comments.docx

Click here for additional data file. (186.5KB, docx)

Decision Letter 2

Joseph Najbauer

30 Jun 2021

Pécs, Hungary

June 29, 2021

Whole brain and deep gray matter structure segmentation: quantitative comparison between MPRAGE and MP2RAGE sequences

PONE-D-20-40185R2

Dear Dr. Droby,

We’re pleased to inform you that your manuscript (R2 version) has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

PLEASE NOTE:

References 1, and 10 are not complete (the last names of the authors are missing, also the year of publication and the journal volume and page naumbers are missing - PLEASE CORRECT.

PLEASE CHECK ALL REFERENECES FOR COMPLETENESS AND ACCURACY.

PLEASE CHECK THE ENTIRE MANUSCRIPT FOR COMPLETENESS AND ACCURACY.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Joseph Najbauer, Ph.D.

Academic Editor

PLOS ONE

------------------------------------

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: Yes: Roy AM Haast

Acceptance letter

Joseph Najbauer

21 Jul 2021

PONE-D-20-40185R2

Whole brain and deep gray matter structure segmentation: quantitative comparison between MPRAGE and MP2RAGE sequences

Dear Dr. Droby:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Joseph Najbauer

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Attachment

    Submitted filename: Reply_to_Reviewers_comments.docx

    Click here for additional data file. (39.6KB, docx)
    Attachment

    Submitted filename: Reviewers_comments.docx

    Click here for additional data file. (186.5KB, docx)

    Data Availability Statement

    De-identified data set supporting this work are available from the upon institutional review board (IRB) at Tel Aviv Sourasky Medical Center (Tel: +972-3-6974003; FAX: +972-3-6973974; Email: anatmi@tlvmc.gov.il) upon reasonable request by researchers who meet the criteria for access to confidential data.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES