Spiral gradient echo versus cartesian turbo spin echo imaging for sagittal contrast-enhanced fat-suppressed T ₁ weighted spine MRI: an inter-individual comparison study

Abstract

Objectives:

To compare a novel 3D spiral gradient echo (GRE) sequence with a conventional 2D cartesian turbo spin echo (TSE) sequence for sagittal contrast-enhanced (CE) fat-suppressed (FS) T ₁ weighted (T ₁W) spine MRI.

Methods:

In this inter-individual comparison study, 128 patients prospectively underwent sagittal CE FS T ₁W spine MRI with either a 2D cartesian TSE (“TSE”, 285 s, 64 patients) or a 3D spiral GRE sequence (“Spiral”, 93 s, 64 patients). Between both groups, patients were matched in terms of anatomical region (cervical/thoracic/lumbar spine and sacrum). Three readers used 4-point Likert scales to assess images qualitatively in terms of overall image quality, presence of artifacts, spinal cord visualization, lesion conspicuity and quality of fat suppression.

Results:

Spiral achieved a 67.4% scan time reduction compared to TSE. Interreader agreement was high (alpha=0.868-1). Overall image quality (4;[3,4] vs 3;[3,4], p<0.001 – p=0.002 for all readers), presence of artifacts (4;[3,4] vs 3;[3,4] p=0.027 – p=0.046 for all readers), spinal cord visualization (4;[4,4] vs 4;[3,4], p<0.001 for all readers), lesion conspicuity (4;[4,4] vs 4;[4,4], p=0.016 for all readers) and quality of fat suppression (4;[4,4] vs 4;[4,4], p=0.027 – p=0.033 for all readers), were all deemed significantly improved by all three readers on Spiral images as compared to TSE images

Conclusion:

We demonstrate the feasibility of a novel 3D spiral GRE sequence for improved and rapid sagittal CE FS T ₁W spine MRI.

Advances in knowledge:

A 3D spiral GRE sequence allows for improved sagittal CE FS T ₁W spine MRI at very short scan times.

Introduction

Contrast-enhanced (CE) fat-suppressed (FS) T ₁ weighted (T ₁W) MRI sequences are essential components of CE spine MRI protocols in patients with conditions such as tumors or infections. Spine imaging is challenging due to difficulties in depicting small anatomical structures such as spinal nerve roots and intervertebral foramina. Furthermore, image quality may be impaired by patient motion, respiration, heart motion and blood or cerebrospinal fluid (CSF) flow. ^1,2 2D cartesian turbo spin echo (TSE) sequences are currently the reference standard for CE FS T ₁W MRI of the spine. ^1–4 While these sequences exhibit certain benefits, such as a high signal-to-noise (SNR) and contrast-to-noise ratio (CNR), they are susceptible to artifacts. Furthermore, the scan times of these sequences can be quite long, especially in the case of whole-spine sagittal imaging. Thus, there is considerable interest in improving the image quality and in reducing scan time for sagittal CE FS T ₁W spinal MRI. ²

Recent hard- and software improvements and developments have made spiral MRI feasible for structural MR imaging on clinical scanners. ^5–12 Spiral MRI relies on a non-cartesian readout scheme thus exhibiting certain benefits over conventional cartesian readout schemes. Specifically, sequences with spiral sampling exhibit an exceptionally high scan speed and a reduced sensitivity towards artifacts. Thus, spiral MRI may improve image quality at a fraction of the scan time required for conventional cartesian imaging. ^5,6,9,13,14 Recent pilot studies have shown the feasibility of spiral time-of-flight (TOF) MR angiography (MRA) and T ₁W gradient echo (GRE) sequences as applied for routine clinical brain MR imaging. ^7–11

However, with the exception of a single study, ¹² the value of spiral MRI as applied in the context of routine clinical spine MR imaging has to be defined. Furthermore, in the latter study, a spiral sequence acquired in axial direction was used. In clinical routine however, sagittal images of the spine are considered the most important images as they allow for the depiction of the entire vertebral column and its pathologies. In contrast, axial images are only used to selectively gain information on specific structures.

Given the challenging nature of spine MR imaging and the potential benefits of spiral MRI, we sought to assess a novel 3D spiral GRE sequence for routine clinical sagittal CE FS T ₁W spine MR imaging.

Methods and materials

Study design and subjects

This institutional review board approved inter-individual comparison study was performed between January 2020 and April 2021. In total, 128 patients referred to our tertiary center for contrast-enhanced spinal MR imaging (cervical spine (CS), thoracic spine (TS), lumbar spine (LS), sacrum) were prospectively included.

First, 64 consecutive patients (mean age: 58 years, age range: 16–91, 35 male, 29 female) were scanned with the conventional 2D cartesian T ₁W post-contrast TSE sequence. Then, 64 consecutive and anatomically matched patients (i.e. same number and distribution of anatomical regions as in the TSE cohort) (mean age: 60 years, age range: 16–87, 34 male, 30 female) were scanned with the 3D spiral T ₁W post-contrast GRE sequence.

For both groups, the anatomical distribution was as follows: In each group, 18 patients underwent whole-spine imaging (CS, TS, LS), 21 patients underwent imaging for two anatomical regions (TS and LS - 9 patients, CS and TS - 11 patients, LS and Sacrum - 1 patient) and 25 patients underwent imaging for one anatomical region (TS - 5 patients, CS - 3 patients, LS - 17 patients). Of the 64 patients in the spiral GRE group, 19 patients had previously also undergone cartesian TSE imaging, and were thus also included in the cartesian TSE group. Thus, for a fraction of patients we performed an intra-individual rather than an inter-individual comparison. In the cartesian TSE group, metastases (in 26 patients), other tumorous lesions such as chordoma, chondrosarcoma, lipoma, hibernoma, myeloma, meningioma, schwannoma (in 12 patients), intramedullary ischemia (in 1 patient), spondylodiscitis and/ or soft tissue abscess (in 10 patients), myelitis and myeloradiculitis (in 4 patients) , hydrosyringomyelia (in 1 patient), degenerative changes (in 7 patients) and normal findings (in 3 patients) were the indications for imaging or were the final diagnosis/imaging findings. Importantly, 6 patients presented with intramedullary pathologies. In the spiral GRE group, metastases (in 20 patients), other tumorous lesions such as chordoma, chondrosarcoma, lipoma, hibernoma, myeloma, meningioma, schwannoma (in 10 patients), intramedullary ischemia (in 3 patients), spondylodiscitis and/ or soft tissue abscess (in 10 patients), myelitis and myeloradiculitis (in 7 patients), hydrosyringomyelie (in 1 patient), degenerative changes (in 11 patients) and normal findings (in 2 patients) were the indications for imaging or were the final diagnosis/imaging findings. Importantly, 11 patients presented with intramedullary pathologies.

MR imaging

All patients underwent routine spinal MRI examinations on a standard clinical 1.5 T scanner (Ingenia; Philips Healthcare, Best, the Netherlands) with a 16-channel spine coil and product software release 5.6.

Sagittal CE FS T ₁W 2D cartesian TSE (named “TSE”) or 3D spiral GRE (named “Spiral”) sequences were acquired after contrast administration (Gadovist [Gadobutrol] 1.0 mmol ml⁻¹ by Bayer HealthCare Pharmaceuticals at 0.1 ml/kg bodyweight). The spiral sequence represents a prototypical, “work-in-progress” sequence that uses a stack of spirals with an in-plane spiral-out readout scheme. Blurring due to off-resonance was corrected during reconstruction based on a magnetic field map (B0 map) acquired before the spiral scans. ⁵ Sequences were acquired on the standard product configuration without additional enhancements. Eddy current calibration of B0 eddy currents, linear, and cross-term eddy currents was performed as part of the standard system tuning procedure. Compensation of those eddy current contributions was performed in run-time as part of the vendor’s product acquisition software. ^9,15

TSE sequence parameters were chosen based on longstanding and well established routine clinical protocols. The imaging parameters for the novel Spiral sequence were implemented based on the vendor’s recommendation after initial optimization of this sequence and were adapted to the parameters of the TSE sequence wherever possible. ^16,17 Certain sequence parameters (i.e. number of slices, slice thickness and FOV) were adapted to patient size and anatomical location (cervical/thoracic/lumbar spine). ^1,2 Imaging was performed in free-breathing and without physiological triggering or gating of pulse sequences. ¹

Sequence parameters are provided in Table 1.

Table 1.

Sequence parameters

Whole-spine (three anatomical regions /arameters per anatomical region)	TSE (2D Cartesian T1W TSE mDIXON)	Spiral (3D Spiral T1W TFE mDIXON)
Technique	2D T 1W TSE	3D T1 TFE
FoV)AP/FH/RL	200 × 264 x 49 mm3	270 × 270 x 78 mm3
Acquired voxel size	0.9 × 1.25 x 3.0 mm3	0.9 × 0.9 x 3.0 mm3
Reconstructed voxel size	0.6 × 0.6 x 3.0 mm3	0.42 × 0.42 x 3.0 mm3
Slice orientation	Sagittal	Sagittal
Number of slices	15	26
Phase encoding direction	FH	-
Repetition time	534 ms	19 ms
Echo time	7.2 ms	0.91 ms
Flip angle	90 °	18 °
TFE or TSE Factor	7	33
Fatsupression	DIXON (two acquisitions)	DIXON (two acquisitions)
Spiral direction	-	Spiral-out
3D spiral distribution	-	Stack of spirals
NSAs	2	2
Receiver bandwidth	775 Hz / pixel	120 Hz / pixel
Acquisition time [mm:ss]	04:45	01:33

FOV, field of view; NSA, Number of signal average; T ₁W, T ₁ weighted; TSE, turbo spin ech; TFE, turbo field echo

Image analysis

Three readers (board-certified neuroradiologist with 30 years of post-fellowship experience and two trainees with each 3 years of experience in medical imaging) individually performed image analyses in a randomized and blinded manner. The readers were blinded both towards patient and sequence.

The image sets from all patients were completely shuffled and readers were asked to review and score one image set at a time. No restrictions were set for time, window level settings or the ability to scroll through the images of a given image set. ² Rating categories were adapted from similar studies. ^1,2,18,19 Specifically, the readers were asked to use 4-point Likert scales to grade the overall image quality (i.e. sharpness, noise, homogeneity and contrast of anatomical structures - 1: poor and non-diagnostic, 2: acceptable, 3: good, 4: excellent), presence of artifacts (i.e. motion artifacts from respiration, pulsation, swallowing, bowel and cardiac motion and patient movement - 1: severe, 2: moderate, 3: mild, 4: absent), spinal cord visualization (i.e. delineation, homogeneity and conspicuity of spinal cord and visualization relative to surrounding CSF - 1: poor and non-diagnostic, 2: acceptable, 3: good, 4: excellent), lesion conspicuity (i.e. delineation and visualization of lesion relative to surrounding parenchyma - 1: unable to see, 2: blurry but acceptable visualization, 3: good visualization, 4: excellent visualization) and quality of fat suppression (i.e. homogeneity and uniformity of fat suppression - 1: poor, 2: acceptable, 3: good, 4: excellent) for each image set.

Statistical analysis

Mann–Whitney U tests were used to assess differences in scores from image analysis. Interreader agreement for all three readers was assessed using Krippendorff’s α. The following scale was used to indicate the level of agreement: 0.0 to 0.20, poor agreement; 0.21 to 0.40, fair agreement; 0.41 to 0.60, moderate agreement; 0.61 to 0.80, substantial agreement; and 0.81 to 1.00, almost perfect agreement. ⁹ p-values were adjusted with the Benjamini-Hochberg procedure to account for multiple comparisons. A two-sided p-value of p < 0.05 was considered significant. Unless indicated otherwise, data are presented as median; [interquartile range]. All analyses were performed in the R programming language (v. 4.0.2) (R Core Team, 2020) using the packages “ggplot2” and ”irr”.

Results

Comparison of acquisition times

Acquisition times are presented for one anatomical region. The acquisition times for two anatomical regions or for whole-spine imaging are proportional to those for one anatomical region. The TSE sequence exhibited a scan time of 04:45 min while the Spiral sequence exhibited a scan time of 01:33 min. Thus, the Spiral sequence achieved a scan time reduction of 67.4% as compared to the TSE sequence.

Image analysis

A detailed overview is provided as a visual representation in Figure 1 . Representative image examples are provided in Figures 2–5.

Figure 1.

: Overview of rating scores. TSE, turbo spin echo.

Figure 2.

: Representative example of a patient referred for contrast-enhanced whole-spine imaging. This patient underwent cartesian TSE imaging as well as spiral GRE imaging. The patient is a 69-year-old male with metastatic adenoidcystic carcinoma. Multiple osseous contrast enhancing metastases at levels Th6, Th11, Th12, L3-L5, with contrast enhancing epidural tumor mass at level Th5 to Th6 can be seen on the TSE images. The Spiral images were acquired as part of follow-up MRI examination 11 months later, after decompression operation at level Th5 to Th6, after local radiotherapy with 10 × 3 Gy at level Th4 to Th7 and after eight cycles of chemotherapy. Extensively progressive inhomogeneous contrast enhancing osseous spinal metastases in cervical, thoracic and lumbar and sacral vertebral bodies and posterior vertebral elements and progressive dural infiltration with diffuse and linear dural enhancement, especially of the posterior dura. Note the improved overall image quality on Spiral images, especially across the lumbar vertebral bodies. GRE, gradient echo; TSE, turbo spin echo.

Figure 3.

: Representative examples of patients referred for contrast-enhanced imaging of the cervical and thoracic spine. (TSE) - 55-year-old female patient with neurosarcoidosis and punctate discrete leptomeningeal enhancement at the dorsal surface of the spinal cord at levels Th8 to Th11 Note the slightly noisy appearance of the TSE images, particularly across the thoracic vertebral bodies. (Spiral) - 52 year old male patient after recovery of unclassified multisegmental myelitis (without medullary contrast enhancement). Note the crisp appearance of the spinal cord and the vertebral bodies on Spiral images. TSE, turbo spin echo.

Figure 4.

: Representative examples of patients referred for contrast-enhanced imaging of the lumbar spine. (TSE) - 76-year-old male patient with acute spondylitis/spondylodiscitis at levels L4/L5 and L5/S1 with punctate and linear contrast enhancement of the end plates, the intervertebral disc, the vertebral bodies, the ventral and dorsal epidural space with epidural abscess formation and the prevertebral soft tissues as seen on the TSE image. Osteochondrosis type Modic I and II at level L3-L4 and type Modic I at level L5-S1. (Spiral) - Acute spondylodiscitis in a 76-year-old male patient at level L2/L3 with punctate and linear contrast enhancement of the end plates, the vertebral bodies, the ventral epidural space and the prevertebral soft tissues at level L2 and L3 as seen on the Spiral image. Severe degenerative changes with osteochondrosis Modic II at levels Th12/L1, L1/L2, L3/L4, L4/L5 and L5/S1. Spondylolisthesis of L5 with bilateral spondylolysis L5. TSE, turbo spin echo.

Figure 5.

Representative examples of patients referred for contrast-enhanced whole-spine imaging. (TSE) - 32-year-old male patient with osseous metastasis of sigmoid carcinoma at level Th3. The metastasis shows inhomogeneous contrast enhancement on the TSE images. (Spiral) - 64-year-old male patient with multiple myeloma and multiple intense and homogeneous contrast enhancing osseous lesions (posterior arch of atlas, vertebral bodies or posterior elements of C5, Th1, Th2, Th4, Th5, Th6, Th10) on the sagittal postcontrast fat-saturated 3D spiral GRE image. Pathological vertebral body fracture of Th5. Pathological lumbar retroperitoneal lymph node at level L4. Note, despite the challenging anatomy of this patient, the excellent overall image quality, nerve root depiction and lesion visualization on the Spiral sequence. GRE, gradient echo; TSE, turbo spin echo.

Interreader agreement was high (α = 0.868 – α = 1). As for the scores from image analysis, overall image quality (Reader 1/Reader 2/Reader 3 – Spiral vs TSE: 4; [3,4]/ 4; [3,4]/ 4; [3,4] vs 3; [3,4]/ 3; [3,4]/ 3; [3,4], p<0.001 – p=0.002 for all readers), presence of artifacts (Reader 1/ Reader 2/ Reader 3 – Spiral vs TSE: 4; [3,4]/ 4; [3,4]/ 4; [3,4] vs 3; [3,4]/ 3; [3,4]/ 3; [3,4], p=0.027 – p=0.046 for all readers), spinal cord visualization (Reader 1/ Reader 2/ Reader 3 – Spiral vs TSE: 4; [4,4]/ 4; [4,4]/ 4; [4,4] vs 4; [3,4]/ 4; [3,4]/ 4; [3,4], p<0.001 for all readers), lesion conspicuity (Reader 1/Reader 2/Reader 3 – Spiral vs TSE: 4; [4,4]/ 4; [4, 4]/ 4; [4, 4] vs 4; [4, 4]/ 4; [4, 4]/ 4; [4, 4], p = 0.016 for all readers) and finally quality of fat suppression (Reader 1/ Reader 2/ Reader 3 – Spiral vs TSE: 4; [4,4]/ 4; [4,4]/ 4; [4,4] vs 4; [4,4]/ 4; [4,4]/ 4; [4,4], p=0.027 – p=0.033 for all readers), were all deemed significantly improved by all three readers on Spiral images as compared to TSE images. Thus, while all five metrics were improved on Spiral images as compared to TSE, the largest benefits of the Spiral sequence were found in terms of overall image quality, presence of artifacts and spinal cord visualization.

Discussion

In this feasibility study, we show that a 3D spiral GRE sequence allows for improved sagittal contrast-enhanced fat-suppressed T ₁W spinal MRI in comparison to a conventional 2D cartesian TSE sequence at very short scan times. Specifically, while all qualitative metrics were deemed superior on Spiral images as compared to TSE images, the largest benefits of the Spiral sequence were observed in terms of overall image quality, presence of artifacts and spinal cord visualization. Furthermore and importantly, the Spiral sequence exhibited a 67.4% scan time reduction compared to the TSE sequence.

Due to high technological demands, spiral MRI has only recently been shown to be feasible for routine clinical MR imaging on standard clinical scanners. ^5,6 Select pilot studies have explored its potential for brain MR imaging: one study assessed the feasibility of a spiral spin echo (SE) sequence for post-contrast pediatric brain MRI. 24 pediatric patients were scanned with the spiral SE sequence and a conventional cartesian TSE sequence. The spiral sequence was deemed superior in terms of subjective preference and flow artifact reduction and exhibited a scan time reduction of 48% as compared to its cartesian counterpart. ⁵

Another study assessed the performance of two spiral time-of-flight MRA sequences for clinical intracranial vessel imaging at 1.5 T. 44 patients underwent both spiral TOF imaging as well as cartesian compressed sensing accelerated TOF imaging. Two readers determined that the spiral TOFs were superior to their cartesian counterpart for the visualization of small intracranial vessels and that they exhibited comparable diagnostic performance yet at much shorter scan times. ⁸ These findings were largely confirmed by three further studies performed at 1.5 T or 3 T. ^7,10,11

Lastly, a recent study reported the use of an isotropic, axially acquired spiral GRE sequence for brain MR imaging. The spiral sequence enabled high-quality clinical brain MR imaging at very short scan times in comparison to conventional cartesian sequences. ⁹

Concerning the value of spiral MRI for spine imaging, promising results were recently reported in a pilot study that assessed the value of an axially acquired spiral GRE sequence. ¹² Here, we largely confirm these findings as we also observed an improved performance of our spiral GRE sequence despite it being acquired in sagittal direction. Sagittal images are largely considered the most important images for spine MRI as they allow clinicians and radiologists to rapidly gain an overview over the whole vertebral column. The same commonly applies for brain imaging, where morphological T ₁W images are also often acquired in sagittal direction. However, interestingly in previous studies with spiral sequences as used for structural imaging, ^9,12 images were acquired in axial direction rather than in sagittal direction. In sagittal spiral imaging, difficulties such as fold-over artifacts may arise more frequently, yet here we provide evidence that high quality sagittal spiral images can be achieved even when imaging a challenging anatomical region such as the spine.

MRI of the spine is considered challenging as firstly, small anatomical structures must be visualized accurately and secondly, patient motion as well as physiological processes such as respiration, fluid pulsation or cardiac motion may lead to a loss of image quality. Furthermore, with some patients presenting for imaging in the post-operative setting, image quality may be degraded by metallic objects and implants thus causing severe susceptibility artifacts. While certain compensation techniques, such as presaturation bands, flow compensation techniques, cardiac and respiratory gating and specialized metal artifact reduction MRI techniques such as SEMAC ²⁰ may be used to improve image quality, scan times are prolonged considerably when such techniques are added to the scan. ¹ In clinical routine, this may not be tolerable or feasible as the scan time may be long already, especially in case of whole-spine sagittal imaging where three separate stacks are acquired over the cervical, thoracic and lumbar spine.

Spiral MRI may thus be particularly valuable for (sagittal) spine MRI. Spiral MRI is a non-cartesian technique that utilizes a spiral k-space trajectory and exhibits key advantages over conventional cartesian MRI. Firstly, sequences with spiral k-space sampling are more robust towards various types of artifacts (flow, fold-over aliasing, geometric distortions, Gibbs ringing etc.). This is due to spiral trajectory’s inherently reduced gradient moments, central k-space oversampling, the non-dedicated phase-encoding direction and the incoherent dispersion of unwanted signal changes between spiral arms. Secondly, spiral sequences achieve an unparalleled k-space sampling efficiency as the spiral traverses the k-space more efficiently per unit of time than in cartesian trajectories. This can be leveraged into a decrease in scan time and/or an increase in SNR. ^5,6 Incidentally, the spiral GRE sequence presented in this study exhibited improved qualitative image quality metrics at a 67.4% scan time reduction as compared to its conventional cartesian counterpart.

As for short scan times, one should consider that they may be valuable in various ways: firstly, short scan times may lead to improved patient comfort and consequently to less stress. Secondly, while not investigated in this study, short scan times offer the possibility to either increase SNR or to improve spatial resolution by adjusting the imaging protocol without prolonging scan time in comparison to standard cartesian sequences.

Concerning the novel spiral GRE sequence used in this study, it should be noted that this sequence can be acquired both at higher (and lower) field strengths and may also be used for a variety of other anatomical areas and scenarios. Specifically, its applicability for rapid high-resolution musculoskeletal imaging or improved abdominal imaging should be explored in future studies. Moreover, the 3D nature of the Spiral sequence should be highlighted. This property may enable the acquisition of contiguous thin slices and the straightforward computation of axial and coronal reformats from the sagittal-acquired data. ¹ Additonally, while not currently feasible on standard clinical scanners, spiral MRI may potentially be accelerated even further by means of parallel imaging or compressed sensing techniques. ^21–23 Lastly, while not analyzed explicitly in this study, special consideration should be given to imaging in the post-operative setting. In particular, it must be considered that our Spiral sequence is a GRE sequence and not a TSE sequence. In patients where metal induced artifacts can be expected, TSE sequences are often preferred as the artifacts appear smaller than with GRE sequences. ²⁴ This should be taken into account with regards to the novel spiral GRE sequence described in this study.

There are certain limitations to this study that must be acknowledged: Firstly, while in line with similar studies, ^21,25 the two sequences were not compared intra-individually in all patients. However, it should be noted that we acquired images after contrast administration and in such scenarios an intra-individual study design may itself be less optimal as the acquisition of the sequences relative to the time point of administration of the contrast agent may have an influence on the image impression. ⁹ Furthermore, especially for whole-spine imaging, the acquisition of both sequences would have not been reasonable for patients due to the considerable prolongment in scan time. Secondly, while in the range of similar inter-individual feasibility studies the sample size was limited. ^21,25 Thirdly, the patient cohort was quite inhomogeneous. In particular, only few patients presented with intramedullary lesions. However, our aim was to demonstrate and assess the general feasibility of this novel sequence rather than assessing its potential for the imaging of particular pathologies. Fourthly, due to the heterogeneity of the study cohort and the lack of a standard of reference, we did not evaluate the diagnostic benefit of the sequences. Ultimately, while in line with similar studies such as those by Hu et al ¹ and Cho et al, ² it should be mentioned that we compared a GRE with a TSE sequence. This inherently influences the image contrast and visualization of structures, especially after contrast administration as can be witnessed in the representative image examples. Whether this has an influence on diagnostic performance remains unclear and must be evaluated in future studies.

To conclude, we demonstrate the feasibility of a novel 3D spiral GRE sequence for improved and rapid sagittal CE FS T ₁W spine MRI. The insights gained from this investigation justify the further development of spiral MRI as a novel MRI technique for rapid and improved clinical imaging in a variety of anatomical areas.