Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Aug 1.
Published in final edited form as: Magn Reson Med. 2024 Mar 12;92(2):751–760. doi: 10.1002/mrm.30083

Inline automatic quality control of 2D phase-contrast flow MR imaging for subject-specific scan time adaptation

Pierre Daudé 1, Rajiv Ramasawmy 1, Ahsan Javed 1, Robert J Lederman 2, Kelvin Chow 3, Adrienne E Campbell-Washburn 1
PMCID: PMC11142871  NIHMSID: NIHMS1977528  PMID: 38469944

Abstract

Purpose

To develop an inline automatic quality control to achieve consistent diagnostic image quality with subject-specific scan time, and to demonstrate this method for 2D phase-contrast flow MRI to reach a pre-determined signal-to-noise ratio (SNR).

Methods

We designed a closed-loop feedback framework between image reconstruction and data acquisition to intermittently check SNR (every 20s) and automatically stop the acquisition when a target SNR is achieved. A free-breathing 2D pseudo-golden angle spiral phase contrast sequence was modified to listen for image quality messages from the reconstructions. Ten healthy volunteers and one patient were imaged at 0.55T. Target SNR was selected based on retrospective analysis of cardiac output error, and performance of the automatic SNR-driven “stop” was assessed inline.

Results

SNR calculation and automated segmentation was feasible within 20s with inline deployment. The SNR-driven acquisition time was 2m39s ± 67s (aorta) and 3m ± 80s (main pulmonary artery) with a min / max acquisition time of 1m43s / 4m52s (aorta) and 1m43s / 5m50s (main pulmonary artery) across 6 healthy volunteers, while ensuring a diagnostic measurement with relative absolute error in quantitative flow measurement lower than 2.1% (aorta) and 6.3% (main pulmonary artery).

Conclusion

The inline quality control enables subject-specific optimized scan times while ensuring consistent diagnostic image quality. The distribution of automated stopping times across the population revealed the value of a subject-specific scan time.

Keywords: quality assessment, inline processing, flow, scan time

Introduction

Conventional MRI uses a fixed acquisition duration to provide acceptable image quality for most patients. Indeed, MRI protocols are optimized to provide diagnostic images as fast as possible while maintaining sufficient information for evaluating the targeted structures. However, image quality can vary between patients. In cardiac imaging, image quality is worse in patients with larger body habitus, patients who move, patients with irregular breathing, or when coils are improperly positioned. Whereas, in other patients, sufficient image quality is achieved quickly and the default scan time is unnecessarily long. Insufficient image quality can result in longer exam time due to scan repetition, and in the worst case, the patient may need to return for a repeat exam. Rather than waiting until the end of the scan to identify poor quality, we hypothesized that suboptimal images could be detected and prevented by intermittent automated inline evaluation during the scan.

In recent years, automatic quality control tools have been designed to detect non diagnostic images. For example, methods have been presented that automatically detect artifacts1,2, identify missing apical and basal slices on cine CMR3, cardiac motion4, detect inter-slice motion along with heart coverage and image contrast estimation5, and segmentation failures6,7. However, these have been primarily applied offline after the exam, or online after a completed acquisition, but never during the acquisition. Meanwhile, low-latency and real-time processing applications have emerged. Jaubert et al.8 implemented low-latency inline flow measurements monitoring. Frueh et al.9 developed real-time landmark detection and tracking. Huttinga et al.10 realized a real-time low-latency 3D non rigid motion field thanks to an offline preparation from a previous acquisition. Furthermore, subject-specific adaptive acquisitions have been proposed. Contijoch et al.11 developed a closed-loop system to choose optimal sampling for segmented cine radial acquisitions within the sequence controller. Breutigam et al.12 developed an automatic feedback for adjusting the post-labeling delay in ASL. Vidya Shankar et al.13 developed a method using automatic slice tracking to follow a catheter for MR-guided cardiac catheterization. A closed-loop image-based assessment of scan quality will bridge modern image reconstruction techniques and novel computer vision programs with the clinical workflows to maximize the effectiveness of these tools.

We proposed an inline automatic quality control based on a generalizable closed-loop feedback framework between image reconstruction and data acquisition to efficiently achieve consistent diagnostic image quality based on a pre-determined metric. Signal-to-noise ratio (SNR) is directly related to the confidence of flow measurements14. Therefore, we applied this framework for cardiac flow measurements with 2D phase-contrast MRI using an SNR threshold as a stop criterion for achieving accurate flow measurement across subjects with subject-specific imaging duration.

Methods

Theory

We implemented closed-loop feedback between the image reconstruction and data acquisition. Two-way communication between the image acquisition software and the Gadgetron15 reconstruction software is handled using “FIRE” research framework16 (Siemens Healthineers AG, Erlangen, Germany), which supports streaming of MRI raw data (MRD)17. Periodically during the acquisition, the quality of the image is automatically evaluated, and a feedback message is sent to the sequence controller, i.e., a small data packet which reports the quality metric. If the quality requirement is met, the acquisition will automatically stop itself; if not, the acquisition will continue.

Automatic SNR-driven inline quality control workflow

Figure 1 provides a schematic diagram of our workflow. We applied the closed-loop feedback framework for 2D pseudo-golden spiral phase contrast flow acquisition that will automatically stop when the target SNR is achieved.

Figure 1 :

Figure 1 :

Open in a new tab

Schematic diagram demonstrating the inline automatic quality control based on assessment of SNR and feedback messaging. Every 20s, SNR maps are generated by Gadgetron and the SNRfeedback is extracted automatically in the targeted tissue and sent using FIRE framework to the sequence controller. As soon as sufficient SNR is reached in the targeted tissue, a stop message is sent to the acquisition and cardiac cycle-resolved magnitude, phase images are reconstructed with their segmentations.

Every 20s, SNR maps were estimated using the pseudo-replica method18 which is a Monte Carlo approach that emulates the gold standard repeated image signal-to-noise measurement (scan, rescan). Random gaussian noise, based on measured noise statistics is added to k-space before image reconstruction, generating a pseudo replica image. The process is repeated N times (N=100, here), with different synthetic noise, resulting in a stack of independent pseudo-replica images. Then the pixel-wise SNR is calculated using the ratio of the image over the standard deviation of the image replicas.

To minimize computation time, data from the entire cardiac cycle was used for interim SNR analysis instead of cardiac resolved SNR maps. This time-averaged SNR was correlated with cardiac-resolved SNR to validate this approach. The SNR of the targeted tissue, either ascending aorta (AAo) or main pulmonary artery (MPA), was then extracted by automatic segmentation using a nnUNet19 and sent to the sequence controller. The operator selected an image reconstruction pipeline from a drop-down menu in order to ensure that the appropriate segmentation network for AAo or MPA was used inline.

The stopping criterion was defined as a minimum SNR threshold, i.e., after the target SNR is achieved, the sequence will finish executing the current loop and stop. If the target SNR never reached the threshold, the acquisition will stop at the maximum number of prescribed averages. At the end of the scan, a higher quality image reconstruction was performed inline using Gadgetron. The acquired data was retrospectively self-gated to 25 cardiac frames and reconstructed using T-CG-SENSE with a spatial and temporal constraints (λs=0.1, λt=1) and images are returned to the scanner host along with the segmentation of the target tissue for each frame using a nnUNet. The T-CG-SENSE reconstruction used to generate the final images following automatic stop will have inherently higher SNR than the NUFFT reconstruction used for the rapid inline quality assessment.

Healthy volunteer and patient imaging

Institutional Review Board approval and written informed consent from all study participants was obtained (ClinicalTrails.gov identifier NCT03331380). Ten healthy volunteers (body mass index (BMI)=25.4±2.5, age=30±8 years, male/female=4/6) were imaged on a 0.55T MRI scanner (MAGNETOM Free.Max, Siemens Healthineers AG, Erlangen, Germany) with prototype gradient coil. We used the vendor body array and the spine coil array, totaling up to 21 channels. A free-breathing gradient echo single slice pseudo-golden-angle spiral flow sequence (TE/TR=2.0/10.5ms, FA=25°, (1.7mm)2 resolution, 8mm slice thickness, through plane venc=200cm/s, FOV=(384mm)2) was modified to listen for and process the feedback messages. Two scans were performed in each subject: 1) full acquisition time and 2) a SNR-driven automatic stop acquisition with a maximum scan time of 4min50s (AAo) or 6min10s (MPA). Interim pseudo-replica SNR estimation, automatic segmentation, and image reconstructions were performed using Gadgetron on a computer equipped with 4 GPUs (NVIDIA A100-SXM, 80Gb) and 128 CPUs cores (2x AMD EPYC 7H12 64-core processors).

To test the clinical robustness of the inline automatic quality control, one patient (BMI=29.3, age=73 years old, female) with a prosthetic aortic valve was also recruited.

Image Quality Assessment: automatic segmentation

We chose the nnUNet19 framework to automatically segment the ascending aorta and the main pulmonary artery in our workflow. The nnUNet has already shown good performance for CMR segmentation challenge20. To train and evaluate neural network models, a mono-centric database with 138 patients was defined, divided into 128 patients for training and validation and 10 for testing, all of whom underwent a cardiac MRI exam at 1.5 T (MAGNETOM Aera, Siemens Healthineers AG, Erlangen, Germany) and/or 0.55 T (prototype MAGNETOM Aera or MAGNETOM Free.Max, Siemens Healthineers AG, Erlangen, Germany). For all enrolled subjects, 2D cartesian phase contrast MRI were acquired with the following parameters for 0.55T: TE/TR=4.3/14.1ms, FA=30°, (1.56–2.2 mm)2 resolution, 6 mm slice thickness, through plane venc=200 cm/s, FOV=270×360 mm2, 3 averages, and at 1.5T: TE/TR=2.7/10.0ms, flip angle (FA)=20°, (1.40–1.56 mm)2 resolution, 6mm slice thickness, through plane, venc=200 cm/s, FOV=270×360 mm2, 3 averages. As a result, 454 acquisitions were obtained, as some patients had more than one acquisition per targeted tissue, divided in 59 (AO: 30 / MPA: 29) at 1.5T and 395 (AO: 200, MPA: 195) at 0.55T. For reference, AAo and MPA were segmented for all 25 retrogated cardiac phases using an automatic tool provided by a commercial software (suiteHEART software version 5.1.0; NeoSoft LLC) and revised by experts.

The default 2D nnUNet networks were trained for AAo and MPA independently with five-fold cross-validation, stratified by field strength and subject independence. The models were trained over 1000 epochs with a batch size of 106 using a stochastic gradient descent with Nesterov momentum (mu=0.99) and an initial learning rate of 0.01. Data augmentation was performed on the fly. All images were resampled to (1.875 mm)2 in-plane resolution with an interpolation of order 3.

Image Quality Assessment: Stopping criterion

The stopping criterion was defined by a target SNR threshold. A retrospective analysis of the SNR feedback messages sent every 20s during the full acquisition duration was conducted in 10 healthy volunteers to determine an optimal SNR stopping threshold. The optimal stopping threshold was chosen to produce an accurate measurement of cardiac output (CO), defined as <5% absolute relative error compared to the full acquisition time, which was chosen to be long enough to provide precise measurement (4min50s (AAo) or 6min10s (MPA)). This clinical criterion has been chosen empirically to represent the concept of a diagnostic measurement criterion, but it is equivalent to previously reported inter-site variability of 5% in cardiac exams21. This stopping threshold was then applied inline for 6 volunteers.

Statistical Analysis and evaluations metrics

For the automatic segmentation, the nnUNet network was evaluated on a test dataset including 10 healthy volunteers using conventional segmentation metrics (Dice Similarity Coefficient, Haussdorff distance, absolute relative surface error). Flow analysis was performed using MATLAB R2021a (The MathWorks, Natick, Massachusetts, USA). Statistical analysis has been conducted using R (version 4.3.0).

Results

Feasibility of inline quality control

For the automatic segmentation, 2D nnUNet provided accurate segmentation of the AAo and MPA with mean Dice Similarity Coefficient of 0.95±0.02. Segmentation cross-validation results are provided in Supporting information Table S1 and Supporting information Table S2. The automatic segmentation required 1.08±0.09s and SNR map computation (100 pseudo-replicas) required 12.98 ± 5.49s. The inline control computation time grows throughout the scan as more data is collected, starting at 4s and ending at 20s for the full acquisition. The SNR map reconstruction time increased based on the density compensation and NUFFT operations that scale with the number of spiral shots. SNR estimation using the pseudo replica method requires a trade-off between computer processing time and SNR measurement precision according to the number of replicas (Supporting Information Figure S1). The total latency of the computation was always compatible within the 20s assessment interval, which is crucial to avoid a growing lag between acquisition and image quality assessment that would result in incorrect stopping time and protocol inefficiency. The final images reconstruction using T-CG-SENSE was done inline using Gadgetron, such that final images return to the scanner after the acquisition, and required 1 min. Our reconstruction, SNR estimation and segmentation implementations are available open-source using Gadgetron (https://github.com/NHLBI-MR/SNR-driven-flow).

SNR-based stopping criterion: retrospective analysis

Retrospective analysis demonstrated that by choosing an SNR threshold of 175 for AAo and 140 for MPA (Figure 2.A/B.), we ensured sufficient image quality to maintain accurate quantitative CO measurements with an error ≤ 5% relative to the full duration measurement (4min50s for AAo and 6min10s for MPA). These SNR threshold values are higher than typical cardiac resolved 2D phase contrast because the intermittent SNR assessment is made on the time-averaged data (i.e., un-binned data, not cardiac-resolved). The time-averaged SNR correlated well (R2=0.99) with the mean cardiac-resolved SNR (Supporting Information Figure S2). As expected, the time-averaged and cardiac-resolved SNR values are proportional by a factor corresponding to the square root of the number of cardiac frames.

Figure 2 :

Figure 2 :

Open in a new tab

Retrospective analysis of absolute error in cardiac output for ascending aorta (A) and main pulmonary artery (B) to determine the SNR threshold used to stop the acquisition and then applied it retrospectively across ten healthy volunteers for the ascending aorta (C) and main pulmonary artery (D). Cardiac output error is calculated relative to the data from the full scan duration (4m50s for AAo and 6min10s for MPA). SNR is calculated from un-binned images, equivalent to the sum of all 25 cardiac frames. Open markers indicate an error <5% and the region highlighted yellow shows unstable cardiac output measurements. In C./D., the dotted horizontal lines represent the selected SNR threshold and the dashed vertical lines represent the “stop” message for each healthy volunteer. For one subject (HV 2), the optimal SNR threshold would have never been reached in the ascending aorta acquisition. HV=healthy volunteer.

By applying these optimal SNR thresholds retrospectively (Figure 2.C./D.), acquisition would have automatically stopped at 2min41s±62s and 2min39s±63s, saving 41±23% and 57±18% of scan time for AAo and MPA, respectively. Compared to the full acquisition, the retrospective stopped acquisition had a CO%error of 1.3±1.6% / 1.4±1.1% with a maximum of 5.0%/3.3% for AAo/MPA. For one healthy volunteer of BMI=28.8, the full acquisition (4m50s) in the AAo did not reach the target SNR (max SNR=164).

Inline application of SNR-based stopping criterion

Deploying the closed-loop feedback inline, target SNR was reached at 2min27s±53s/2min50s±69s with SNR=181±5/145±3 (Figure 3). The acquisition was stopped at 2min39s±67s (range 1min43s to 4min52s)/ 3min±80s (range 1min43s, 5min50s) saving 43% / 51% of the scan duration for AAo/ MPA, retrospectively. Compared to the full acquisition, the early stopped acquisition CO%error was 2.1±2.0%/6.3±3.7% with a maximum of 5.4%/11.5% for AAo/MPA. (Figure 3).

Figure 3 :

Figure 3 :

Open in a new tab

Example SNR-driven quality control applied inline. The final SNR maps along with the magnitude and phase images after automatic stop, the resulting flow curves compared to the full acquisition time, and the resulting error in cardiac output are provided. Results are displayed for (A,B) the ascending aorta, (B,C) the main pulmonary artery of different healthy volunteers. The contour of the automatic segmentation is displayed in red.

Figure 4 compares MPA flow maging and measurements between a fixed 2-minute acquisition and SNR-driven automatic stop acquisition, both compared to the reference the full acquisition (6min10s), in two different healthy subjects. Compared to the reference, the fixed 2-minute acquisition provided in the top row inaccurate flow measurements and 5 minutes was the required acquisition time determined by the SNR-driven automatic stop method. In the other subject, 2 minutes was sufficient and, in fact, the SNR-driven automatic stop occurred at 1min40s. This illustrates the added value of SNR-driven acquisition to reduce protocol inefficiency.

Figure 4 :

Figure 4 :

Open in a new tab

Comparison of phase images and flow measurements between a fixed 2-minute acquisition and SNR-driven automatic stop acquisition (SNR threshold=140 for main pulmonary artery) with as the reference the full acquisition (6min10 s). In the top row (A.), the SNR-driven scan stopped the acquisition at 5min for a female HV of BMI=28.8 whereas in the bottom row (B.), the acquisition stopped at 1min40s for a male HV of BMI=23.8. This illustrates the value of an SNR-driven stop criterion.

As illustrated in Figure 5, the SNR-driven automatic stop was also deployed inline in one patient with a prosthetic aortic valve and visible metallic artifacts caused by sternal wires. The acquisition stopped at 2min with a SNR = 178 and it generated diagnostic flow measurements with a CO%error of 3.7% compared to the full acquisition. The automatic segmentation of the aorta for the inline quality control was robust to an artifact induced by a metallic implant.

Figure 5 :

Figure 5 :

Open in a new tab

Inline automatic quality control of the ascending aorta of a patient, displaying the SNR map, and magnitude and phase images after early stop and flow measurements compared to the full acquisition time. The automatic segmentation for the quality control performed well even in the presence of an artifact induced by a sternal wire (blue arrow).

Discussion

This study aimed to develop a framework for inline automatic quality control based on a pre-determined image quality metric. We provided an illustrative example of an adaptive subject-specific MRI acquisition time for 2D phase contrast MR flow measurements in the heart with an SNR-driven stop criterion. We demonstrated a generalizable framework for intermittent closed-loop communication between the image reconstruction software and the data acquisition software, sending messages about image quality inline to the sequence controller in this application. For pseudo-golden angle spiral 2D phase contrast flow, the standard deviation of automatic stop times (±67s for AAo, ±80s for MPA) revealed the value of subject-specific acquisition time for consistent image quality. It resulted in saving approximately 50% of acquisition time while ensuring a diagnostic measurement with average error in quantitative measurements <2.1% / 6.3% for AAo / MPA which is in the order of variation for scan-rescan (3%) or inter-site variability (5%) found in the literature21.

We chose 2D phase contrast flow measurement as an example application because these scans can be time-consuming for specific clinical indications. For example, during MRI-guided invasive catheterization procedures, flow is measured repeatedly while the patient is instrumented. Similarly, several flow measurements may be required in pediatric and adult patients with congenital heart disease. Therefore, optimal patient-specific scan durations of 2D phase contrast flow measurements may improve the efficiency of these exams. We used SNR as the image quality metric for 2D phase contrast MRI. SNR is directly linked to velocity-noise ratio (VNR)22, therefore the standard deviation of the velocity (σv) is given by this formula:

σv=2πVencSNR

and enables calculation of the confidence interval of the flow measurement14, and therefore is an appropriate metric for this application. We chose a pseudo-golden angle spiral sequence to provide flexibility to have fine control over the automatic stopping time, compared to Cartesian imaging where a full image average requires almost 1 minute. 2D spiral phase contrast measurements have previously been validated against Cartesian sequences23. The SNR thresholds used here were specifically optimized for our sequence and application and should be adapted for each application.

Inline quality control imposed the need for low-latency SNR calculation. For speed purposes, only the SNR of the time-averaged data was reconstructed in our study because it is correlated (R2=0.99) to the mean cardiac frame resolved SNR. The pseudo replica method used to estimate SNR is computationally intensive for non-Cartesian imaging and required 12.98 ± 5.49s and automatic segmentation required 1.08 ± 0.09s. The required computation time restricted the maximum frequency of intermittent image quality assessment (fixed at 20s for our study). Instead, another possible implementation would be to calculate SNR once and extrapolate the predicted SNR versus scan time.

Our implementation of SNR calculation is available open-source using Gadgetron (https://github.com/NHLBI-MR/SNR-driven-flow). We used FIRE, which is a proprietary Siemens package, for messaging between the reconstruction and acquisition, but this could be similarly achieved using the Gadgetron streaming capabilities, custom implementation, and/or vendor-provided software. This technology could benefit low latency offline automatic image quality control present in the literature2–7,24,25 by enabling rapid inline implementation.

The main limitation affecting our method is that inline quality control is computationally intensive due to the use of the pseudo-replica method for SNR map estimation, restraining therefore the deployment of this method in clinical settings with limited resources. It should also be noted that the choice of the clinical stopping criterion (CO%error ≤ 5%) was only intended as a proof-of-concept to illustrate the concept of an automatic stop when a certain level of diagnostic certainty was achieved. Defining such criterion from widespread application would require a larger cohort with diversity of patients profiles which is beyond the scope of our study. In addition, the use of a temporal constraint in the image reconstruction may influence the flow measurement, especially when SNR is insufficient.

The concept of automatic scan termination based on a pre-determined image quality metric is widely applicable. Indeed, the inline quality control has been designed for a single slice acquisition and could be extended to a multi-slice approach where each slice could have their own stopping time (based on a single SNR threshold) in order to ensure consistent quality across the whole volume. Moreover, the quality control could be extended with multiple targets, multiple image quality metrics thereby increasing the complexity to design the stopping criterion. To alleviate this issue, the image quality could be based on a singular quality score or a binary classification (i.e., continue or stopping scanning). However, those approaches will likely require large database with expert annotations to train a classification network. For other applications, different image quality metrics may be relevant such as contrast-to-noise ratio (CNR) or sharpness metric. We applied this method on a contemporary 0.55T system where, given the intrinsically lower SNR, the ability to ensure consistent image quality is desirable. However, this approach is also valuable at other field strengths. Moreover, the closed-loop feedback workflow could be also extended to detect artifacts in real-time and correct them by automatically adjusting sequence parameters. For instance, a feedback module that detects velocity aliasing and corrects it in real-time may enable subject-specific optimal venc for our 2D phase contrast application.

Conclusions

We demonstrated a framework for automatic real-time quality control for subject-specific acquisition timing adaptation and applied it using SNR-driven imaging on phase contrast MRI as a proof-of-concept. We observed a wide distribution of automatic stopping times across the population, which revealed the value of subject-specific acquisition time for consistent image quality.

Supplementary Material

Supinfo

Supporting Information Figure S1 : A. Total computational processing time of the inline quality control depending of the number of pseudo replicas. B. SNR absolute error depending of the number of replicas. The reference corresponds to the mean SNR across 10 repeated measurements for each inline quality control with 125 replicas.

Supporting Information Figure S2: Regression plot between SNR-time-averaged scale by number of cardiac frames against mean cardiac resolved SNR again using retrospective database of cartesian acquisitions reconstructed at different averages.

Supporting information Table S1 : Mean values and standard deviations (in parenthesis) of segmentation results on the test dataset.

Supporting information Table S2 : Mean values and standard deviations (in parenthesis) of segmentation results for the cross-validation.

Acknowledgements

Many thanks to Christine Mancini and Matthew Thurston for their expertise for revising segmentation.

Funding

This work was supported by the National Heart, Lung, and Blood Institute (NHLBI) Division of Intramural Research (Z01-HL006257, Z01-HL006213).

Footnotes

Ethics declarations

Ethics approval and consent to participate

Imaging was approved by the local Institutional Review Board (ClinicalTrials.gov identifier NCT03331380) and written informed consent was obtained from all participants.

Competing interests

The authors are investigators on a US Government Cooperative Research and Development Agreement (CRADA) with Siemens Healthcare.

Data availability statement

Data are available from the corresponding author on reasonable request. Code is freely available for download at https://github.com/NHLBI-MR/SNR-driven-flow.

References

  • 1.Küstner T, Gatidis S, Liebgott A, et al. A machine-learning framework for automatic reference-free quality assessment in MRI. Magnetic Resonance Imaging. 2018;53:134–147. doi: 10.1016/j.mri.2018.07.003 [DOI] [PubMed] [Google Scholar]
  • 2.Manso Jimeno M, Ravi KS, Jin Z, Oyekunle D, Ogbole G, Geethanath S. ArtifactID: Identifying artifacts in low-field MRI of the brain using deep learning. Magnetic Resonance Imaging. 2022;89:42–48. doi: 10.1016/j.mri.2022.02.002 [DOI] [PubMed] [Google Scholar]
  • 3.Zhang L, Gooya A, Dong B, et al. Automated Quality Assessment of Cardiac MR Images Using Convolutional Neural Networks. In: Tsaftaris SA, Gooya A, Frangi AF, Prince JL, eds. Simulation and Synthesis in Medical Imaging. Vol 9968. Lecture Notes in Computer Science. Cham: Springer International Publishing; 2016:138–145. doi: 10.1007/978-3-319-46630-9_14 [DOI] [Google Scholar]
  • 4.Oksuz I, Ruijsink B, Puyol-Antón E, et al. Automatic CNN-based detection of cardiac MR motion artefacts using k-space data augmentation and curriculum learning. Medical Image Analysis. 2019;55:136–147. doi: 10.1016/j.media.2019.04.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Tarroni G, Oktay O, Bai W, et al. Learning-Based Quality Control for Cardiac MR Images. IEEE Trans Med Imaging. 2019;38(5):1127–1138. doi: 10.1109/TMI.2018.2878509 [DOI] [PubMed] [Google Scholar]
  • 6.Robinson R, Valindria VV, Bai W, et al. Automated quality control in image segmentation: application to the UK Biobank cardiovascular magnetic resonance imaging study. J Cardiovasc Magn Reson. 2019;21(1):18. doi: 10.1186/s12968-019-0523-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Fournel J, Bartoli A, Bendahan D, et al. Medical image segmentation automatic quality control: A multi-dimensional approach. Medical Image Analysis. 2021;74:102213. doi: 10.1016/j.media.2021.102213 [DOI] [PubMed] [Google Scholar]
  • 8.Jaubert O, Montalt-Tordera J, Brown J, et al. FReSCO : Flow Reconstruction and Segmentation for low-latency Cardiac Output monitoring using deep artifact suppression and segmentation. Magnetic Resonance in Med 2022;88(5):2179–2189. doi: 10.1002/mrm.29374 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Frueh M, Schilling A, Gatidis S, Kuestner T. Real Time Landmark Detection for Within- and Cross Subject Tracking With Minimal Human Supervision. IEEE Access. 2022;10:81192–81202. doi: 10.1109/ACCESS.2022.3195211 [DOI] [Google Scholar]
  • 10.Huttinga NRF, Bruijnen T, Van Den Berg CAT, Sbrizzi A. Real-Time Non-Rigid 3D Respiratory Motion Estimation for MR-Guided Radiotherapy Using MR-MOTUS. IEEE Trans Med Imaging. 2022;41(2):332–346. doi: 10.1109/TMI.2021.3112818 [DOI] [PubMed] [Google Scholar]
  • 11.Contijoch F, Han Y, Kamesh Iyer S, et al. Closed-loop control of k-space sampling via physiologic feedback for cine MRI. Bauer WR, ed. PLoS ONE. 2020;15(12):e0244286. doi: 10.1371/journal.pone.0244286 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Breutigam NJ, Hoinkiss DC, Konstandin S, et al. Subject-specific timing adaption in time-encoded arterial spin labeling imaging. Magn Reson Mater Phy September 2023. doi: 10.1007/s10334-023-01121-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Vidya Shankar R, Huang L, Neji R, et al. Real-time automatic image-based slice tracking of gadolinium-filled balloon wedge catheter during MR -guided cardiac catheterization: A proof-of-concept study. Magnetic Resonance in Med 2024;91(1):388–397. doi: 10.1002/mrm.29822 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Hansen MS, Olivieri LJ, O’Brien K, Cross RR, Inati SJ, Kellman P. Method for calculating confidence intervals for phase contrast flow measurements. J Cardiovasc Magn Reson 2014;16(1):46. doi: 10.1186/1532-429X-16-46 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Hansen MS, Sørensen TS. Gadgetron: An open source framework for medical image reconstruction: Gadgetron. Magn Reson Med 2013;69(6):1768–1776. doi: 10.1002/mrm.24389 [DOI] [PubMed] [Google Scholar]
  • 16.Chow K, Kellman P, Xue H. Prototyping Image Reconstruction and Analysis with FIRE. In: SCMR. Virtual Scientific Sessions.; 2021:838972. [Google Scholar]
  • 17.Inati SJ, Naegele JD, Zwart NR, et al. ISMRM Raw data format: A proposed standard for MRI raw datasets. Magnetic Resonance in Med 2017;77(1):411–421. doi: 10.1002/mrm.26089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Robson PM, Grant AK, Madhuranthakam AJ, Lattanzi R, Sodickson DK, McKenzie CA. Comprehensive quantification of signal-to-noise ratio and g -factor for image-based and k -space-based parallel imaging reconstructions. Magn Reson Med 2008;60(4):895–907. doi: 10.1002/mrm.21728 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Isensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods. December 2020. doi: 10.1038/s41592-020-01008-z [DOI] [PubMed] [Google Scholar]
  • 20.Bernard O, Lalande A, Zotti C, et al. Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis: Is the Problem Solved? IEEE Trans Med Imaging. 2018;37(11):2514–2525. doi: 10.1109/TMI.2018.2837502 [DOI] [PubMed] [Google Scholar]
  • 21.Zhang Y, Tang XQ, Ma Y, et al. A test–retest multisite reproducibility study of cardiovascular four-dimensional flow MRI without respiratory gating. Clinical Radiology. 2021;76(3):236.e1–236.e8. doi: 10.1016/j.crad.2020.09.014 [DOI] [PubMed] [Google Scholar]
  • 22.Callaghan FM, Kozor R, Sherrah AG, et al. Use of multi-velocity encoding 4D flow MRI to improve quantification of flow patterns in the aorta. Magnetic Resonance Imaging. 2016;43(2):352–363. doi: 10.1002/jmri.24991 [DOI] [PubMed] [Google Scholar]
  • 23.Kowalik GT, Knight D, Steeden JA, Muthurangu V. Perturbed spiral real-time phase-contrast MR with compressive sensing reconstruction for assessment of flow in children. Magnetic Resonance in Med 2020;83(6):2077–2091. doi: 10.1002/mrm.28065 [DOI] [PubMed] [Google Scholar]
  • 24.Piccini D, Demesmaeker R, Heerfordt J, et al. Deep Learning to Automate Reference-Free Image Quality Assessment of Whole-Heart MR Images. Radiology: Artificial Intelligence. 2020;2(3):e190123. doi: 10.1148/ryai.2020190123 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Lei K, Syed AB, Zhu X, Pauly JM, Vasanawala SS. Artifact- and content-specific quality assessment for MRI with image rulers. Medical Image Analysis. 2022;77:102344. doi: 10.1016/j.media.2021.102344 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supinfo

Supporting Information Figure S1 : A. Total computational processing time of the inline quality control depending of the number of pseudo replicas. B. SNR absolute error depending of the number of replicas. The reference corresponds to the mean SNR across 10 repeated measurements for each inline quality control with 125 replicas.

Supporting Information Figure S2: Regression plot between SNR-time-averaged scale by number of cardiac frames against mean cardiac resolved SNR again using retrospective database of cartesian acquisitions reconstructed at different averages.

Supporting information Table S1 : Mean values and standard deviations (in parenthesis) of segmentation results on the test dataset.

Supporting information Table S2 : Mean values and standard deviations (in parenthesis) of segmentation results for the cross-validation.

NIHMS1977528-supplement-Supinfo.docx (191.9KB, docx)

Data Availability Statement

Data are available from the corresponding author on reasonable request. Code is freely available for download at https://github.com/NHLBI-MR/SNR-driven-flow.

RESOURCES