Deep‐Learning‐Based Preprocessing for Quantitative Myocardial Perfusion MRI

Scannell, Cian M.; Veta, Mitko; Villa, Adriana; Sammut, Eva; Lee, Jack; Breeuwer, Marcel; Chiribiri, Amedeo

doi:10.1002/jmri.26983

Cited by 68 publications

(54 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, water and fat images were generated using mDixon reconstruction 23 . A rectangular region of interest surrounding the left ventricle was automatically computed using our previously described deep learning-based processing pipeline 25 and the fat images were then registered in an iterative manner using a rigid (translation and rotation) transformation to optimise the mean squared error (MSE) cost function. The reference frame was taken to be the mean frame of the image series.…”

Section: Respiratory Motion Correctionmentioning

confidence: 99%

See 1 more Smart Citation

Feasibility of free-breathing quantitative myocardial perfusion using multi-echo Dixon magnetic resonance imaging

Scannell

Correia

Villa

et al. 2020

Sci Rep

Self Cite

View full text Add to dashboard Cite

Dynamic contrast-enhanced quantitative first-pass perfusion using magnetic resonance imaging enables non-invasive objective assessment of myocardial ischemia without ionizing radiation. However, quantification of perfusion is challenging due to the non-linearity between the magnetic resonance signal intensity and contrast agent concentration. furthermore, respiratory motion during data acquisition precludes quantification of perfusion. While motion correction techniques have been proposed, they have been hampered by the challenge of accounting for dramatic contrast changes during the bolus and long execution times. in this work we investigate the use of a novel free-breathing multi-echo Dixon technique for quantitative myocardial perfusion. the Dixon fat images, unaffected by the dynamic contrast-enhancement, are used to efficiently estimate rigid-body respiratory motion and the computed transformations are applied to the corresponding diagnostic water images. this is followed by a second non-linear correction step using the Dixon water images to remove residual motion. the proposed Dixon motion correction technique was compared to the stateof-the-art technique (spatiotemporal based registration). We demonstrate that the proposed method performs comparably to the state-of-the-art but is significantly faster to execute. Furthermore, the proposed technique can be used to correct for the decay of signal due to T2* effects to improve quantification and additionally, yields fat-free diagnostic images. Myocardial perfusion can be assessed using dynamic MRI during first-pass of a contrast agent 1. While the images are routinely reviewed visually in the clinic, quantification is desirable as it is user-independent 2. Quantification of myocardial perfusion can be challenging mainly due to the non-linearity between the signal intensity and contrast agent concentration at concentrations necessary to observe potential perfusion abnormalities within the myocardium 3. Furthermore, respiratory motion makes the quantification of perfusion difficult, and may even preclude it. Although most commonly used to mitigate respiratory motion, breath-holding for at least 40 s (to cover the full first pass of the contrast bolus) is challenging for many patients, particularly under stress conditions. Moreover, long breath-holds can lead to changes in heart rate, resulting in images being acquired at slightly different cardiac phases thus introducing cardiac motion. Conversely, if patients are allowed to breathe normally, respiratory motion tends to be more regular and shallower when compared to the large gasps that may occur when the patient can no longer hold their breath. Therefore, acquisitions in free-breathing are easier to correct for respiratory motion using image registration compared to breath-holding with intermittent breathing. The problem of registering myocardial perfusion MR images is, however, challenging due to the rapid change in signal intensity during the contrast bolus transit. The dynamic contrast-enhancement invalidate...

show abstract

Section: Respiratory Motion Correctionmentioning

confidence: 99%

“…Perfusion quantification. The perfusion images generated with the different methods were processed automatically using our deep learning processing pipeline 25 and MBF is quantified using the dual-bolus AIF. Pixel-wise time signal intensity curves were then extracted from the myocardial mask.…”

Section: In Vivo Experiments the Proposed Perfusion Technique Was Pementioning

confidence: 99%

Feasibility of free-breathing quantitative myocardial perfusion using multi-echo Dixon magnetic resonance imaging

Scannell

Correia

Villa

et al. 2020

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

“…The published perfusion myocardial segmentation studies, on the contrary, used smaller data sets (e.g. 175 patients in Scannell et al 63 and 1034 patients in Xue et al 64 ). An advantage of this decoupling is the accuracy of AIF detection can be independently optimized and evaluated.…”

Section: Discussionmentioning

confidence: 99%

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Xue

Tseng

Knott

et al. 2020

Magnetic Resonance in Med

View full text Add to dashboard Cite

Purpose Quantification of myocardial perfusion has the potential to improve the detection of regional and global flow reduction. Significant effort has been made to automate the workflow, where one essential step is the arterial input function (AIF) extraction. Failure to accurately identify the left ventricle (LV) prevents AIF estimation required for quantification, therefore high detection accuracy is required. This study presents a robust LV detection method using the convolutional neural network (CNN). Methods CNN models were trained by assembling 25,027 scans (N = 12,984 patients) from three hospitals, seven scanners. Performance was evaluated using a hold‐out test set of 5721 scans (N = 2805 patients). Model inputs were a time series of AIF images (2D+T). Two variations were investigated: (1) two classes (2CS) for background and foreground (LV mask), and (2) three classes (3CS) for background, LV, and RV. The final model was deployed on MRI scanners using the Gadgetron reconstruction software framework. Results Model loading on the MRI scanner took ~340 ms and applying the model took ~180 ms. The 3CS model successfully detected the LV in 99.98% of all test cases (1 failure out of 5721). The mean Dice ratio for 3CS was 0.87 ± 0.08 with 92.0% of all cases having Dice >0.75. The 2CS model gave a lower Dice ratio of 0.82 ± 0.22 (P < 1e−5). There was no significant difference in foot‐time, peak‐time, first‐pass duration, peak value, and area‐under‐curve (P > .2) comparing automatically extracted AIF signals with signals from manually drawn contours. Conclusions A CNN‐based solution to detect the LV blood pool from the arterial input function image series was developed, validated, and deployed. A high LV detection accuracy of 99.98% was achieved.

show abstract

“…Bland-Altman analysis demonstrated a 95% CI for global MBF of 0.05 mmol/ min/g compared with manual labeling, which was sufficient for automated detection and reporting. Prior work on segmenting perfusion (19) used much smaller datasets that resulted in much higher variance. A weighted sum loss function was used in this study and gave good accuracy.…”

Section: Cnn Speed Performancementioning

confidence: 99%

Automated Inline Analysis of Myocardial Perfusion MRI with Deep Learning

Xue¹,

Davies²,

Brown³

et al. 2020

Radiology: Artificial Intelligence

View full text Add to dashboard Cite

To develop a deep neural network-based computational workflow for inline myocardial perfusion analysis that automatically delineates the myocardium, which improves the clinical workflow and offers a "one-click" solution. Materials and Methods: In this retrospective study, consecutive adenosine stress and rest perfusion scans were acquired from three hospitals between October 1, 2018 and February 27, 2019. The training and validation set included 1825 perfusion series from 1034 patients (mean age, 60.6 years 6 14.2 [standard deviation]). The independent test set included 200 scans from 105 patients (mean age, 59.1 years 6 12.5). A convolutional neural network (CNN) model was trained to segment the left ventricular cavity, myocardium, and right ventricle by processing an incoming time series of perfusion images. Model outputs were compared with manual ground truth for accuracy of segmentation and flow measures derived on a global and per-sector basis with t test performed for statistical significance. The trained models were integrated onto MR scanners for effective inference. Results: The mean Dice ratio of automatic and manual segmentation was 0.93 6 0.04. The CNN performed similarly to manual segmentation and flow measures for mean stress myocardial blood flow (MBF; 2.25 mL/min/g 6 0.59 vs 2.24 mL/min/g 6 0.59, P = .94) and mean rest MBF (1.08 mL/min/g 6 0.23 vs 1.07 mL/min/g 6 0.23, P = .83). The per-sector MBF values showed no difference between the CNN and manual assessment (P = .92). A central processing unit-based model inference on the MR scanner took less than 1 second for a typical perfusion scan of three slices. Conclusion: The described CNN was capable of cardiac perfusion mapping and integrated an automated inline implementation on the MR scanner, enabling one-click analysis and reporting in a manner comparable to manual assessment.

show abstract

Deep‐Learning‐Based Preprocessing for Quantitative Myocardial Perfusion MRI

Cited by 68 publications

References 27 publications

Feasibility of free-breathing quantitative myocardial perfusion using multi-echo Dixon magnetic resonance imaging

Feasibility of free-breathing quantitative myocardial perfusion using multi-echo Dixon magnetic resonance imaging

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Automated Inline Analysis of Myocardial Perfusion MRI with Deep Learning

Contact Info

Product

Resources

About