scholarly journals Fully-automated global and segmental strain analysis of DENSE cardiovascular magnetic resonance using deep learning for segmentation and phase unwrapping

2021 ◽  
Vol 23 (1) ◽  
Author(s):  
Sona Ghadimi ◽  
Daniel A. Auger ◽  
Xue Feng ◽  
Changyu Sun ◽  
Craig H. Meyer ◽  
...  
Keyword(s):  
Cardiovascular Magnetic Resonance ◽  
Deep Learning ◽  
Magnetic Resonance ◽  
Healthy Subjects ◽  
Circumferential Strain ◽  
Strain Analysis ◽  
Phase Unwrapping ◽  
Insertion Point ◽  
Fully Automatic ◽  
Myocardial Segmentation

Abstract Background Cardiovascular magnetic resonance (CMR) cine displacement encoding with stimulated echoes (DENSE) measures heart motion by encoding myocardial displacement into the signal phase, facilitating high accuracy and reproducibility of global and segmental myocardial strain and providing benefits in clinical performance. While conventional methods for strain analysis of DENSE images are faster than those for myocardial tagging, they still require manual user assistance. The present study developed and evaluated deep learning methods for fully-automatic DENSE strain analysis. Methods Convolutional neural networks (CNNs) were developed and trained to (a) identify the left-ventricular (LV) epicardial and endocardial borders, (b) identify the anterior right-ventricular (RV)-LV insertion point, and (c) perform phase unwrapping. Subsequent conventional automatic steps were employed to compute strain. The networks were trained using 12,415 short-axis DENSE images from 45 healthy subjects and 19 heart disease patients and were tested using 10,510 images from 25 healthy subjects and 19 patients. Each individual CNN was evaluated, and the end-to-end fully-automatic deep learning pipeline was compared to conventional user-assisted DENSE analysis using linear correlation and Bland Altman analysis of circumferential strain. Results LV myocardial segmentation U-Nets achieved a DICE similarity coefficient of 0.87 ± 0.04, a Hausdorff distance of 2.7 ± 1.0 pixels, and a mean surface distance of 0.41 ± 0.29 pixels in comparison with manual LV myocardial segmentation by an expert. The anterior RV-LV insertion point was detected within 1.38 ± 0.9 pixels compared to manually annotated data. The phase-unwrapping U-Net had similar or lower mean squared error vs. ground-truth data compared to the conventional path-following method for images with typical signal-to-noise ratio (SNR) or low SNR (p < 0.05), respectively. Bland–Altman analyses showed biases of 0.00 ± 0.03 and limits of agreement of − 0.04 to 0.05 or better for deep learning-based fully-automatic global and segmental end-systolic circumferential strain vs. conventional user-assisted methods. Conclusions Deep learning enables fully-automatic global and segmental circumferential strain analysis of DENSE CMR providing excellent agreement with conventional user-assisted methods. Deep learning-based automatic strain analysis may facilitate greater clinical use of DENSE for the quantification of global and segmental strain in patients with cardiac disease.

scholarly journals Comparison between conventional and compressed sensing cine cardiovascular magnetic resonance for feature tracking global circumferential strain assessment

2021 ◽  
Vol 23 (1) ◽  
Author(s):  
Tomoyuki Kido ◽  
Kuniaki Hirai ◽  
Ryo Ogawa ◽  
Yuki Tanabe ◽  
Masashi Nakamura ◽  
...  
Keyword(s):  
Cardiovascular Magnetic Resonance ◽  
Magnetic Resonance ◽  
Compressed Sensing ◽  
Feature Tracking ◽  
Circumferential Strain ◽  
Strain Analysis ◽  
Global Circumferential Strain ◽  
Volume Assessment ◽  
Lv Volume ◽  
Cine Cmr

Abstract Background Feature tracking (FT) has become an established tool for cardiovascular magnetic resonance (CMR)-based strain analysis. Recently, the compressed sensing (CS) technique has been applied to cine CMR, which has drastically reduced its acquisition time. However, the effects of CS imaging on FT strain analysis need to be carefully studied. This study aimed to investigate the use of CS cine CMR for FT strain analysis compared to conventional cine CMR. Methods Sixty-five patients with different left ventricular (LV) pathologies underwent both retrospective conventional cine CMR and prospective CS cine CMR using a prototype sequence with the comparable temporal and spatial resolution at 3 T. Eight short-axis cine images covering the entire LV were obtained and used for LV volume assessment and FT strain analysis. Prospective CS cine CMR data over 1.5 heartbeats were acquired to capture the complete end-diastolic data between the first and second heartbeats. LV volume assessment and FT strain analysis were performed using a dedicated software (ci42; Circle Cardiovasacular Imaging, Calgary, Canada), and the global circumferential strain (GCS) and GCS rate were calculated from both cine CMR sequences. Results There were no significant differences in the GCS (− 17.1% [− 11.7, − 19.5] vs. − 16.1% [− 11.9, − 19.3; p = 0.508) and GCS rate (− 0.8 [− 0.6, − 1.0] vs. − 0.8 [− 0.7, − 1.0]; p = 0.587) obtained using conventional and CS cine CMR. The GCS obtained using both methods showed excellent agreement (y = 0.99x − 0.24; r = 0.95; p < 0.001). The Bland–Altman analysis revealed that the mean difference in the GCS between the conventional and CS cine CMR was 0.1% with limits of agreement between -2.8% and 3.0%. No significant differences were found in all LV volume assessment between both types of cine CMR. Conclusion CS cine CMR could be used for GCS assessment by CMR-FT as well as conventional cine CMR. This finding further enhances the clinical utility of high-speed CS cine CMR imaging.

Abstract P345: A MRI Displacement-based Deep-learning Semantic Segmentation Tool For Left-ventricular Longitudinal Strain Analysis In Cardiotoxicity

2021 ◽  
Vol 129 (Suppl_1) ◽  
Author(s):  
Julia Kar ◽  
Michael V Cohen ◽  
Teja Poorsala ◽  
Samuel A McQuiston ◽  
Cheri Revere ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Deep Learning ◽  
Cancer Patients ◽  
Healthy Subjects ◽  
Longitudinal Strain ◽  
Strain Analysis ◽  
Semantic Segmentation ◽  
Left Ventricular ◽  
Phase Unwrapping ◽  
Breast Cancer Patients

Global longitudinal strain (GLS) computed in the left-ventricle (LV) is an established metric for detecting cardiotoxicity in breast cancer patients treated with antineoplastic agents. The purpose of this study was to develop a novel, MRI-based, deep-learning semantic segmentation tool that automates the phase-unwrapping for LV displacement computation in GLS. Strain analysis via phase-unwrapping was conducted on 30 breast cancer patients and 30 healthy females acquired with the Displacement Encoding with Stimulated Echoes (DENSE) sequence. A ResNet-50 deep convolutional neural network (DCNN) architecture for automated phase-unwrapping, a previously validated ResNet-50 DCNN for chamber quantification and the Radial Point Interpolation Method were used for GLS computation (Figure 1). The DCNN's performance was measured with F1 and Dice scores, and validated in comparison to the robust transport of intensity equation (RTIE) and quality guided phase-unwrapping (QGPU) conventional algorithms. The three techniques were compared by intraclass correlation coefficient with Cronbach’s alpha (C-alpha) index. Classification accuracy with the DCNN was F1 score of 0.92 and Dice score of 0.89. The GLS results from RTIE, QGPU and DCNN were -16.0 ± 2%, -16.1 ± 3% and -15.9 ± 3% (C-alpha = 0.89) for patients and -18.9 ± 3%, -19.0 ± 4% and -18.9 ± 3% (C-alpha = 0.92) for healthy subjects. Comparable validation results from the three techniques show the feasibility of a DCNN-based approach to LV displacement and GLS analysis. The dissimilarities between patients and healthy subjects demonstrate that DCNN-based GLS computation may detect LV abnormalities related to cardiotoxicity.

scholarly journals Clinical validation of three cardiovascular magnetic resonance techniques to measure strain and torsion in patients with suspected coronary artery disease

2020 ◽  
Vol 22 (1) ◽  
Author(s):  
Johan Kihlberg ◽  
Vikas Gupta ◽  
Henrik Haraldsson ◽  
Andreas Sigfridsson ◽  
Sebastian I. Sarvari ◽  
...  
Keyword(s):  
Coronary Artery Disease ◽  
Cardiovascular Magnetic Resonance ◽  
Coronary Artery ◽  
Magnetic Resonance ◽  
Circumferential Strain ◽  
Myocardial Strain ◽  
Radial Strain ◽  
Left Ventricular ◽  
Operating Characteristics ◽  
Artery Disease

Abstract Background Several cardiovascular magnetic resonance (CMR) techniques can measure myocardial strain and torsion with high accuracy. The purpose of this study was to compare displacement encoding with stimulated echoes (DENSE), tagging and feature tracking (FT) for measuring circumferential and radial myocardial strain and myocardial torsion in order to assess myocardial function and infarct scar burden both at a global and at a segmental level. Method 116 patients with a high likelihood of coronary artery disease (European SCORE > 15%) underwent CMR examination including cine images, tagging, DENSE and late gadolinium enhancement (LGE) in the short axis direction. In total, 97 patients had signs of myocardial disease and 19 had no abnormalities in terms of left ventricular (LV) wall mass index, LV ejection fraction, wall motion, LGE or a history of myocardial infarction. Thirty-four patients had myocardial infarct scar with a transmural LGE extent (transmurality) that exceeded 50% of the wall thickness in at least one segment. Global circumferential strain (GCS) and global radial strain (GRS) was analyzed using FT of cine loops, deformation of tag lines or DENSE displacement. Results DENSE and tagging both showed high sensitivity (82% and 71%) at a specificity of 80% for the detection of segments with > 50% LGE transmurality, and receiver operating characteristics (ROC) analysis showed significantly higher area under the curve-values (AUC) for DENSE (0.87) than for tagging (0.83, p < 0.001) and FT (0.66, p = 0.003). GCS correlated with global LGE when determined with DENSE (r = 0.41), tagging (r = 0.37) and FT (r = 0.15). GRS had a low but significant negative correlation with LGE; DENSE r = − 0.10, FT r = − 0.07 and tagging r = − 0.16. Torsion from DENSE and tagging had a weak correlation (− 0.20 and − 0.22 respectively) with global LGE. Conclusion Circumferential strain from DENSE detected segments with > 50% scar with a higher AUC than strain determined from tagging and FT at a segmental level. GCS and torsion computed from DENSE and tagging showed similar correlation with global scar size, while when computed from FT, the correlation was lower.

Sign in / Sign up

Export Citation Format

Share Document