Automated Segmental Analysis of Fully Quantitative Myocardial Blood Flow Maps by First-Pass Perfusion Cardiovascular Magnetic Resonance

Abstract

First pass gadolinium-enhanced cardiovascular magnetic resonance (CMR) perfusion imaging allows fully quantitative pixel-wise myocardial blood flow (MBF) assessment, with proven diagnostic value for coronary artery disease. Segmental analysis requires manual segmentation of the myocardium. This work presents a fully automatic method of segmenting the left ventricular myocardium from MBF pixel maps, validated on a retrospective dataset of 247 clinical CMR perfusion studies, each including rest and stress images of three slice locations, performed on a 1.5T scanner. Pixel-wise MBF maps were segmented using an automated pipeline including region growing, edge detection, principal component analysis, and active contours to segment the myocardium, detect key landmarks, and divide the myocardium into sectors appropriate for analysis. Automated segmentation results were compared against a manually defined reference standard using three quantitative metrics: Dice coefficient, Cohen Kappa and myocardial border distance. Sector-wise average MBF and myocardial perfusion reserve (MPR) were compared using Pearson’s correlation coefficient and Bland-Altman Plots. The proposed method segmented stress and rest MBF maps of 243 studies automatically. Automated and manual myocardial segmentation had an average (± standard deviation) Dice coefficient of 0.86 ± 0.06, Cohen Kappa of 0.86 ± 0.06, and Euclidian distances of 1.47 ± 0.73 mm and 1.02 ± 0.51 mm for the epicardial and endocardial border, respectively. Automated and manual sector-wise MBF and MPR values correlated with Pearson’s coefficient of 0.97 and 0.92, respectively, while Bland-Altman analysis showed bias of 0.01 and 0.07 ml/g/min. The validated method has been integrated with our fully automated MBF pixel mapping pipeline to aid quantitative assessment of myocardial perfusion CMR.

Predicting post-contrast information from contrast agent free cardiac MRI using machine learning: Challenges and methods

“…Given the complex nature of this problem and the reported accuracy of 95.03% mean Dice score of the trained model for the identification and segmentation of the scar from cine SAX images, it is most likely that the model – trained on the dataset of 165 cine CMR patients (140 diagnosed with MI and 25 control cases) – suffers from overfitting issues. Moreover, even simpler DL-based image segmentation problems involving CMR images hardly achieved this level of accuracy to date [for example, in Bai et al ( 52 ), endocardium and epicardium segmentation models achieved 0.88 (0.03) and 0.94 (0.04) mean (±SD) Dice scores, respectively; in Jacobs et al ( 56 ), myocardial segmentation model achieved 0.86 (±0.06) Dice score on gadolinium-enhanced CMR images; and in Zhuang et al ( 57 ), myocardial segmentation of the mid-ventricular slice achieved 0.86 (±0.07) inter-observer Dice score on LGE CMR images]. It should be noted that in Zhang et al ( 58 ), researchers have used the framework given in Xu et al ( 29 ) (i.e., with the following main components: LV localization; the motion feature extraction layers which use LSTM and optical flow techniques; and the fully connected layers) for a training dataset that consists of only chronic MI ( n = 169) and control ( n = 69) patients.…”

“…Given the complex nature of this problem and the reported accuracy of 95.03% mean Dice score of the trained model for the identification and segmentation of the scar from cine SAX images, it is most likely that the model -trained on the dataset of 165 cine CMR patients (140 diagnosed with MI and 25 control cases) -suffers from overfitting issues. Moreover, even simpler DL-based image segmentation problems involving CMR images hardly achieved this level of accuracy to date [for example, in Bai et al (52), endocardium and epicardium segmentation models achieved 0.88 (0.03) and 0.94 (0.04) mean (±SD) Dice scores, respectively; in Jacobs et al (56) The Dice score of 86.1% (±5.7) was reported in the study but this was the result of a small test set [chronic MI (n = 43) and control (n = 18) patients] from a single vendor and single center (i.e., the same vendor and center as the training set). The approach and dataset presented in this article are not limited to chronic MI cases only.…”

Clinical Application of Dynamic Contrast Enhanced Perfusion Imaging by Cardiovascular Magnetic Resonance

“…Perfusion images must first be reconstructed from the raw data. The dynamic image series require correction for respiratory motion in addition to correction for coil sensitivity bias (80,81). Segmentation of the left ventricle and myocardium is required to enable extraction of the AIF and myocardial tissue curves.…”

“…If a dual sequence approach has been employed, SI data requires conversion to CA concentration. Only at this stage are quantification models applied to the perfusion data to calculate MBF (81). Until recently, these multiple processing steps required time-consuming and laborious manual input, which restricted the application of quantitative perfusion CMR from mainstream clinical practise.…”

Global Context and Enhanced Feature Guided Residual Refinement Network for 3D Cardiovascular Image Segmentation

“…As an intuitive manner, cardiovascular images can give detailed visual morphology presentation for the blood pool and the corresponding surrounding myocardium. Segmenting the heart in cardiovascular images plays an important and crucial role in cardiovascular disease diagnosing and treatment planning [2]- [4]. However, manually accomplishing this task is laborious, tedious and much time is needed, especially when medical resources are scarce.…”