ABSTRACT
Objectives: Currently, administering contrast agents is necessary for accurately visualizing and quantifying presence, location, and extent of myocardial infarction (MI) with cardiac magnetic resonance (CMR). In this study, our objective is to investigate and analyze pre- and post-contrast CMR images with the goal of predicting post-contrast information using pre-contrast information only. We propose methods and identify challenges. Methods: The study population consists of 272 retrospectively selected CMR studies with diagnoses of MI (n = 108) and healthy controls (n = 164). We describe a pipeline for pre-processing this dataset for analysis. After data feature engineering, 722 cine short-axis (SAX) images and segmentation mask pairs were used for experimentation. This constitutes 506, 108, and 108 pairs for the training, validation, and testing sets, respectively. We use deep learning (DL) segmentation (UNet) and classification (ResNet50) models to discover the extent and location of the scar and classify between the ischemic cases and healthy cases (i.e., cases with no regional myocardial scar) from the pre-contrast cine SAX image frames, respectively. We then capture complex data patterns that represent subtle signal and functional changes in the cine SAX images due to MI using optical flow, rate of change of myocardial area, and radiomics data. We apply this dataset to explore two supervised learning methods, namely, the support vector machines (SVM) and the decision tree (DT) methods, to develop predictive models for classifying pre-contrast cine SAX images as being a case of MI or healthy. Results: Overall, for the UNet segmentation model, the performance based on the mean Dice score for the test set (n = 108) is 0.75 (±0.20) for the endocardium, 0.51 (±0.21) for the epicardium and 0.20 (±0.17) for the scar. For the classification task, the accuracy, F1 and precision scores of 0.68, 0.69, and 0.64, respectively, were achieved with the SVM model, and of 0.62, 0.63, and 0.72, respectively, with the DT model. Conclusion: We have presented some promising approaches involving DL, SVM, and DT methods in an attempt to accurately predict contrast information from non-contrast images. While our initial results are modest for this challenging task, this area of research still poses several open problems.
ABSTRACT
AIMS: To determine population-related and technical sources of variation in cardiac magnetic resonance (CMR) reference ranges for left ventricular (LV) quantification through a formal systematic review and meta-analysis. METHODS AND RESULTS: This study is registered with the International Prospective Register of Systematic Reviews (CRD42019147161). Relevant studies were identified through electronic searches and assessed by two independent reviewers based on predefined criteria. Fifteen studies comprising 2132 women and 1890 men aged 20-91 years are included in the analysis. Pooled LV reference ranges calculated using random effects meta-analysis with inverse variance weighting revealed significant differences by age, sex, and ethnicity. Men had larger LV volumes and higher LV mass than women [LV end-diastolic volume (mean difference = 6.1 mL/m2, P-value = 0.014), LV end-systolic volume (MD = 4 mL/m2, P-value = 0.033), LV mass (mean difference = 12 g/m2, P-value = 7.8 × 10-9)]. Younger individuals had larger LV end-diastolic volumes than older ages (20-40 years vs. ≥65 years: women MD = 14.0 mL/m2, men MD = 14.7 mL/m2). East Asians (Chinese, Korean, Singaporean-Chinese, n = 514) had lower LV mass than Caucasians (women: MD = 6.4 g/m2, P-value = 0.016; men: MD = 9.8 g/m2, P-value = 6.7 × 10-5). Between-study heterogeneity was high for all LV parameters despite stratification by population-related factors. Sensitivity analyses identified differences in contouring methodology, magnet strength, and post-processing software as potential sources of heterogeneity. CONCLUSION: There is significant variation between CMR normal reference ranges due to multiple population-related and technical factors. Whilst there is need for population-stratified reference ranges, limited sample sizes and technical heterogeneity precludes derivation of meaningful unified ranges from existing reports. Wider representation of different populations and standardization of image analysis is urgently needed to establish such reference distributions.
