Discrimination between hypertrophic cardiomyopathy and hypertensive heart disease using deep learning-based magnetic resonance imaging
Year: 2026
Type: article
Abstract: Background Discrimination between HCM and HHD can be challenging due to overlapping imaging presentation and clinical features. This study aimed to develop a deep learning pipeline for discrimination between hypertrophic cardiomyopathy (HCM) and hypertensive heart disease (HHD) using cardiovascular magnetic resonance (CMR) images. Materials and methods EARLY-MYO-HHD registry is a prospective cardiac imaging study that enrolled patients with HCM and HHD. A framework comprised of convolutional neural networks (CNN) and support vector machine (SVM) classifiers was used to estimate the probability of HCM or HHD based on CMR cine sequence and CMR parameters (including LVEF and scar size). Results A total of 368 patients were included (303 patients for model development and 65 patients for independent testing). With 10-fold cross-validation, our proposed CNN model correctly classified 88.1 ± 6.9% of patients with HCM and HHD. By combining CNN model prediction and scar size, the SVM model achieved an overall classification accuracy of 91.1 ± 5.5%, and the predictive performance showed great performance among all HCM subgroups (with or without wall thickness < 15 mm or hypertension, AUCs >0.93). Also, the CNN prediction and scar size-based model achieved a classification accuracy of 96.9% (AUC 0.99, CI 0.96–1.00) in the testing dataset. Conclusion A patient-level deep learning method was developed for HCM and HHD classification using CMR cine images. Our proposed method provided a useful approach for classification of HCM and HHD, especially in clinically ambiguous cases. Adding semi-quantitative scar size information further improved the classification accuracy of HCM and HHD patients. Clinical Trial Registration identifier NCT03271385.
Source: Frontiers in Medicine
Language: English
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7225829923