Diagnostic accuracy of echomics in patients with transthyretin-related cardiacamyloidosis

Le résumé fourni par la source

Abstract Background/Introduction The diagnosis of transthyretin-related cardiac amyloidosis (ATTR-CA), while often suggested by red flags observed in echocardiography, still requires a lengthy and multi-step process. Recent evidence indicates the feasibility of using radiomics on echocardiographic images (i.e., echomics) for the early characterization of myocardial texture anomalies in ATTR-CA. Aims This study aims to develop a radiomic-based model for the diagnosis of transthyretin-related cardiac amyloidosis (ATTR-CA) by characterizing myocardial tissue through echocardiography (echomics). Methods We retrospectively collected echocardiographic images in parasternal long-axis and apical 4-chamber views from ATTR-CA and control patients at four Italian centers. The largest possible region-of-interest (ROI) within the interventricular septum was delineated. Ninety-four radiomic features were extracted, including 9 shape-related features and 85 texture features, 21 of which depend on ROI dimensions and 73 purely texture-based (ROI-independent). Five logistic regression models were used to analyse data from three centres (229 ATTR-CA, 224 controls). The models estimated the diagnostic accuracy and area-under-the-curve (AUC) of different sets of radiomic features. The developed five models were finally tested on a cohort of patients from the fourth centre (32 ATTR-CA and 32 age- and sex-matched control patients) for external validation. Results The models that included shape and all texture features, whether separately or in combination, demonstrated very high cross-validated accuracies (93-95%) and AUC values (0.97-0.99). However, when using a fixed-size 0.5x0.5 cm ROI, these values decreased to 85% and 0.91, respectively. This confirms that some measurement depends on ROI size. The chosen model used 73 ROI independent features on the largest ROI and demonstrated significantly better accuracy and AUC (92% and 0.97, respectively, p<0.001). These results were confirmed in the external validation cohort (87% and 0.95, respectively). Removing the least informative features resulted in a slight model improvement 90% accuracy and 0.95 precision when 40 out of 73 features were used. Conclusions This study demonstrates the potential of echomics (i.e., radiomics applied to echocardiography data) to distinguish between ATTR-CA and control patients by identifying disease-specific textural features that are independent of ROI dimensions. Workflow Results

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.