Myomics

Myomics is a software application for viewing, processing, and analyzing cardiovascular magnetic resonance (MR) images in DICOM format. It provides both manual and semi-automated machine learning tools to assist physicians in qualitative assessment of cardiac images and quantitative measurements of the heart and adjacent vessels, intended for use by qualified medical professionals with 3.0 T MR scanners as part of comprehensive diagnostic decision-making.

Myomics adds machine learning-based semi-automatic segmentation of myocardium across seven imaging protocols (Native T1, Post T1, T2, CINE, LGE PSIR, CINE RV, LGE Magnitude), whereas the primary predicate Myomics Q supports only manual segmentation. Both run on Microsoft Windows, are DICOM-compliant, and perform endocardium/epicardium contour segmentation; the secondary predicate cvi42 Auto additionally processes CT images and runs on Mac OS.

ISO 13485:2016, IEC 62304:2015, ISO 14971:2019 for software development and risk management. AI modules were validated using DICE Score metric (threshold >0.7) across 728 anonymized patient images from multiple MR manufacturers, with training/validation/test split of 80/10/10.

Myomics achieves substantial equivalence through functional and performance parity with predicates: it preserves all core worklist functions (import, display, export, manual contour drawing) and quantitative cardiac measurements of the primary predicate Myomics Q, while adding AI-powered semi-automatic segmentation capability that mirrors the secondary predicate cvi42 Auto's machine learning approach. Bench testing confirmed equivalent performance in segmentation tasks, and AI validation demonstrated consistent accuracy (DICE >0.7) across diverse MR vendors and imaging protocols, demonstrating that the addition of automation does not change the fundamental safety or effectiveness profile relative to the manually-operated predicate.