Quantitative Imaging Analysis in Hypertrophic Cardiomyopathy: Phenotypic Differentiation to Prognostic Stratification

Figures

• 

  Figure 1. PRISMA ScR flowchart. Records were identified through searches of Embase, MEDLINE, and Scopus. Following removal of duplicate records, titles and abstracts were screened, with eligible full-text articles assessed against the predefined inclusion and exclusion criteria. Reasons for full-text exclusion included incorrect study population, absence of quantitative radiomics methodology, inappropriate publication type, and insufficient data. A total of 23 studies met the eligibility criteria and were included in the final systematic review.

• 

  Figure 2. Radiomics pipeline in hypertrophic cardiomyopathy. Cardiac images (CMR, TTE or CT) are segmented to delineate the myocardial region of interest (Step 1), from which high-throughput quantitative features are extracted, either hand-crafted radiomic features or deep learning representations (Step 2). Feature selection or dimensionality reduction mitigates overfitting (Step 3) before classification with conventional or ensemble machine learning models (Step 4). Outputs may inform clinical decisions, such as contrast-sparing imaging or ICD implantation, and interpretability methods such as SHAP identify the features that drive each prediction (Step 5).

• 

  Figure 3. Clinical integration pathway for radiomics in left ventricular hypertrophy (LVH) assessment. This figure illustrates a proposed workflow for incorporating radiomics into the diagnostic evaluation of patients with left ventricular hypertrophy or unexplained myocardial wall thickening. Following initial clinical assessment with electrocardiography, transthoracic echocardiography, and routine blood tests, patients undergo comprehensive cardiac magnetic resonance (CMR) imaging, including cine sequences, T1/T2 mapping, and late gadolinium enhancement (LGE). A radiomics pipeline can then be applied to generate a predicted myocardial phenotype, such as hypertrophic cardiomyopathy (HCM), hypertensive heart disease (HHD), cardiac amyloidosis (CA), or other aetiologies. These outputs are integrated with conventional clinical and imaging findings to support refined diagnosis and guide downstream, disease-specific management pathways.