Differentiating cardiac amyloidosis from hypertrophic cardiomyopathy is clinically challenging, as both conditions share overlapping native T1 values on cardiac magnetic resonance imaging. A study presented at the European Society of Cardiology (ESC) Congress 2025, evaluated whether radiomics-based machine learning could improve diagnostic accuracy. Early and precise identification of cardiac amyloidosis is crucial, since delayed diagnosis is linked to worse outcomes.

The study included 35 patients—11 with cardiac amyloidosis and 24 with hypertrophic cardiomyopathyThe analysis extracted 137 radiomics features from native T1 mapping and refined them using least absolute shrinkage and selection operator regression. Six machine learning models were tested: support vector machines, logistic regression, Naïve Bayes, decision trees, random forests, and k-nearest neighbors.

The support vector machine achieved the highest accuracy at 92 percent, with strong precision, recall, and area under the curve values. Logistic regression and Naïve Bayes also performed well, while decision trees and random forests showed lower accuracy. Further validation in larger cohorts is needed before widespread clinical implementation.