A speckle-tracking strain-based artificial neural network model to differentiate cardiomyopathy type

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Invasive angiography is standard for differentiating ischaemic and non-ischaemic cardiomyopathy in heart failure.
• Accurate differentiation is crucial for guiding appropriate treatment strategies.
• Non-invasive methods to predict cardiomyopathy aetiology are highly desirable.

Purpose of the Study:

• To evaluate the predictive capability of echocardiographic strain features for differentiating ischaemic from non-ischaemic cardiomyopathy.
• To compare the performance of artificial neural network (ANN) models against logistic regression.
• To assess the combined utility of strain parameters with other clinical and electrocardiographic features.

Main Methods:

• Retrospective analysis of 204 heart failure patients with ejection fraction <50% and diagnostic angiograms.
• Development of logistic regression and ANN models using echocardiographic strain, regional wall motion abnormalities, ECG, and demographic features.
• Models were trained on a derivation cohort and validated on a separate validation cohort.

Main Results:

• A strain-based ANN model achieved an F1 score of 0.82.
• A full feature ANN model achieved an F1 score of 0.79.
• The full feature logistic regression model yielded an F1 score of 0.63.

Conclusions:

• ANN models demonstrated superior accuracy in predicting cardiomyopathy type compared to logistic regression.
• The strain-based ANN model shows significant potential for aiding heart failure aetiology diagnosis.
• Further validation in diverse cohorts and integration into clinical tools like online calculators or echocardiogram software are recommended.