Cardiomyopathy Classification with Echocardiogram Videos Using Deep Learning Techniques

Cardiomyopathy is a group of diseases that weaken the heart muscle making it harder for the heart to pump blood. Cardiomyopathy reduces blood output and may lead to heart failure. A variety of qualitative methods and measures have emerged to examine the cardiac function. However, the assessment of cardiomyopathy is bought to create a high progressive method with help of echocardiogram videos for the measurement of ejection fraction in the dilated left ventricle valve. Training on echo videos, proposed model accurately subdivides into three classes as normal, low ejection fraction and arrythmia. In the further analysis, using suitable deep learning techniques the CNN model provides high accuracy result, the evaluation metrics for multiclass classification include precision, recall, f1- score and confusion matrix with probability of 0.99. As a resource to promote great innovation we considered publicly available Econet-dynamic dataset of 195 annotated echocardiogram videos to predict the attack of cardiomyopathy. On model training and testing the CNN algorithm achieves 99.56% as overall accuracy score.