Machine Learning Based Detection and Classification of Heart Abnormalities

Heena, A.; Biradar, N.; Maroof, N. M.

Various types of heart abnormalities require various parameters of consideration for effective detection, analysis, processing and classification of heart abnormalities. Like any other disease, prevention is the key to avoid heart diseases. Heart abnormalities are very common not only in India but all over the world, but with better understanding of different symptoms of heart problems, proper diagnosis and prognosis is possible, so that early and optimal treatment is initiated. Particularly in these Pandemic situations of COVID 19 which led to more stress and as a consequence many heart related issues. Deep learning is also vastly employed technique in recent researches for classification of heart images and biomedical images in general. Neural network and support vector machine (SVM) algorithms are also frequently used in classification because of higher accuracy in results. The use of machine learning resulted in improved accuracy and reduced variability in comparison to manual quantification of echocardiographic parameters. A hybrid algorithm (Image enhancement and denoising both carried out in preprocessing task) is developed in this article which can efficiently and effectively classify the heart abnormalities by detecting whether the image is normal or abnormal, if abnormal whether mild stage or critical stage. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.