Drop Out Rate Optimized MobileNet for Health State Classification Using Heart MRI

Cardiovascular disease is a serious ailment that affects many individuals, and it is an acronym used to describe various cardiac disorders. The categorization of heart disease based on MRI signal has gained importance in the field of heart disease diagnostics, since it can be obtained with a basic diagnostic technology at a cheap cost. Accurate cardiac disease identification with MRI is crucial since early discovery of the condition can help with therapy and even save lives. By employing cardiac MRI data, the technical advancements might be used to autonomously assess the status of the heart. This work recommends Dropout Rate Optimized MobileNet (DOMN) to classify the heart state effectively. The 5000 heart MRI images of CAD Cardiac MRI Dataset from Kaggle is used for the implementation of DOMN. The DOMN model's initial phase is an analysis of image distribution. After data preparation, the optimal CNN model is chosen by fitting the cardiac MRI images with existing CNN models. The accuracy of the MobileNet was 91.22% when compared to other CNN models after implementation. Using the MobileNet CNN, the dropout rate was now adjusted to boost accuracy. The cardiac MRI images are fitted to DOMN with dropout rates of 0.0, 0.2, 0.5, and 0.8. The experimental results show that the recommended MODN with a dropout rate of 0.2 exhibits the best Accuracy of 98.45% when compared to other dropout rates.