A Deep Learning Approach For Heart Disease Classification Using Convolutional Neural Network

• Main Author: Arulmolly, Annathurai
• Format: Final Year Project Report
• Language: English; English
• Published: Universiti Malaysia Sarawak (UNIMAS) 2019
• Subjects: BF Psychology; R Medicine (General)
• Online Access: http://ir.unimas.my/id/eprint/27551/1/A%20deep%20learning%20approach%20for%20heart%20disease%20classification%20using...%2824%20pgs%29.pdf; http://ir.unimas.my/id/eprint/27551/7/Arulmolly%20Annathurai%20ft.pdf; http://ir.unimas.my/id/eprint/27551/

Cardiovascular disease is the leading cause of death, and specialists estimate that roughly half of all heart attacks and strokes happen in individuals who have not been flagged as 'at risk. Hence, there's an urgent need to progress the exactness of heart infection diagnosis. To this end, we explore the potential of utilizing information examination, and in specific the plan and utilize of convolutional neural networks (CNN) for classify heart disease based on ultrasound heart images. Our fundamental commitment is the plan, assessment, and optimization of CNN models of expanding profundity for heart disease classification. Moreover, a system with high precision and accuracy is required to analyse heart disease classification. This study utilized ultrasound scanned heart images which are collected from local hospital. A total of 50 images which includes 20 normal and 30 abnormal heart images. The model has been run successfully without any errors and produce a high accuracy of 96% after running for 46 epochs. Furthermore, a comparative study has been done between SVM and CNN using the similar dataset to analyse performance based on the accuracy.