Enhanced Multimode DBN for Optimal Classification of Heterogeneous Cancer Images for HealthCare System

Abstract

Medical data has grown tremendously in recent years. Health and medical science are advancing through big data and deep learning techniques to predict outcomes. Existing interventions focus on Brain, Breast and Bone cancer separately. In this study, we proposed an e-MDBN (enhanced Multimode Deep Belief Network) model, which is primarily based on an optimized parallel CRBM (Clipped Restricted Boltzmann Machine) Algorithm that trains the layer, then fine-tunes the e-MDBN model and classifies the image into Brain, Breast and Bone cancer. This deep learning-based image classification workflow delivers efficient results. Pyspark's distribution platform, which has an improved e-MDBN network structure, has the fastest and highest accuracy rate. Accuracy, precision, mean squared error, and recall of the recommended methodological configurations are superior to the state-of-the-art alternatives.