REVIEW OF VARIOUS AI ASSISTED MULTIPLE MYELOMA DISEASE DETECTION BASED ON CLINICAL FEATURES AND IMAGING MODALITIES

Abstract

Multiple myeloma (MM) is a malignant illness that results in the uncontrolled growth of plasma cells in the bone marrow, which produce abnormal levels of mono- clonal immunoglobulin. However it has a lack of standardized protocol dependency on radiologist which requires the need for automated systems to improve the diag- nostic precision and reduce delays. Advances in clinical imaging approaches using AI-driven diagnostic frameworks are examined for assessing the research gap in or- der to improve the identification of multiple myeloma using artificial intelligence (AI). Recent advancements in imaging techniques and clinical data, including ma- chine learning algorithms for predictive modelling and a sophisticated deep learning architecture for automated multiple myeloma analysis, are examined in this study. Following hybrid/ensemble techniques for robust tumour classification and efficient feature extraction, transfer learning techniques are used for successful adaptation in data-constrained environments. The experimental result indicate that among evaluated techniques deep learning methods such as Disruption-based Salp-Swarm and Cat Swarm based optimized Convolutional Neural Networks (DSSCSCNN) achieved 98% precision and using clinical attribute autoencoder (AE) delivered accuracy of 97% outperforming traditional methods. This review highlights by integrating AI techniques significantly enhances the Multiple myeloma detection accuracy with minimal data resources making it highly practical for maintaining high performance.