Deep Learning-Based Self-Supervised Transfer Learning for Medical Image Classification

Abstract

Self-supervised switch getting to know is a practical approach for the scientific image type. It includes using unlabeled statistics (from non-clinical images) to teach models to categorize medical photos with low attempts and excessive accuracy. This approach has been utilized in numerous duties, including clinical photo segmentation, computer-aided prognosis, and disease category. To summarize the research studies carried out in this discipline, a complete survey of self-supervised switch learning for scientific image type was performed in 2020. The survey evaluations numerous transfer mastering techniques often used in healthcare, which include area adaptation, multitasking studying, and zero/one-shot mastering. It additionally gives an in-intensity analysis of the modern-day challenges and capability solutions.