Medical Image Segmentation Using Machine Learning

Abstract

Accurate medical image segmentation is critical for diagnosis and treatment planning, yet manual delineation is time consuming and prone to error. Automatic methods like thresholding and edge detection often fail in the presence of noise, artifacts, and complex textures. This study addresses the need for robust segmentation of CT scans and wound images by combining traditional machine learning with lightweight deep learning, enabling precise region-of-interest extraction for clinical analysis.