Classical Machine Learning-Based Approaches for Medical Image Analysis and Computer-Aided Diagnosis

From MRI and CT to ultrasound and mammography, diagnostic imaging produces a steady flow of clinically meaningful data - and it's precisely this volume of information that has pushed automated image analysis to the forefront of AI research in healthcare. This paper focuses on classical machine learning methods for computer-aided diagnosis and clearly separates them from deep learning approaches. The theoretical sections discuss the general workflow of ML-based medical image analysis, including preprocessing, feature extraction, feature selection, classification, and evaluation. The experimental section was redesigned as an independent simulation and is no longer based on the experimental results reported in the reviewed literature. A synthetic dataset of 1,000 patients over a three-month period was generated to evaluate several classical ML classifiers under controlled conditions.