Machine Learning Driven Ultrasound Image Classification with Filter Optimization

Breast Cancer (BC) ranks being the most prevalent form of cancer, streessing the significance of accurate diagnosis a priority in medical research. Determine if a tumor is benign or malignant is crucial for effective treatment and reducing healthcare costs. Ultrasound imaging is a widely adopted method for breast tumor screening, but the presence of scattered noise in these images often hinders diagnostic accuracy, complicating radiologists efforts to detect abnormalities. This study emphasizes enhancing image quality as a preprocessing step before training machine learning models. To achieve this, the entropy of three filters, Gaussian, and Sobel, are evaluated for their ability to minimize noise and enhance edges. The filtered images are classified using Support Vector Machine (SVM), Logistic Regression, and K-Nearest Neighbors (KNN). The performance of these classifiers is assessed through metrics like accuracy, precision, and recall to determine the most effective approach for cell classification.