Texture-based Classification of Ultrasound Breast Cancer Images using Machine Learning

A study on texture-based classification of ultrasound breast cancer images using machine learning demonstrates a promising approach to automate breast cancer diagnosis. This research introduces a computer-aided diagnostic (CAD) system for categorizing breast images into benign and malignant tumors. Machine learning techniques are utilized to extract texture features and classify images based on these features. Evaluation metrics such as accuracy, precision, sensitivity, specificity, and F1-Score are employed to assess performance, with confusion matrices created for each metric. Support Vector Machine (SVM) classifiers with various kernels are employed due to the complexities of medical images, achieving an average accuracy of 95%. The findings suggest that the proposed CAD system could effectively aid radiologists in distinguishing between benign and malignant tumors.