An machine learning based Comparative Analysis of Various Feature Extraction Methods for Improved Prediction Accuracy of Breast Cancer CT Scan Images

This study investigates breast cancer detection through a detailed analysis of feature extraction methods and machine learning models applied to CT scan images. By utilizing the advanced features of the Wavelet Transform, Gray-Level Co-occurrence Matrix (GLCM), and Histogram of Oriented Gradients (HOG), the proposed model aimed to enhance the representation of breast tissue properties that are critical for an accurate diagnosis. The comparative analysis includes support vector machines (SVM), k-Nearest neighbors (KNN), convolutional neural networks (CNN), recurrent neural networks (RNN), and a custom model. With an astounding accuracy rate of 98.8%, CNN showed off its amazing performance and capacity to identify intricate patterns in medical images. By elucidating the degree to which each model distinguishes between cancerous and noncancerous regions, confusion matrices provide a comprehensive analysis. The findings emphasize how crucial it is to select particular machine learning algorithms and feature extraction techniques when making a breast cancer diagnosis. Even though CNN outperformed the other models, it was evident that these techniques should be applied cautiously because each model had benefits of its own. This work contributes to the field of medical image analysis and establishes the foundation for future research that will lead to new opportunities and increase diagnosis accuracy. The use of medical imaging and machine learning technologies to improve patient outcomes through improved breast cancer diagnosis is one exciting application of the research findings.