Enhancing Breast Cancer Screening: Mammogram Image Analysis with Multiclass Support Vector Machine

Mammography is an essential screening test for the early detection of breast cancer, significantly increasing the patient's chances of complete recovery. In this paper, a Multi-Class Support Vector Machines (MSVM) is presented to address the overfitting problem and enhance the technique's performance in detecting locations and areas of breast cancer. This approach proves to be particularly effective in high-dimensional spaces, offering advantages in the context of breast cancer detection, where datasets may contain a large number of features. The MIAS dataset in the classification stage utilizing 60 breast cancer images including 100 malignant and 100 benign breast cancer images. All these images have been labeled with 63 images labeled as normal and 37 images as benign or micro, with an original pixel size of
$\boldsymbol{1024}\times \boldsymbol{1024}$
. After segmentation using Particle Swarm Optimizer (PSO) to segment nuclei and non-nuclei cells and involved parallel computing and global optimization to achieve superior results for the Super pixel. Feature extraction was performed using the Grey Level Co-Occurrence Matrix (GLCM). Finally, the (MSVM) classifier was utilized to grade the histology breast images. The experimental results demonstrate that the MSVM achieved an accuracy rate of 98.90% surpassing other existing models such as Convolutional Neural Network, Deep Neural Network, and Support Vector Machine.