Breast Pathology Changes Extraction and Measurement Based on Machine Learning and DWT

Abstract

In recent years, medical image analysis has witnessed significant advancements in aiding accurate diagnosis and treatment planning. Breast tumor segmentation is a critical task in medical imaging, as it facilitates the identification and characterization of tumors for effective clinical decisions. This paper proposes a novel approach for breast tumor segmentation and analysis by integrating Fuzzy C-Means Clustering (FCM) with Discrete Wavelet Transform (DWT), called FCMDWT. This method is effective in breast diagnosis analysis, tumor size measurements, and diagnosing reports and does not require prior training in segmentation. Initially, the DWT is applied to the mammography image, decomposing it into different frequency subbands. FCM is employed on the DWT coefficients to ensure robust clustering by accommodating uncertainty and overlapping regions in the image. The experimental evaluation conducted on a comprehensive dataset and comparative analyses demonstrates the superiority of the FCMDWT approach. Furthermore, the proposed method extends beyond segmentation, incorporating tumor analysis by extracting relevant features such as size, shape, and texture. The results indicate the potential of the FCMDWT approach in not only accurate segmentation but also in providing valuable insights for clinical decision-making.