Image Filtering Using Anisotropic Diffusion for Brain Tumor Detection

Abstract

Brain tumor analysis is a critical aspect of medical applications, offering valuable structural and functional insights crucial for disease diagnosis. Early detection of tumors significantly enhances treatment outcomes and patient survival rates. However, the manual segmentation of numerous magnetic resonance images poses challenges due to the increased risk of human error. Therefore, there is a pressing need for computer-aided detection systems to ensure higher accuracy and faster tumor identification. In our work, we propose computer-aided techniques utilizing anisotropic diffusion filtering, Otsu threshold segmentation, and morphological procedures for noise reduction, segmentation, and tumor area detection in MR images. Our approach aims to streamline the process of tumor identification by automating key steps through advanced image processing methods. Notably, simulation results highlight the superiority of anisotropic diffusion and Otsu thresholding over other filtering and segmentation combinations, underscoring their effectiveness in enhancing tumor detection accuracy.