Brain Tumor Analysis of Rician Noise Affected MRI Images

Magnetic Resonance Imaging (MRI) established itself as a key imaging modality in diagnosis and treatment of brain tumors. Automatic segmentation of tumors becomes a tedious task due to complex anatomical brain structure. In addition, presence of noise degrades the quality of MRI scans. MRI images are usually corrupted by Rician noise which would mislead the image analysis algorithms and results in improper diagnosis of the diseases. Also, poor tumor boundary becomes a major hurdle for the subsequent stages of tumor analysis such as: feature extraction, classification and quantification. Classification accuracy mainly depends on quality of the denoised images and sharpness of the tumor boundary. This paper investigates the performance evaluation of different image matting techniques to extract tumor from Rician noise affected MRI brain images. General Terms Medical Image Processing