Detecting and Locating of Brain Abnormality in MR Images Using Texture Feature Analysis and Improved Probabilistic Relaxation Methods

Medical imaging has become a major tool in clinical trials since it enables rapid diagnosis with visualization and quantitative assessment. In the study, a detecting method of brain abnormality is proposed through magnetic resonance imaging. The proposed method is composed of four procedures. First the preprocessing is employed to remove noises and enhance the homogeneity of soft tissues. After preprocessing, we adopt the spatial gray level dependence method to compute four texture features of each image. Then, the improved probability relaxation method is applied to discriminate the brain abnormality with extracted texture information. The isolated noises are removed by using neighborhood processing. Final the performance of the improved method has been evaluated and compared to the original method. This proposed method performs better results than the other one, which can be used in further processing stages. We have developed a computer-aided detection system to distinguish the tumor and find the location and coarse contour from brain MRIs. The system can assist doctors to diagnose whether the brain has abnormal and train inexperienced doctors. The proposed algorithm can play a useful role for storage, filtering and indexing of mass MRI data, and furthermore it provides an initial step to find accurate tumor boundaries. Key-Words: Computer-Aided Detection System, Texture Feature Analysis, Spatial Gray Level Dependence, Probability Relaxation Method, Magnetic Resonance Image, Brain Tumor