Hybrid Approach for Segmentation of Tissues in MR Brain Images using Adaptive Neuro Fuzzy Inference System

Abstract

MR brain images are widely used in medical applications for research, diagnosis, treatment, surgical planning and image guided surgeries. Segmentation and Classification of brain tissue in MRI is a crucial process in several medical researchers, clinical applications and involving measurement of tissue volume. Manual segmentation and classification of MR Brain image is a tedious task and time consuming process which is impractical for large amount of data. Because the brain has a complicated structure and are often corrupted with Intensity Inhomogeneity artifact that cause unwanted intensity variation due to non- uniformity in RF coils and Rician noise, the dominant noise in MRI due to thermal vibrations of electrons, ions and movement of objects during acquisition which may affect the performance of image processing techniques used for brain image analysis. Due to this difficulties and obstacles like noise and INU artifact, sometimes one type of normal tissue in MRI may be misclassified as other type of normal tissue and it leads to error in quantifying volume changes of normal tissues for clinical diagnosis to identify the diseases due to delineation of tissues by the medical practitioner. Fully automatic and Soft Computing Techniques eliminate this problem. The proposed method consists of two preprocessing steps (1) wrapping based curvelet transform to remove noise (2) Skull stripping using Mathematical Morphology and then normal tissues such as White Matter, Gray Matter and Cerebrospinal Fluid are segmented from noise removal skull stripped image using Modified Spatial Possibilistic Fuzzy C Means. The textural features are extracted from segmented tissues using Local Binary Pattern. The extracted features are trained by Adaptive Neuro-Fuzzy Inference System to classify unknown MR brain image normal tissues in testing phase after extracting feature from testing brain images. The Performance of segmentation results are compared with the expert’s results and analyzed. The classifier differentiates the normal tissues with relatively high accuracy.