A Novel Algorithm to Detect Brain Tumor using Staged-Type-II Fuzzy Classifier

Automatic detection of brain tumor is a crucial step in the domain of medical image processing. Classification of the tumor is a vital part in brain tumor diagnosis to aid accurate treatment. However, manual detection with the help of human interpretation is time taking and also subject to inaccurate diagnosis. Based on these facts, an automated brain tumor classification algorithm is proposed in this work. The present work is divided into the following stages, viz. preprocessing, segmentation, feature extraction, feature selection, ranking of the selected features and finally classification of the segmented tumor. Gray-Level-Co-Occurrence Matrix (GLCM), Law’s Texture and Mass Effect features are extracted from the brain tumor and feature selection is carried out for each individual type followed by the ranking of the individual feature types. The final step comprises of the classification algorithm where a three stage classifier using Interval Type-II Fuzzy Logic System is designed in order to classify the segmented tumor into benign or malignant class. Finally, the work is validated with the help of BRATS 12 dataset and the superiority of the model is showcased in comparison with Type-I Fuzzy Inference System.