A Study on Prognosis of Brain Tumors Using Fuzzy Logic and Genetic Algorithm Based Techniques

In present study attempt has been taken to determine the degree of malignancy of brain tumors using artificial intelligence. The suspicious regions in brain as suggested by the radiologists have been segmented using fuzzy c-means clustering technique. Fourier descriptors are utilized for precise extraction of boundary features of the tumor region. As Fourier descriptors introduce a large number of feature vectors that may invite the problem of over learning and chance of misclassifications, the proposed diagnosis system efficiently search the significant boundary features by genetic algorithm and feed them to the adaptive neuro-fuzzy based classifier. In addition to shape based features, textural compositions are also incorporated to achieve high level of accuracy in diagnosis of tumors. The study involves 100 brain images and has shown 86% correct classification rate.