Application of Neuro-Fuzzy Model for MR Brain Tumor Image Classification

Brain tumor image classification and segmentation are an important but inherently difficult problem in magnetic resonance (MR) medical images. Artificial neural networks employed for image classification problems do not guarantee high accuracy besides being computationally heavy. The necessity for a large training set to achieve high accuracy is another drawback of ANN. On the other hand, fuzzy logic technique which promises better accuracy depends heavily on expert knowledge, which may not always available. Even though it requires less convergence time, it rely on trial and error method in selecting either the fuzzy membership functions or the fuzzy rules. These problems are overcome by the hybrid model namely, neurofuzzy model. This system removes the stringent requirements since it enjoys the benefits of both ANN and the fuzzy logic systems. In this paper, the application of Adaptive neuro-fuzzy inference systems (ANFIS) for MR brain tumor classification has been demonstrated. Abnormal brain tumor images from four classes namely metastase, meningioma, glioma and astrocytoma are used in this work. A comprehensive feature set and fuzzy rules are selected to classify an abnormal image to the corresponding tumor type. Experimental results illustrate promising results in terms of classification accuracy and convergence rate. A comparative analysis is performed with the representatives of ANN and fuzzy systems to show the superior nature of ANFIS systems.