Brain Tumor Classification using the Probabilistic-Deformable Fuzzy system and RideNN

Abstract

In the medical field, brain tumor classification is a very significant treatment process. The manual segmentation and classification of brain tumors is a tedious and time-consuming process. Hence, automatic classification and segmentation methods are very important. This paper presents brain tumor classification using a Probabilistic-deformable fuzzy system (PDFS) with RideNN. The overall procedure of the proposed approach involves four steps, such as pre- processing, segmentation, feature extraction, and classification. Initially, the input image is subjected to pre-processing. Then, the brain tumor segmentation is done using a new model called the Probabilistic-Deformable Fuzzy system. It is developed by combining a deformable model and a fuzzy system. Then the result is combined using probability-based fusion. After the segmentation, feature extraction is carried out by extracting texture-based features such as Local Gradient Pattern (LGP) and tetrolet transform. Finally, the classification is performed based on the features extracted using RideNN which employs Rider optimization Algorithm (ROA) for training the Neural network (NN). The performance of the brain tumor classification is based on PDFS-RideNN and evaluated based on sensitivity, specificity, and accuracy. The proposed PDFS-RideNN method achieves a maximal sensitivity of 93.18%, a maximal specificity of 98.67%, and a maximal accuracy of 98.25%.