Optimal Feature Selection based on Hybridization of MSFLA and Gabor filters for Enhanced MR Brain Image Recognition using SVM
Anis Ladgham, Anis Sakly, Abdellatif Mtibaa · International journal of tomography and simulation · 2014
This paper presents a novel technique of recognition of MRI brain tumor based on the hybridization of the Support Vector Machines (SVM) and the Modified Shuffled Frog Leaping Algorithm (MSFLA). This technique is termed MSFLA-SVM. Firstly, we use the Gabor filter for textural feature extraction from MRI. Secondly, the feature selection stage is performed using the MSFLA. This stage is used to obtain significantly reduced feature subset and improved recognition rate. Finally, the optimal features are given as input to the SVM classifier to detect the brain MRI as normal or infected. The proposed approach has been compared with other recent methods. Experimental results show that the proposed MSFLA-SVM is able to achieve better recognition quality and execution time than the other methods.