Classification performed on Magnetic Resonance (MR) brain images by automated means and with accuracy is highly essential for medical evaluation and diagnosis. MR brain images‟ classification is gaining rising significance in the medical field as it is critical for planning the treatment and diagnosing the abnormalities, measurement of the tissue volume for investigating the tumor development and analysing the anatomical structure and the following procedure of the patient. The key to this framework is the adoption of fruit fly optimization algorithm (FFOA) improved by Levy flight (LF) mechanism (LFFOA) for optimizing the two core parameters of support vector machine (SVM) and LFFOA-based SVM (LFFOA-SVM) is built for the diagnosis of the breast cancer. In fact, the basic step behind the classifiers involved with magnetic resonance imaging (MRI) brain scans lies in their capability of extracting useful features. Consequently, several works have introduced various techniques for features extraction in order to differentiate the irregular developments observed in the brain MR images. Quite lately, the usage of deep learning algorithms for medical imaging presented interesting performance improvements in the classification and diagnosis of sophisticated pathologies, like tumors in the brain. Moreover, a deep learning feature extraction algorithm is introduced for the extraction of the useful features acquired from MR brain images. Concurrently, manmade features‟ extraction are carried out with the help of the Modified Gray Level Co-Occurrence Matrix (MGLCM) technique. In the next level, the useful features extracted are integrated with manmade features in order to boost the classification procedure of MR brain images and LFFOA-SVM is employed in the form of the classifier. The experimental results show that a fusion of MGLCM and deep learning features and LFFOA-SVM classifier acts as a good-performing and generic model for MRI brain image classification.
Volume 11 | 11-Special Issue