Mix-model for optimisation of textural features applied to multiple sclerosis lesion-tumour segmentation

Abstract

Segmentation of biomedical images plays an important role in many applications especially in medical imaging, forming an important step in enabling qualification in the field of medical research as well as clinical practices. Magnetic resonance imaging is normally used to distinguish and enumerate multiple sclerosis lesions in the brain. Recently multiple sclerosis lesion of segmentation is the challenging issue due to special variation, low size and unclear boundaries. Since usual technique for brain MRI tumour detection and classification is manual investigation but it is varied from person to person and also very time consuming. Many new methods have been proposed to segment lesions automatically. This paper proposed segmentation of MRI brain tumour using cellular automata and classification of tumour by pointing kernel classifier (PKC). The utilisation of modified Cuckoo search with the priority values and the PKC in proposed mix model for optimisation of textural features (M-MOTF) provides the significant improvement in classification performance with low dimensionality. The proposed system has been validated with the support of real time data set from Frederick National Laboratory and the experimental results showed improved performance.