Segmentation of Neonatal Brain using MR Images in an Efficient Manner

Abstract

Image analysis using updated technology of magnetic resonance for finding, measuring and studying various tissue related structure of brain and thus discovering its medical region is an important application of segmentation process. In order to analyze the specific regions of brain, brain image segmentation plays a significant role for researchers and clinicians. In this work, we make an attempt to design an efficient segmentation model of neonatal brain MRI images of preterm infants. Initially, the dataset is collected from an eminent public repository that composes of numerous training and testing datasets. The proposed framework comprises of six phases, viz, pre-processing using FANFMF, Contrast enhancement using AAIHE, Feature extraction using PBDLFL, Affinity information using SCMMAL, Dictionary creation using DCAD and clustering using SSMLC. The main aim of this paper is to increase segmentation accuracy in the given MR images. The extraction of local features is a complex task which is simply achieved by the proposed PBDLFL via DCAD. The formation of selfsimilarity map from the probabilistic dictionary creation helps for better segmentation process. Finally clustering based segmentation process using SSMLC algorithm is used that that helps in decreasing uncertainty and sparsity of data so that an efficient diagnosis system can be obtained. Segmentation process that is proposed in this paper can be proved as accurate and efficient by various experimental result.