Superpixel-based C-SVC for Brain Tissue Classification in MRI Scans.

Tareef, Afaf

Accurate identification of brain tissue is an ill-posed problem due to the inhomogeneous intensity and the extremely complicated and irregular border between endocrine tissues. This study introduces a superpixelbased approach to brain tissue classification in MRI scans. The proposed approach starts with image smoothing and feature highlighting, followed by image splitting based on the SLIC superpixel and merging strategy. Then, distinct superpixel-based appearance and boundary features are extracted and refined by minimizing redundancy and maximizing relevance technique before sending to the C-support vector classifier. Finally, a refinement step is adopted based on morphological characteristics and the distance regularized level set evolution model to modify the matter contour. The proposed approach was evaluated and compared with ten existing algorithms using the publicly accessible IBSR dataset. The experimental results show the better efficiency of the proposed approach in delimiting the contour of each matter than the other approaches in the literature.

FEATURE extraction; FEATURE selection; MAGNETIC resonance imaging; CLASSIFICATION; TISSUES

Engineering, Technology & Applied Science Research, 2024, Vol 14, Issue 6, p18271

2241-4487

Academic Journal

10.48084/etasr.9080