An Automated Brain Tumor Segmentation Technique in MRI Scans Using Modified Fuzzy C-Means Clustering

International Journal of Computer Science and Technology | Vol. 11, No. 10, October 2009 | pp. 73–80

DOI: 10.46882/2009/IJCST/000019

Article Type: Research Paper

Title: An Automated Brain Tumor Segmentation Technique in MRI Scans Using Modified Fuzzy C-Means Clustering

Names of Authors: Dr. Sophia Laurent¹, Jean-Pierre Dubois²

Authors’ Affiliations: ¹Medical Imaging Processing Laboratory, Sorbonne University, Paris, France; ²Department of Bioengineering, University of Lyon, Lyon, France

Abstract: Accurate segmentation of brain tumors from Magnetic Resonance Imaging (MRI) is vital for radiotherapy planning and tumor growth monitoring. Manual segmentation remains time-consuming and subjective to inter-observer variability. This paper introduces an automated segmentation methodology utilizing a Modified Fuzzy C-Means (MFCM) clustering algorithm that incorporates spatial contextual constraints to handle intensity inhomogeneities. The proposed algorithm minimizes objective functions by adjusting fuzzy membership grades based on immediate neighborhood pixel attributes. We validated the approach on 45 clinical T1-weighted, T2-weighted, and Fluid-Attenuated Inversion Recovery (FLAIR) MRI data slices. The MFCM approach yielded an average Dice Similarity Coefficient of 0.912 and a Jaccard Index of 0.835 against manual segmentations performed by expert radiologists. The statistical analysis demonstrates that spatial bounding integration limits false cluster assignments caused by salt-and-pepper noise by 34.2%. We conclude that the MFCM framework serves as an effective diagnostic aid for rapid, objective oncological boundaries assessment.

Keywords: Brain Tumor Segmentation, Medical Image Processing, Fuzzy C-Means, Magnetic Resonance Imaging, Spatial Constraints, Computer-Aided Diagnosis

Manuscript Timeline: Received: February 15, 2009; Revised: April 3, 2009; Accepted: May 2, 2009; Published: October 19, 2009

Citation: Laurent, S., & Dubois, J. (2009). An Automated Brain Tumor Segmentation Technique in MRI Scans Using Modified Fuzzy C-Means Clustering. International Journal of Computer Science and Technology, 11(10), 73–80.

Subscribe to read the full article: https://www.internationalscholarsjournals.org/subscribe-to-read