Related Experiment Video
Updated: Apr 5, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
43.8K
Improved Fuzzy C-Means based Particle Swarm Optimization (PSO) initialization and outlier rejection with level set
Abdenour Mekhmoukh1, Karim Mokrani1
1Laboratoire de Technologie Industrielle et de l'Information (LTII), Faculté de Technologie, Université de Bejaia, 06000 Bejaia, Algeria.
Computer Methods and Programs in Biomedicine
|August 25, 2015
Summary
This study introduces an improved image segmentation method for Magnetic Resonance (MR) images using Particle Swarm Optimization (PSO) and outlier rejection with level sets. The novel approach enhances accuracy by optimizing cluster centers and incorporating spatial information, outperforming traditional methods.
Area of Science:
- Medical Imaging
- Computer Vision
- Computational Intelligence
Background:
- Traditional Fuzzy C-Means (FCM) for Magnetic Resonance (MR) image segmentation is sensitive to noise, outliers, and image inhomogeneities.
- FCM lacks integration of spatial information and is dependent on random initialization of cluster centers, limiting its accuracy.
Purpose of the Study:
- To propose a novel image segmentation method for MR images that overcomes the limitations of conventional FCM.
- To enhance outlier rejection and reduce noise sensitivity in MR image segmentation.
Main Methods:
- A new method combining Particle Swarm Optimization (PSO) for optimal cluster center initialization and outlier rejection with a level set approach is presented.
- An extended Fuzzy C-Means (FCM) algorithm incorporating spatial neighborhood information into the cost function is developed.
- The optimized fuzzy clustering results are utilized to define the initial contour for the level set segmentation.
Main Results:
- The proposed method demonstrates improved outlier rejection and reduced sensitivity to noise and inhomogeneities compared to traditional FCM.
- Optimal initialization of cluster centers using PSO significantly enhances segmentation performance.
- The integration of spatial information further refines the segmentation accuracy for MR images.
Conclusions:
- The developed PSO-based, outlier-robust FCM with level sets offers a more effective approach for MR image segmentation.
- The method provides a robust and accurate solution for segmenting MR images, addressing key limitations of existing algorithms.
