[Research on K-means clustering segmentation method for MRI brain image based on selecting multi-peaks in gray...

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Traditional K-means clustering for MRI brain image segmentation suffers from random initialization, leading to inefficiencies and inaccuracies.
• Accurate segmentation of White Matter, Gray Matter, and Cerebrospinal Fluid is crucial for neurological studies.

Purpose of the Study:

• To develop a novel K-means segmentation algorithm for MRI brain images that overcomes the limitations of random initial center selection.
• To improve the effectiveness, accuracy, and robustness of brain tissue segmentation.

Main Methods:

• Proposed a new K-means algorithm utilizing robust selection of histogram 'peaks' from multi-peak gray histograms of MRI brain images.
• Utilized the gray values of selected histogram 'peaks' as initial clustering centers for K-means.
• Segmented MRI brain images into White Matter, Gray Matter, and Cerebrospinal Fluid.

Main Results:

• The proposed algorithm effectively segments MRI brain images into three tissue types.
• Demonstrated superior effectiveness, accuracy, and robustness compared to traditional K-means clustering.
• Overcame shortcomings of traditional methods including low efficiency, veracity, and time consumption.

Conclusions:

• The novel K-means algorithm based on histogram peak selection provides a more effective and reliable method for MRI brain image segmentation.
• The histogram 'peak' selection strategy offers broad applicability for segmentation tasks.