Automated segmentation of multiple sclerosis lesions in multispectral MR imaging using fuzzy clustering

Area of Science:

• Medical Imaging
• Neurology
• Artificial Intelligence

Background:

• Multiple Sclerosis (MS) diagnosis relies on identifying lesions in Magnetic Resonance (MR) imaging.
• Manual lesion detection is time-consuming and prone to inter-observer variability.
• Automated methods are needed to improve the efficiency and accuracy of MS lesion detection.

Purpose of the Study:

• To develop and evaluate a fully automated method for detecting Multiple Sclerosis (MS) lesions.
• To leverage multispectral MR imaging and Fuzzy C-Means (FCM) algorithm for enhanced lesion segmentation.
• To assess the performance of the automated method in a clinical setting.

Main Methods:

• A novel automated detection method based on the Fuzzy C-Means (FCM) algorithm.
• Initial segmentation to extract a CSF/lesions mask with local image contrast enhancement.
• Application of FCM to a masked image, followed by anatomical knowledge-based refinement to remove artifacts.
• Inclusion of a lesion size threshold to filter out small, irrelevant structures.

Main Results:

• The method successfully detected Multiple Sclerosis (MS) lesions in multispectral MR images.
• Automated segmentation and refinement steps were effective in isolating lesions.
• Initial tests on 10 patients demonstrated the potential of the automated approach.

Conclusions:

• The presented automated method shows promise for accurate and efficient Multiple Sclerosis (MS) lesion detection.
• Further validation and refinement are necessary to address limitations and optimize clinical application.
• This approach could aid in objective MS diagnosis and monitoring.