Advanced Multiple Sclerosis Data Analysis Using Intensity-Guided Skull Removal and Level Set Method for Enhanced Accuracy

Abstract

Edge detection and segmentation are an important medical image processing task based on many mathematical frameworks like level set method. It provides many advancements over conventional approaches, including the ability to handle complicated topological changes. Efficient differentiation of the lesions associated with multiple sclerosis from medical imaging is vital for the diagnosis and management of many clinical diseases. This study introduces a method to automatically segmenting lesions using an adaptive thresholding followed by a level-set method. The suggested approach tried to help solve the difficulties caused by intensity inhomogeneity, which typically occures in magnetic resonance images of multiple secleroses lesions. The experimental results showed that the proposed method is efficient, which shows good improvements in segmentation accuracy and outlining lesion boundaries. The integration of adaptive thresholding in the pre-processing stage helped in edge enhancement, resulting in more efficient edge detection in the second step. The sensitivity of the suggested method was 0.923. which means that this method is helpful for lesion detection. while, specificity is 0.936, which means this method is accurate in identifying non-lesion areas. The Dice Coefficient, which measures the overlap between the segmented lesions and the ground truth around 0.994, which is quite good. This study tried to enhance the dominion of medical image analysis by introducing a new method for the automatic segmentation of multiple sclerosis lesions from magnetic resonance images. This technique provided valuable support to clinicians in enhancing the precision of disease evaluation and treatment planning.