A hybrid approach based on logistic classification and iterative contrast enhancement algorithm for hyperintense...

Area of Science:

• Neuroimaging
• Medical image analysis
• Neurology

Background:

• Multiple sclerosis (MS) is a growing neurodegenerative disease impacting patients worldwide.
• Accurate assessment of MS lesions on MRI is crucial for disease management.
• Current automated lesion segmentation methods require improvement for clinical utility.

Purpose of the Study:

• To introduce and evaluate a novel hybrid lesion segmentation approach, LC+ICE, for MS.
• To compare the performance of LC+ICE against existing FLAIR lesion segmentation techniques.
• To assess the robustness and applicability of LC+ICE across different MS disease stages.

Main Methods:

• A hybrid approach combining logistic classification (LC) and iterative contrast enhancement (ICE) was developed.
• The LC+ICE method utilized T1 and FLAIR MRI images from 32 secondary progressive MS patients.
• Manual segmentation served as the ground truth for evaluating segmentation accuracy.

Main Results:

• The LC+ICE method demonstrated precise and robust lesion segmentation, validated by DICE, Sensitivity, Specificity, AUC, and Volume Similarity metrics.
• Performance was superior when compared to two recent FLAIR lesion segmentation approaches.
• Segmentation stability was observed across varying lesion loads, indicating broad applicability.

Conclusions:

• The LC+ICE method offers a precise and robust solution for hyperintense MS lesion segmentation.
• This automated technique serves as a valuable alternative to manual segmentation, aiding healthcare professionals.
• The LC+ICE approach shows potential for wide applicability across different stages of multiple sclerosis.