Automatic segmentation of gadolinium-enhanced multiple sclerosis lesions

Area of Science:

• Medical Imaging
• Neurology
• Biomedical Engineering

Background:

• Accurate characterization of disease state in multiple sclerosis (MS) relies on detecting contrast-enhanced lesions on MRI.
• Automated analysis of lesion enhancement is complicated by enhancing structures like cerebral vasculature and blood-brain barrier disruptions.

Purpose of the Study:

• To develop and evaluate an automated method for detecting and quantifying contrast-enhanced lesions in multiple sclerosis (MS) patients using MRI.
• To overcome challenges in automated analysis caused by non-lesion enhancing structures.

Main Methods:

• A novel MRI pulse sequence incorporating stationary and marching saturation bands with gradient dephasing was used to suppress vascular enhancement.
• Automatic image segmentation was employed as a postprocessing technique to eliminate non-lesion enhancing structures, such as the choroid plexus.
• The developed technique was evaluated on 13 patients with MS.

Main Results:

• The automated method successfully identified all multiple sclerosis (MS) lesions larger than 5 mm3.
• No false-positive or false-negative lesions exceeding 5 mm3 were detected in the evaluated patients.
• The technique effectively suppressed enhancements from cerebral vasculature and eliminated structures like the choroid plexus.

Conclusions:

• The described automated MRI acquisition and postprocessing techniques are effective for accurate detection and quantitation of contrast-enhanced MS lesions.
• This method shows promise for reliable disease state characterization in multiple sclerosis (MS) by overcoming common image analysis artifacts.