Multiple sclerosis: identification of temporal changes in brain lesions with computer-assisted detection software.

Area of Science:

• Neurology
• Medical Imaging
• Computer Science

Background:

• Multiple sclerosis (MS) is a chronic, progressive neurological disease.
• Serial magnetic resonance imaging (MRI) is crucial for monitoring MS.
• Tracking changes in brain lesions is essential for patient management.

Purpose of the Study:

• To evaluate the performance of a developed computer-assisted detection (CAD) software for tracking temporal changes in MS brain lesions.
• To compare the diagnostic accuracy and interpretation time of CAD-assisted versus routine clinical MRI interpretation in MS patients.

Main Methods:

• Developed a CAD software pipeline to track T2-bright MS lesions on 3D FLAIR MR sequences.
• Compared CAD-assisted interpretation with routine clinical interpretation using 98 paired MR examinations from 88 MS patients.
• Established ground truth through a three-expert panel consensus.

Main Results:

• CAD interpretation demonstrated higher case-wise sensitivity (87%) compared to clinical reports (77%).
• Lesion-wise analysis showed improved sensitivity of CAD over routine interpretation by 40%-48%.
• Mean interpretation time for CAD-assisted analysis was 2.7 minutes.

Conclusions:

• The developed CAD software shows potential for integration into neuroradiology workflows for detecting MS lesion changes.
• Automated quantification of MS lesion load changes could benefit clinical research, including drug trials.