Validating Nonlinear Registration to Improve Subtraction Images for Lesion Detection and Quantification in Multiple...

Area of Science:

• Medical Imaging
• Neurology
• Image Processing

Background:

• Multiple Sclerosis (MS) lesion detection and volume quantification are crucial for treatment monitoring.
• Subtraction imaging is a valuable tool but can be affected by artifacts.
• Improving registration techniques is key to enhancing subtraction imaging accuracy.

Purpose of the Study:

• To propose and validate nonlinear registration techniques for generating MRI subtraction images.
• To assess the ability of nonlinear registration to reduce artifacts and improve lesion detection and volume quantification in MS patients.

Main Methods:

• Postcontrast T1-weighted and T2-weighted MRI scans were acquired monthly for 20 relapsing-remitting MS patients over one year.
• Four registration algorithms (linear, halfway linear, nonlinear, nonlinear halfway) were evaluated.
• Subtraction images were generated and analyzed for lesion activity, artifacts, and volume changes by two independent observers.

Main Results:

• Lesion activity detection performance was similar across all registration techniques.
• Nonlinear registration provided lesion volume measurements closer to T2-weighted images.
• Nonlinear registration significantly reduced yin-yang artifacts compared to other methods.

Conclusions:

• Nonlinear registration is a promising technique for generating subtraction images in MS.
• It improves lesion activity detection and provides more accurate lesion volume estimates.
• Reduced artifacts with nonlinear registration can enhance statistical power in subtraction imaging studies.