Deep Learning Framework for Automated MRI Planimetry in Multiple Sclerosis.

Area of Science:

• Neuroimaging
• Artificial Intelligence in Medicine
• Neurology

Background:

• Brain volume changes and infratentorial involvement are crucial for predicting disability in multiple sclerosis (MS).
• Magnetic resonance imaging (MRI) planimetry offers a robust method for assessing these changes, but traditional manual measurements are time-consuming and prone to bias.
• Current planimetry relies on manual, unblinded expert analysis, limiting its scalability and reproducibility.

Purpose of the Study:

• To develop and validate a fully automated deep learning framework for deriving brainstem planimetric measurements from MRI.
• To enhance the objectivity, reliability, and scalability of MRI planimetry for MS assessment.
• To support more accurate disease progression and treatment response monitoring in MS patients.

Main Methods:

• A deep learning pipeline was created, integrating automated midsagittal plane (MSP) detection.
• A convolutional neural network (CNN) was trained to perform segmentations essential for planimetry.
• The automated framework was validated against manual measurements across different MRI scanners and protocols.

Main Results:

• The automated framework demonstrated strong agreement with manual planimetric measurements.
• The method proved robust and consistent across various scanners and acquisition protocols.
• The deep learning approach successfully automated the complex segmentation tasks required for planimetry.

Conclusions:

• The proposed automated framework enables reliable, reproducible, and scalable MRI planimetry.
• This AI-driven approach can overcome limitations of manual measurements, reducing bias and time.
• The tool supports objective assessment of disease progression and treatment efficacy in multiple sclerosis.