BrainAgeNeXt: Advancing brain age modeling for individuals with multiple sclerosis.

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Neurology

Background:

• Aging impacts brain structure and cognitive function, often preceding neurodegenerative diseases.
• Brain age, derived from MRI, quantifies structural aging and may predict neurodegeneration.
• Existing brain age prediction methods vary in accuracy and robustness.

Purpose of the Study:

• To develop and validate BrainAgeNeXt, a novel convolutional neural network for brain age prediction from T1-weighted MRI.
• To compare BrainAgeNeXt's performance against state-of-the-art methods across diverse datasets and image qualities.
• To investigate the utility of brain age as a prognostic biomarker in multiple sclerosis (MS).

Main Methods:

• BrainAgeNeXt, inspired by MedNeXt, was trained on 11,574 MRI scans (ages 5-95) from 33 datasets (3T and 7T).
• Performance was evaluated using Mean Absolute Error (MAE) against three established methods.
• Methods were tested on varying image quality, including motion artifacts and 7T data.
• Brain age was analyzed in three longitudinal MS cohorts (273 individuals).

Main Results:

• BrainAgeNeXt achieved a superior MAE of 2.78 ± 3.64 years, outperforming existing methods (MAE 3.55-4.16 years).
• BrainAgeNeXt demonstrated robustness across different image qualities, including 7T MRI and motion artifacts.
• In MS cohorts, brain age exceeded chronological age by 4.21 ± 6.51 years, increasing by 1.15 years per chronological year.
• Worsening disability in early MS correlated with a higher annual increase in brain age (1.24 vs 0.75 years/year).

Conclusions:

• BrainAgeNeXt is an accurate and robust tool for brain age prediction from MRI.
• Brain age serves as a sensitive indicator of accelerated aging in multiple sclerosis.
• Brain age shows potential as a prognostic biomarker for MS progression and a clinical trial endpoint.