Infant Brain Age Estimation With T1w/T2w Ratio MRI: A Myelination- Aware Deep Learning Approach

• Background Brain age estimation provides a noninvasive MRI biomarker of neurodevelopment. In infancy, rapid regionally ordered myelination reflects brain maturation, yet early- life brain age estimation remains underexplored, particularly with myelination- sensitive MRI and biologically informed modeling.
• Purpose To develop and evaluate a biologically informed deep learning framework for infant brain age estimation using T1w/ T2w ratio MRI.
• Study Type Retrospective.
• Population Internal cohort: 629 infants aged 0–24 months (626 with age- appropriate myelination, train/validation/ test = 376/125/125), 3 with myelin- related developmental abnormalities for qualitative review. External cohort: 10 healthy infants aged 0–15 months (5 females, 5 males).
• Field Strength/Sequence Internal: 3T; 3D gradient- echo or 2D spin- echo T1w, and 2D turbo spin- echo T2w. External: 3T; 3D gradient- echo T1w and 2D turbo spin- echo T2w.
• Assessment 3D convolutional neural networks were trained with T1w, T2w, and T1w/T2w ratio inputs using manually defined biological age labels from visual myelination assessment. The model incorporated multi- task learning for age regression, white matter segmentation, and image reconstruction.
• Statistical Tests Performance was evaluated using five- fold cross- validation with repeated random splits. Metrics included mean absolute error, root mean squared error, R2, and Pearson and Spearman correlations. Modality differences were tested using one- way ANOVA, t- tests, and Mann–Whitney U, with Cohen's d and 95% confidence intervals. In the external co hort, absolute prediction errors were compared using the Wilcoxon signed- rank test. Statistical significance was defined as p<0.05.
• Results T1w/T2w ratio models achieved the best overall performance (MAE: 1.489 ±0.302 months; r = 0.966 ± 0.012), com pared with T1w (2.055 ± 0.944; 0.933 ± 0.061), T2w (1.794 ± 0.434; 0.947 ± 0.023), T1w+T2w (1.546 ± 0.291; 0.960 ± 0.013), and T1w+T2w+RI (1.498 ± 0.313; 0.963±0.012). Modality effects were significant for MAE, RMSE, R2, r, but not for (p=0.250). Auxiliary- task and multi- scale modeling numerically improved performance (MAE, 1.203 months; r = 0.979). External vali dation showed the lowest error for the RI- based model (MAE, 1.16 months), and Grad- CAM highlighted myelination- relevant white matter. © 2026 International Society for Magnetic Resonance in Medicine

https://doi.org/10.1002/jmri.70376