MRI - Based Brain Age Prediction Using 3D CNN and XAI

Abstract

Brain age is an important biomarker that quantifies age-related structural changes in the human brain, with the potential for early disease diagnosis and monitoring of healthy aging. However, conven-tional three-dimensional (3D) convolutional neural networks require substantial computational resources to achieve high accuracy. In this study, we propose a computationally efficient deep learning model based on two-dimensional (2D) projections that balances efficiency and accuracy. We integrated eight publicly available datasets comprising 7,649 healthy participants aged 5–89 years, using T1-weighted magnetic resonance imaging. In addition to the gray matter probability maps, we incorporated full-brain structural information and multiple projection statistics. These statistics include the mean, standard deviation, median, and maximum to capture the comprehensive morphological features. The proposed architecture comprises only three convolutional blocks with 414,785 parameters and an 86% reduction compared to the similarly highperformance simple fully convolutional network (SFCN). To mitigate systematic prediction biases, we implemented age-distribution-weighted training. The experimental results indicated that single-plane models achieved a mean absolute error (MAE) of approximately 2.7–2.8 years, whereas a three-plane ensemble reduced the error to 2.50 years. After bias correction, the error was 2.54 years, effectively mitigating age-related bias while maintaining accuracy. However, prediction reliability in the middle-aged subgroup remains limited owing to data scarcity, with MAE reaching 7.48 years in the 40–49 age range. The model outperformed the existing 2D projection methods with extremely low computational complexity, requiring only approximately 1.5 h of training. This training time is nearly two orders of magnitude faster than that of the 3D approaches. Furthermore, gradient-weighted class activation mapping visualizations confirmed biological plausibility, highlighting aging-related regions such as the ventricles, cortex, and hippocampus.

Introduction

Brain ageing is a natural process that causes gradual structural and functional changes in the brain, such as reduced grey matter, cortical thinning, white matter degradation, and ventricular enlargement. While this process varies across individuals, some experience accelerated brain ageing, which can lead to cognitive decline and increased risk of neurological disorders. Early detection is important but challenging because current methods rely on manual MRI analysis and cognitive tests, which are time-consuming, subjective, and limited in capturing early structural changes.

Traditional machine learning methods depend on handcrafted features and struggle with complex 3D MRI data, while deep learning models improve accuracy but often act as “black boxes” without explainable results. They also lack risk classification, human impact analysis, and personalized recommendations, making them less useful in clinical practice.

Conclusion

In conclusion, this project successfully develops an intelligent, MRI-based brain ageing prediction system that combines a 3D Convolutional Neural Network with Explainable AI techniques for accurate and interpretable analysis. The system effectively processes structural brain MRI scans to estimate biological brain age, compute the brain age gap, and classify individuals into meaningful risk levels. The integration of Grad-CAM enhances transparency by highlighting critical brain regions that influence the prediction, improving trust and clinical interpretability. In addition to numerical prediction, the system provides ageing cause inference, human impact assessment, and personalized recommendations, making it a comprehensive decision-support tool rather than a simple prediction model. Performance evaluation using metrics such as Accuracy, Sensitivity, Specificity, F1-Score, and AUC demonstrates strong classification capability and reliable model generalization. The confusion matrix and ROC analysis further confirm balanced detection of both normal and accelerated brain ageing cases. Thus, the system offers a robust, explainable, and practically applicable framework for early brain health assessment. It supports preventive healthcare strategies, assists clinical decision-making, and contributes to advancements in AI-driven medical imaging research.

References

[1] A. M. Fjell, K. B. Walhovd, C. FennemaNotestine, L. K. McEvoy, D. J. Hagler, D. Holland, J. B. Brewer, and A. M. Dale, ‘‘One-year brain atrophy evident in healthy aging,’’ J. Neurosci., vol. 29, no. 48, pp. 15223–15231, Dec. 2009. [2] H.Neeb, K. Zilles, and N. J. Shah, ‘‘Fullyautomated detection of cerebral water content changes: Study of age-andgender-related H2Opatternswith quantitative MRI,’’ NeuroImage, vol. 29, no. 3, pp. 910–922, Feb. 2006. [3] C. Davatzikos, F. Xu, Y. An, Y. Fan, and S. M. Resnick, ‘‘Longitudi nal progression of [4] Alzheimer’s-like patterns of atrophy in normal older adults: The SPARE-AD index,’’ Brain, vol. 132, no. 8, pp. 2026–2035, Aug. 2009. [5] K. Franke, M. Ristow, and C. Gaser, ‘‘Genderspecific impact of personal health parameters on individual brain aging in cognitively unimpaired elderly subjects,’’ Frontiers Aging Neurosci., vol. 6, May 2014, Art. no. 94. [6] M. Habes, G. Erus, J. B. Toledo, T. Zhang, N. Bryan, L. J. Launer, Y. Rosseel, D. Janowitz, J. Doshi, S. Van der Auwera, B. von Sarnowski, K. Hegenscheid, N. Hosten, G. Homuth, H. Völzke, U. Schminke, W. Hoffmann, H. J. Grabe, and C. Davatzikos, ‘‘White matter hyperinten sities and imaging patterns of brain ageing in the general population,’’ Brain, vol. 139, no. 4, pp. 1164–1179, Apr. 2016. [7] J. H. Cole and K. Franke, ‘‘Predicting age using neuroimaging: Inno vative brain ageing biomarkers,’’ Trends Neurosciences, vol. 40, no. 12, pp. 681–690, Dec. 2017. [8] T.Kaufmannetal.,‘‘Commonbraindisordersareass ociatedwithheritable patterns of apparent aging of the brain,’’ Nat. Neurosci., vol. 22, no. 10, pp. 1617– 1623, Oct. 2019. [9] J. H. Cole, R. P. K. Poudel, D. Tsagkrasoulis, M. W. A. Caan, C. Steves, T. D. Spector, and G. Montana, ‘‘Predicting brain age with deep learning from raw imaging data results in a reliable and heritable biomarker,’’ NeuroImage, vol. 163, pp. 115–124, Dec. 2017. [10] K. Franke and C. Gaser, ‘‘Ten years of BrainAGE as a neuroimaging biomarker of brain aging: What insights have we gained?’’ Frontiers Neurol., vol. 10, Aug. 2019, Art. no. 789. [11] C. A. Raji, O. L. Lopez, L. H. Kuller, O. T. Carmichael, and J. T. Becker, ‘‘Age, Alzheimer disease, and brain structure,’’ Neurology, vol. 73, no. 22, pp. 1899–1905, Dec. 2009. [12] C. Gaser, K. Franke, S. Klöppel, N. Koutsouleris, and H. Sauer, ‘‘BrainAGE in mild cognitive impaired patients: Predicting the conver sion to Alzheimer’s disease,’’ PLoS ONE, vol. 8, no. 6, Jun. 2013, Art. no. e67346. [13] G. M. McAlonan, ‘‘Brain anatomy and sensorimotor gating in Asperger’s syndrome,’’ Brain, vol. 125, no. 7, pp. 1594–1606, Jul. 2002. [14] E. H. Aylward, N. J. Minshew, K. Field, B. F. Sparks, and N. Singh, ‘‘Effects of age on brain volume and head circumference in autism,’’ Neurology, vol. 59, no. 2, pp. 175–183, Jul. 2002. [15] P.Shaw,K.Eckstrand, W. Sharp, J. Blumenthal, J. P. Lerch, D. Greenstein, L. Clasen, A. Evans, J. Giedd, and J. L. Rapoport, ‘‘Attention deficit/hyperactivity disorder is characterized by a delay in cortical maturation,’’ Proc. Nat.Acad.Sci.USA,vol.104,no.49,pp. 19649– 19654, Dec. 2007. [16] M.Hoogmanetal.,‘‘Brain imaging of the cortex in ADHD: Acoordinated analysis of large-scale clinical and population-based samples,’’ Am. J. Psychiatry, vol. 176, no. 7, pp. 531–542, Jul. 2019 [17] S. Shahab, B. H. Mulsant, M. L. Levesque, N. Calarco, A. Nazeri, A. L. Wheeler, G. Foussias, T. K. Rajji, and A. N. Voineskos, ‘‘Brain structure, cognition, and brain age in schizophrenia, bipolar disorder, and healthy controls,’’ Neuropsychopharmacology, vol. 44, no. 5, pp. 898– 906, Apr. 2019. [18] T. Hajek, K. Franke, M. Kolenic, J. Capkova, M. Matejka, L. Propper, R. Uher, P. Stopkova, T. Novak, T. Paus, M. Kopecek, F. Spaniel, and M. Alda,‘‘Brain age in early stages of bipolar disorders or schizophrenia,’’ Schizophrenia Bull., vol. 45, no. 1, pp. 190–198, Jan. 2019. [19] L. K. M. Han, H. G. Schnack, R. M. Brouwer, D. J. Veltman, N. J. A. van der Wee,M.J.vanTol,M.Aghajani,andB. W. J. H. Penninx, ‘‘Contributing factors to advanced brain aging in depression and anxiety disorders,’’ Transl. Psychiatry, vol. 11, no. 1, Jul. 2021, Art. no. 402. [20] W. H. Lee, M. Antoniades, H. G. Schnack, R. S. Kahn, and S. Frangou, ‘‘Brain age prediction in schizophrenia: Does the choice of machine learning algorithm matter?’’ Psychiatry Res., Neuroimaging, vol. 310, Apr. 2021, Art. no. 111270. [21] A. Krizhevsky, I. Sutskever, and G. E. Hinton, ‘‘ImageNet classification with deep convolutional neural networks,’’ in Proc. Adv. Neural Inf. Pro cess. Syst., vol. 25, 2012, pp. 1097–1105. [22] Y. LeCun, Y. Bengio, and G. Hinton, ‘‘Deep learning,’’ Nature, vol. 521, no. 7553, pp. 436–444, May 2015. [23] H. Peng, W. Gong, C. F. Beckmann, A. Vedaldi, and S. M. Smith, ‘‘Accu rate brain age prediction with lightweight deep neural networks,’’ Med. Image Anal., vol. 68, Feb. 2021, Art. no. 101871. [23] V. M. Bashyam et al., ‘‘MRI signatures of brain age and disease over the lifespan based onadeepbrainnetworkand14468individualsworldwi de,’’ Brain, vol. 143, no. 7, pp. 2312–2324, Jul. 2020. [24] U. Gupta, P. K. Lam, G. V. Steeg, and P. M. Thompson, ‘‘Improved brain age estimation with slice-based set networks,’’ in Proc. IEEE 18th Int. Symp. Biomed. Imag. (ISBI), Apr. 2021, pp. 840– 844. [25] J. Jönemo, M. U. Akbar, R. Kämpe, J. P. Hamilton, and A. Eklund, ‘‘Effi cient brain age prediction from 3D MRI volumes using 2D projections,’’ Brain Sci., vol. 13, no. 9, p. 1329, Sep. 2023. [26] H. Wang, M. S. Treder, D. Marshall, D. K. Jones, and Y. Li, ‘‘A skewed loss function for correcting predictive bias in brain age prediction,’’ IEEE Trans. Med. Imag., vol. 42, no. 6, pp. 1577–1589, Jun. 2023. [27] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, ‘‘Grad-CAM: Visual explanations from deep networks via gradient-based localization,’’ in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), Oct. 2017, pp. 618–626. [28] T. Langner, J. Wikström, T. Bjerner, H. Ahlström, and J. Kullberg, ‘‘Identifying morphological indicators of aging with neural networks on large-scale whole-body MRI,’’ IEEE Trans. Med. Imag., vol. 39, no. 5, pp. 1430–1437, May 2020. [29] A. N. V. Ruigrok, G. Salimi-Khorshidi, M.-C. Lai, S. Baron-Cohen, M. V. Lombardo, R. J. Tait, and J. Suckling, ‘‘A meta-analysis of sex dif ferences in human brain structure,’’ Neurosci. Biobehavioral Rev., vol. 39, pp. 34–50, Feb. 2014. [30] [30] P. Hartmann, A. Ramseier, F. Gudat, M. J. Mihatsch, and W. Polasek, ‘‘Normal weight of the brain in adults in relation to age, sex, body height and weight,’’ Der Pathologe, vol. 15, no. 3, pp. 165–170, May 1994. [31] C. E. Coffey, J. F. Lucke, J. A. Saxton, G. Ratcliff, L. J. Unitas, B. Billig, and R. N. Bryan, ‘‘Sex differences in brain aging: A quantitative magnetic resonance imaging study,’’ Arch. Neurol., vol. 55, no. 2, pp. 169–179, Feb. 1998. [32] Y. Wang, Q. Xu, J. Luo, M. Hu, and C. Zuo, ‘‘Effects of age and sex on subcortical, volumes,’’ Frontiers Aging Neurosci., vol. 11, Sep. 2019, Art. no. 259. [33] J. Golomb, M. J. de Leon, A. Kluger, A. E. George, C. Tarshish, and S. H. Ferris, ‘‘Hippocampal atrophy in normal aging: An association with recent memory impairment,’’ Arch. Neurol., vol. 50, no. 9, pp. 967–973, Sep. 1993. [34] N. M. Armstrong, Y. An, L. Beason-Held, J. Doshi, G. Erus, L. Ferrucci, C. Davatzikos, and S. M. Resnick, ‘‘Sex differences in brain aging and predictors of neurodegeneration in cognitively healthy older adults,’’ Neu robiol. Aging, vol. 81, pp. 146–156, Sep. 2019. [35] B. A. Jonsson, G. Bjornsdottir, T. E. Thorgeirsson, L. M. Ellingsen, G. B. Walters, D. F. Gudbjartsson, H. Stefansson, K. Stefansson, and M. O. Ulfarsson,‘‘Brain age prediction using deep learning uncovers asso ciated sequence variants,’’ Nature Commun., vol. 10, no. 1, Nov. 2019, Art. no. 5409. [36] J. Cheng, Z. Liu, H. Guan, Z. Wu, H. Zhu, J. Jiang, W. Wen, D. Tao, and T. Liu, ‘‘Brain age estimation from MRI using cascade networks with ranking loss,’’ IEEE Trans. Med. Imag., vol. 40, no. 12, pp. 3400–3412, Dec. 2021. [37] Y. Joo, E. Namgung, H. Jeong, I. Kang, J. Kim, S. Oh, I. K. Lyoo, S. Yoon, and J. Hwang, ‘‘Brain age prediction using combined deep convolutional neural network and multi-layer perceptron algorithms,’’ Sci. Rep., vol. 13, no. 1, Dec. 2023, Art. no. 22388. [38] H.Jiang, N. Lu, K. Chen, L. Yao, K. Li, J. Zhang, and X. Guo, ‘‘Predicting brain age of healthy adults based on structural MRI parcellation using convolutional neural networks,’’ Frontiers Neurol., vol. 10, Jan. 2020, Art. no. 1346.

Copyright

Copyright © 2026 Nandhini S, Arthi S, Pavya P, Kiruthika M, Mrs. Vinotha Vasuki. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.