Beyond transfer learning: a generative self-supervised framework for fMRI-based diagnosis on small and imbalanced datasets

Diagnosing neurological and psychiatric diseases (NeuroPsyD) from functional Magnetic Resonance Imaging (fMRI) using Deep Neural Networks (DNNs) is challenging, particularly when datasets are small and imbalanced, leading to severe overfitting, instability, and biased sensitivity–specificity trade-offs. Transfer Learning (TL), including supervised and self-supervised pre-training, can partially mitigate these challenges, but its effectiveness remains limited by domain shift, annotation bias, majority-class bias, and limited gains on very small datasets. To address these limitations, we propose a purely in-domain generative self-supervised framework that does not require external pre-training, called Boundary-aware Variational Autoencoder + Self-Supervised Mixup (BVAE+SSup-Mixup).

The framework integrates SSup-Mixup for label-free representation learning, multivariate VAE-based minority-class oversampling, and boundary-aware synthetic sample selection, which retains informative generated samples near the decision boundary while reducing outliers and poorly positioned synthetic samples. The framework is evaluated on five fMRI-based NeuroPsyD diagnosis tasks covering extremely and moderately small, imbalanced datasets. Compared with supervised and self-supervised TL approaches and a state-of-the-art RHVAE-based generative baseline, BVAE+SSup-Mixup improved accuracy, F1-score, and AUC across most tasks, achieving accuracies of 87.7–95.0% and AUCs of 89.3–94.7%.

It also produced a more balanced sensitivity–specificity profile, with both measures mostly above 85%. These balanced gains suggest that boundary-aware augmentation provides targeted minority-class evidence that TL may not fully exploit. These findings provide a proof of concept for a TL-competitive framework in severely data-limited and imbalanced medical imaging settings.

The authors show gratitude towards the contribution of the Iranian National Brain Mapping Laboratory (NBML), Tehran, Iran, for the data acquisition service. We also acknowledge data collection and sharing provided by the Autism Brain Imaging Data Exchange I (ABIDE I) consortium and the Preprocessed Connectomes Project for providing C-PAC derivatives. The second author (Saeed Masoudnia) gratefully acknowledges the Iran National Science Foundation (INSF) for the financial support he received through a postdoctoral research grant.

CIPCE, School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran Ershad Hassanpour Golagani, Saeed Masoudnia, Ahmad Kalhor & Hamid Soltanian-Zadeh School of Cognitive Science, Institute for Research in Fundamental Sciences (IPM), Tehran, Iran Departments of Radiology and Research Administration, Henry Ford Health System, Detroit, MI, USA Correspondence to Hamid Soltanian-Zadeh. The authors declare no competing interests. The TLE dataset was collected at the Iranian National Brain Mapping Laboratory (NMBL), Tehran, Iran.

The study protocol was approved by the Research Ethics Committee of Iran University of Medical Sciences (Ethics code: IR.IUMS.REC.1397.672). All participants provided written informed consent before participation. All methods were performed in accordance with relevant guidelines and regulations.