# Evaluating the Generalization of Neuroimaging Foundation Models on African Brain MRI

arXiv: https://arxiv.org/abs/2609.23983 · HTML: https://mlsanity.com/paper/2609.23983/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 18 references that are in the map.
- First posted on arXiv 21 Sep 2026.

## Builds on (most cited first, 8 of 8)

- [Automated brain extraction of multi-sequence MRI using artificial neural networks](https://mlsanity.com/paper/1901.11341/) (2019)
- [VoCo: A Simple-yet-Effective Volume Contrastive Learning Framework for 3D Medical Image Analysis](https://mlsanity.com/paper/2402.17300/) (2024)
- [An OpenMind for 3D medical vision self-supervised learning](https://mlsanity.com/paper/2412.17041/) (2024)
- [Primus: Enforcing Attention Usage for 3D Medical Image Segmentation](https://mlsanity.com/paper/2503.01835/) (2025)
- [Health system learning achieves generalist neuroimaging models](https://mlsanity.com/paper/2511.18640/) (2025)
- [A labeled Clinical-MRI dataset of Nigerian brains](https://mlsanity.com/paper/2311.04425/) (2023)
- [Benchmarking Foundation Models and Parameter-Efficient Fine-Tuning for Prognosis Prediction in Medical Imaging](https://mlsanity.com/paper/2506.18434/) (2025)
- [Learning Sparse Latent Predictive Foundation Model for Multimodal Neuroimaging](https://mlsanity.com/paper/2606.14957/) (2026)

## Abstract (arXiv)

Neuroimaging foundation models pretrained on large, predominantly western cohorts are increasingly proposed as general-purpose backbones for brain MRI analysis. Yet, their ability to generalize to underrepresented clinical populations remains largely untested. We evaluate four recent foundation models (BrainIAC, Neuro-JEPA, NeuroVFM, and Primus) on a three-way diagnostic classification task (Control, Dementia, Parkinson's disease) using a cohort of 88 subjects from a Nigerian clinical brain MRI dataset, across four modality configurations (T1w, T2w, T1w+T2w, FLAIR), and compare against an end-to-end trained ViT3D baseline. The frozen backbones collapse to majority-class predictions, while Neuro-JEPA on FLAIR shows modest but still limited discrimination. In contrast, the end-to-end trained ViT3D achieves higher accuracy and MCC on every task (up to 53.4% accuracy, MCC=0.27) and is the only model with non-trivial recall. Our findings suggest that these frozen neuroimaging foundation models are insufficient for fine-grained diagnostic classification in small, non-western clinical cohorts, motivating parameter-efficient adaptation and broader multi-site external validation for equitable deployment in global health settings.