FsBAD: Data-efficient feature reconstruction for few-shot brain anomaly detection

Data efficiency remains a central challenge in brain anomaly detection, where annotated datasets are often scarce. Most existing methods are tailored to single-class settings and show limited ability to generalize. We introduce FsBAD, a feature reconstruction-based approach designed for few-shot brain anomaly detection with minimal supervision. FsBAD reconstructs a nominal version of an anomalous brain scan by leveraging a small set of aligned reference samples. To enhance reconstruction quality, we propose a novel feature alignment strategy that integrates regression with distribution regularization, promoting both semantic accuracy and nominal consistency. While FsBAD is optimized for brain imaging, we evaluate its generalization capabilities on liver and retina datasets. Experiments across all three domains show that FsBAD consistently outperforms state-of-the-art methods in both image-wise classification and pixel-wise anomaly localization, even in extremely low-shot (2- to 15-shot) settings. This demonstrates FsBAD’s potential as a scalable, data-efficient solution for brain anomaly detection and its robustness across medical imaging tasks. • We propose FsBAD, a training-free brain anomaly detection method based on feature-level reconstruction and alignment with a query sample. • We introduce a distribution regularization strategy based on ridge regression to enforce nominal characteristics in the reconstructed sample while preserving structural similarity to the query. • We demonstrate strong generalization capabilities of FsBAD across multiple datasets and few-shot settings.