Search papers, labs, and topics across Lattice.
The paper introduces D$^2$4FAD, a dual distillation framework for few-shot anomaly detection in medical imaging, addressing the limitations of existing unsupervised methods that require extensive normal training data and struggle with generalization. D$^2$4FAD employs a pre-trained encoder as a teacher to extract multi-scale features from support and query images, while a student decoder learns via knowledge distillation on query images and self-distillation on support images. Experiments on a new benchmark dataset of 13,084 medical images demonstrate that D$^2$4FAD achieves state-of-the-art performance in few-shot anomaly detection, outperforming existing methods.
Achieve state-of-the-art few-shot anomaly detection in medical imaging by dynamically weighting reference images based on their relevance to the query, enabling effective knowledge transfer from limited normal examples.
Anomaly detection is a critical task in computer vision with profound implications for medical imaging, where identifying pathologies early can directly impact patient outcomes. While recent unsupervised anomaly detection approaches show promise, they require substantial normal training data and struggle to generalize across anatomical contexts. We introduce D$^2$4FAD, a novel dual distillation framework for few-shot anomaly detection that identifies anomalies in previously unseen tasks using only a small number of normal reference images. Our approach leverages a pre-trained encoder as a teacher network to extract multi-scale features from both support and query images, while a student decoder learns to distill knowledge from the teacher on query images and self-distill on support images. We further propose a learn-to-weight mechanism that dynamically assesses the reference value of each support image conditioned on the query, optimizing anomaly detection performance. To evaluate our method, we curate a comprehensive benchmark dataset comprising 13,084 images across four organs, four imaging modalities, and five disease categories. Extensive experiments demonstrate that D$^2$4FAD significantly outperforms existing approaches, establishing a new state-of-the-art in few-shot medical anomaly detection. Code is available at https://github.com/ttttqz/D24FAD.
