Anatomically-guided masked autoencoder pre-training for aneurysm detection
This work addresses the challenge of limited training data for automated aneurysm detection, which is crucial for reducing morbidity and mortality, but it is incremental as it builds on existing masked auto-encoder methods.
The authors tackled the problem of detecting intracranial aneurysms from head CT scans by proposing a novel pre-training strategy using unannotated data, which improved sensitivity by 4-8% absolute at a false positive rate of 0.5 compared to state-of-the-art models.
Intracranial aneurysms are a major cause of morbidity and mortality worldwide, and detecting them manually is a complex, time-consuming task. Albeit automated solutions are desirable, the limited availability of training data makes it difficult to develop such solutions using typical supervised learning frameworks. In this work, we propose a novel pre-training strategy using more widely available unannotated head CT scan data to pre-train a 3D Vision Transformer model prior to fine-tuning for the aneurysm detection task. Specifically, we modify masked auto-encoder (MAE) pre-training in the following ways: we use a factorized self-attention mechanism to make 3D attention computationally viable, we restrict the masked patches to areas near arteries to focus on areas where aneurysms are likely to occur, and we reconstruct not only CT scan intensity values but also artery distance maps, which describe the distance between each voxel and the closest artery, thereby enhancing the backbone's learned representations. Compared with SOTA aneurysm detection models, our approach gains +4-8% absolute Sensitivity at a false positive rate of 0.5. Code and weights will be released.