Anatomically-guided masked autoencoder pre-training for aneurysm detection

TL;DR

This paper introduces a novel 3D Vision Transformer pre-training method using masked auto-encoders tailored for aneurysm detection in head CT scans, improving sensitivity over state-of-the-art models.

Contribution

It proposes a new pre-training strategy with a factorized self-attention mechanism and artery-focused masking, enhancing aneurysm detection performance.

Findings

Achieved +4-8% sensitivity over SOTA models.

Utilized artery distance maps to improve feature learning.

Demonstrated effectiveness of unannotated data pre-training.

Abstract

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.