Synthetic vascular structure generation for unsupervised pre-training in CTA segmentation tasks

TL;DR

This paper introduces a method to generate synthetic vascular structures for pre-training deep learning models in CTA segmentation, reducing reliance on manual labeling and improving accuracy in stroke patient analysis.

Contribution

The study presents a novel synthetic vascular structure generation technique for unsupervised pre-training in CTA segmentation tasks.

Findings

Pre-trained models outperform those trained only on labeled data.

Synthetic data enhances segmentation accuracy.

Unsupervised pre-training reduces manual labeling effort.

Abstract

Large enough computed tomography (CT) data sets to train supervised deep models are often hard to come by. One contributing issue is the amount of manual labor that goes into creating ground truth labels, specially for volumetric data. In this research, we train a U-net architecture at a vessel segmentation task that can be used to provide insights when treating stroke patients. We create a computational model that generates synthetic vascular structures which can be blended into unlabeled CT scans of the head. This unsupervised approached to labelling is used to pre-train deep segmentation models, which are later fine-tuned on real examples to achieve an increase in accuracy compared to models trained exclusively on a hand-labeled data set.