PARSE challenge 2022: Pulmonary Arteries Segmentation using Swin U-Net Transformer(Swin UNETR) and U-Net

TL;DR

This paper presents an ensemble deep learning approach combining Swin UNETR and U-Net models for pulmonary arteries segmentation from CT scans, achieving an 84.36% dice score in the MICCAI PARSE 2022 challenge.

Contribution

The work introduces an ensemble of Swin UNETR and U-Net models specifically designed for pulmonary arteries segmentation, demonstrating improved accuracy in a challenge setting.

Findings

Achieved a multi-level dice score of 84.36%.

Ensemble of models improved segmentation performance.

Code is publicly available for reproducibility.

Abstract

In this work, we present our proposed method to segment the pulmonary arteries from the CT scans using Swin UNETR and U-Net-based deep neural network architecture. Six models, three models based on Swin UNETR, and three models based on 3D U-net with residual units were ensemble using a weighted average to make the final segmentation masks. Our team achieved a multi-level dice score of 84.36 percent through this method. The code of our work is available on the following link: https://github.com/akansh12/parse2022. This work is part of the MICCAI PARSE 2022 challenge.