Sequential Segmentation of the Left Atrium and Atrial Scars Using a Multi-scale Weight Sharing Network and Boundary-based Processing
Published in LAScarQS 2022: Left Atrial and Scar Quantification & Segmentation Challenge in conjunction with STACOM and MICCAI 2022 (Sep 18th, 2022, Singapore), 2022
Our Contributions
- We introduce a new method for segmenting both the left atrium and atrial scars sequentially to improve accuracy and efficiency compared to existing techniques.
- The Multi-scale weight sharing network (MSWS-Net) allows for better feature extraction at different image resolutions, potentially improving the segmentation accuracy.
- Our Boundary2Patches method enhances the precision of segmentation, especially for complex structures like atrial scars.
- The proposed method combines anatomical information and texture features of the left atrium endocardium, leading to more robust segmentation.