Back to Search
Start Over
UR-CarA-Net: A Cascaded Framework With Uncertainty Regularization for Automated Segmentation of Carotid Arteries on Black Blood MR Images
- Source :
- IEEE Access, Vol 11, Pp 26637-26651 (2023)
- Publication Year :
- 2023
- Publisher :
- IEEE, 2023.
-
Abstract
- We present a fully automated method for carotid artery (CA) outer wall segmentation in black blood MRI using partially annotated data and compare it to the state-of-the-art reference model. Our model was trained and tested on multicentric data of patients (106 and 23 patients, respectively) with a carotid plaque and was validated on different MR sequences (24 patients) as well as data that were acquired with MRI systems of a different vendor (34 patients). A 3D nnU-Net was trained on pre-contrast T1w turbo spin echo (TSE) MR images. A CA centerline sliding window approach was chosen to refine the nnU-Net segmentation using an additionally trained 2D U-Net to increase agreement with manual annotations. To improve segmentation performance in areas with semantically and visually challenging voxels, Monte-Carlo dropout was used. To increase generalizability, data were augmented with intensity transformations. Our method achieves state-of-the-art results yielding a Dice similarity coefficient (DSC) of 91.7% (interquartile range (IQR) 3.3%) and volumetric intraclass correlation (ICC) with ground truth of 0.90 on the development domain data and a DSC of 91.1% (IQR 7.2%) and volumetric ICC with ground truth of 0.83 on the external domain data outperforming top-ranked methods for open-source CA segmentation. The uncertainty-based approach increases the interpretability of the proposed method by providing an uncertainty map together with the segmentation.
Details
- Language :
- English
- ISSN :
- 21693536
- Volume :
- 11
- Database :
- Directory of Open Access Journals
- Journal :
- IEEE Access
- Publication Type :
- Academic Journal
- Accession number :
- edsdoj.3b30660a7ca54aec9bb756996e7c2b2b
- Document Type :
- article
- Full Text :
- https://doi.org/10.1109/ACCESS.2023.3258408