Automatic Segmentation of Abdominal Aortic Aneurysms From Time-Resolved 3-D Ultrasound Images Using Deep Learning
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control · 2024-11 · 71 (11) · 1420-1428
DOI: 10.1109/tuffc.2024.3389553 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2024-11. Verfasst von Esther J Maas und 4 weiteren Autor:innen. Hauptthemen laut MEDLINE: Deep Learning, Aortic Aneurysm, Abdominal, Imaging, Three-Dimensional, Image Interpretation, Computer-Assisted. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Union’s Horizon 2020 Research and Innovation Program, Richard G. P. Lopata, 4TU Precision Medicine program.
Abstract (Originalsprache)
Abdominal aortic aneurysms (AAAs) are rupture-prone dilatations of the aorta. In current clinical practice, the maximal diameter of AAAs is monitored with 2-D ultrasound to estimate their rupture risk. Recent studies have shown that 3-D and mechanical AAA parameters might be better predictors for aneurysm growth and rupture than the diameter. These parameters can be obtained with time-resolved 3-D ultrasound (3-D + t US), which requires robust and automatic segmentation of AAAs from 3-D + t US. This study proposes and validates a deep learning (DL) approach for the automatic segmentation of AAAs. A group of 500 patients was included for follow-up 3-D + t US imaging, resulting in 2495 3-D + t US images. Segmentation masks for model training were obtained using a conventional automatic segmentation algorithm (nonDL). Four different DL models were trained and validated by 1) comparison to CT and 2) reader scoring. Performance of the nonDL and different DL segmentation strategies were evaluated by comparing Hausdorff distance, Dice scores, accuracy, sensitivity, and specificity with a sign test. All DL models had higher median Dice scores, accuracy, and sensitivity (all ) compared to nonDL segmentation. The full image-resolution model without data augmentation showed the highest median Dice score and sensitivity ( ). Applying the DL model on an independent test group produced fewer poor segmentation scores of 1 to 2 on a five-point scale (8% for DL, 18% for nonDL). This demonstrates that a robust and automatic segmentation algorithm for segmenting AAAs from 3-D + t US images was developed, showing improved performance compared to conventional segmentation.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Förderung
- European Union’s Horizon 2020 Research and Innovation Program
- Richard G. P. Lopata
- 4TU Precision Medicine program
- High Tech for a Sustainable Future
Zitieren
APA
Maas, E. J., Awasthi, N., van Pelt, E. G., van Sambeek, M. R. H. M., & Lopata, R. G. P. (2024). Automatic Segmentation of Abdominal Aortic Aneurysms From Time-Resolved 3-D Ultrasound Images Using Deep Learning. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 71(11), 1420-1428. https://doi.org/10.1109/tuffc.2024.3389553
Vancouver
Maas EJ, Awasthi N, van Pelt EG, van Sambeek MRHM, Lopata RGP. Automatic Segmentation of Abdominal Aortic Aneurysms From Time-Resolved 3-D Ultrasound Images Using Deep Learning. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 2024;71(11):1420-1428. doi:10.1109/tuffc.2024.3389553
BibTeX
@article{Maas2024,
author = {Maas, Esther J and Awasthi, Navchetan and van Pelt, Esther G and van Sambeek, Marc R H M and Lopata, Richard G P},
title = {Automatic Segmentation of Abdominal Aortic Aneurysms From Time-Resolved 3-D Ultrasound Images Using Deep Learning},
journal = {IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control},
year = {2024},
volume = {71},
number = {11},
pages = {1420--1428},
doi = {10.1109/tuffc.2024.3389553}
}
EUVIMED