Comparison of atlas-based and deep learning methods for organs at risk delineation on head-and-neck CT images using an automated treatment planning system.
Atlas-based methods
Automatic contouring
Automatic planning
Deep-Learning methods
Head-and-Neck Cancer
Organs-at-risk
Journal
Radiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology
ISSN: 1879-0887
Titre abrégé: Radiother Oncol
Pays: Ireland
ID NLM: 8407192
Informations de publication
Date de publication:
12 2022
12 2022
Historique:
received:
20
05
2022
revised:
21
10
2022
accepted:
23
10
2022
pubmed:
4
11
2022
medline:
21
12
2022
entrez:
3
11
2022
Statut:
ppublish
Résumé
To investigate the performance of head-and-neck (HN) organs-at-risk (OAR) automatic segmentation (AS) using four atlas-based (ABAS) and two deep learning (DL) solutions. All patients underwent iodine contrast-enhanced planning CT. Fourteen OAR were manually delineated. DL.1 and DL.2 solutions were trained with 63 mono-centric patients and > 1000 multi-centric patients, respectively. Ten and 15 patients with varied anatomies were selected for the atlas library and for testing, respectively. The evaluation was based on geometric indices (DICE coefficient and 95th percentile-Hausdorff Distance (HD Both DICE and HD DL methods generally showed higher delineation accuracy compared with ABAS methods for AS segmentation of HN OAR. Most ABAS contours had high conformity to the reference but were more time consuming than DL algorithms, especially when considering the computing time and the time spent on manual corrections.
Sections du résumé
BACKGROUND AND PURPOSE
To investigate the performance of head-and-neck (HN) organs-at-risk (OAR) automatic segmentation (AS) using four atlas-based (ABAS) and two deep learning (DL) solutions.
MATERIAL AND METHODS
All patients underwent iodine contrast-enhanced planning CT. Fourteen OAR were manually delineated. DL.1 and DL.2 solutions were trained with 63 mono-centric patients and > 1000 multi-centric patients, respectively. Ten and 15 patients with varied anatomies were selected for the atlas library and for testing, respectively. The evaluation was based on geometric indices (DICE coefficient and 95th percentile-Hausdorff Distance (HD
RESULTS
Both DICE and HD
CONCLUSION
DL methods generally showed higher delineation accuracy compared with ABAS methods for AS segmentation of HN OAR. Most ABAS contours had high conformity to the reference but were more time consuming than DL algorithms, especially when considering the computing time and the time spent on manual corrections.
Identifiants
pubmed: 36328093
pii: S0167-8140(22)04524-8
doi: 10.1016/j.radonc.2022.10.029
pii:
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
61-70Informations de copyright
Copyright © 2022 Elsevier B.V. All rights reserved.
Déclaration de conflit d'intérêts
Declaration of Competing Interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: This work was performed in the framework of a research cooperation agreement with Elekta AB.