Comparison of Deep Learning-Based and Patch-Based Methods for Pseudo-CT Generation in MRI-Based Prostate Dose Planning.
Bone and Bones
/ diagnostic imaging
Deep Learning
Femur Head
/ diagnostic imaging
Humans
Magnetic Resonance Imaging
/ methods
Male
Pelvis
/ diagnostic imaging
Prostate
/ diagnostic imaging
Prostatic Neoplasms
/ diagnostic imaging
Radiotherapy Dosage
Radiotherapy, Intensity-Modulated
/ methods
Rectum
/ diagnostic imaging
Reference Values
Tomography, X-Ray Computed
/ classification
Uncertainty
Urinary Bladder
/ diagnostic imaging
Journal
International journal of radiation oncology, biology, physics
ISSN: 1879-355X
Titre abrégé: Int J Radiat Oncol Biol Phys
Pays: United States
ID NLM: 7603616
Informations de publication
Date de publication:
01 12 2019
01 12 2019
Historique:
received:
14
02
2019
revised:
16
08
2019
accepted:
22
08
2019
pubmed:
11
9
2019
medline:
6
2
2020
entrez:
11
9
2019
Statut:
ppublish
Résumé
Deep learning methods (DLMs) have recently been proposed to generate pseudo-CT (pCT) for magnetic resonance imaging (MRI) based dose planning. This study aims to evaluate and compare DLMs (U-Net and generative adversarial network [GAN]) using various loss functions (L2, single-scale perceptual loss [PL], multiscale PL, weighted multiscale PL) and a patch-based method (PBM). Thirty-nine patients received a volumetric modulated arc therapy for prostate cancer (78 Gy). T Considering the image uncertainties in the whole pelvis, GAN L2 and U-Net L2 showed the lowest mean absolute error (≤34.4 Hounsfield units). The mean errors were not different than 0 (P ≤ .05). The PBM provided the highest uncertainties. Very few DVH points differed when comparing GAN L2 or U-Net L2 DVHs and CT Generating pCT for MRI dose planning with DLMs and PBM provided low-dose uncertainties. In particular, the GAN L2 and U-Net L2 provided the lowest dose uncertainties together with a low computation time.
Identifiants
pubmed: 31505245
pii: S0360-3016(19)33735-6
doi: 10.1016/j.ijrobp.2019.08.049
pii:
doi:
Types de publication
Comparative Study
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
1137-1150Informations de copyright
Crown Copyright © 2019. Published by Elsevier Inc. All rights reserved.