Deep learning-based 4D-synthetic CTs from sparse-view CBCTs for dose calculations in adaptive proton therapy.
4D imaging
adaptive proton therapy
deep learning
synthetic CT
Journal
Medical physics
ISSN: 2473-4209
Titre abrégé: Med Phys
Pays: United States
ID NLM: 0425746
Informations de publication
Date de publication:
Nov 2022
Nov 2022
Historique:
revised:
20
07
2022
received:
02
05
2022
accepted:
08
08
2022
pubmed:
20
8
2022
medline:
15
12
2022
entrez:
19
8
2022
Statut:
ppublish
Résumé
Time-resolved 4D cone beam-computed tomography (4D-CBCT) allows a daily assessment of patient anatomy and respiratory motion. However, 4D-CBCTs suffer from imaging artifacts that affect the CT number accuracy and prevent accurate proton dose calculations. Deep learning can be used to correct CT numbers and generate synthetic CTs (sCTs) that can enable CBCT-based proton dose calculations. In this work, sparse view 4D-CBCTs were converted into 4D-sCT utilizing a deep convolutional neural network (DCNN). 4D-sCTs were evaluated in terms of image quality and dosimetric accuracy to determine if accurate proton dose calculations for adaptive proton therapy workflows of lung cancer patients are feasible. A dataset of 45 thoracic cancer patients was utilized to train and evaluate a DCNN to generate 4D-sCTs, based on sparse view 4D-CBCTs reconstructed from projections acquired with a 3D acquisition protocol. Mean absolute error (MAE) and mean error were used as metrics to evaluate the image quality of single phases and average 4D-sCTs against 4D-CTs acquired on the same day. The dosimetric accuracy was checked globally (gamma analysis) and locally for target volumes and organs-at-risk (OARs) (lung, heart, and esophagus). Furthermore, 4D-sCTs were also compared to 3D-sCTs. To evaluate CT number accuracy, proton radiography simulations in 4D-sCT and 4D-CTs were compared in terms of range errors. The clinical suitability of 4D-sCTs was demonstrated by performing a 4D dose reconstruction using patient specific treatment delivery log files and breathing signals. 4D-sCTs resulted in average MAEs of 48.1 ± 6.5 HU (single phase) and 37.7 ± 6.2 HU (average). The global dosimetric evaluation showed gamma pass ratios of 92.3% ± 3.2% (single phase) and 94.4% ± 2.1% (average). The clinical target volume showed high agreement in D In this study, we have investigated the accuracy of deep learning-based 4D-sCTs for daily dose calculations in adaptive proton therapy. Despite image quality differences between 4D-sCTs and 3D-sCTs, comparable dosimetric accuracy was observed globally and locally. Further improvement of 3D and 4D lung sCTs could be achieved by increasing CT number accuracy in lung tissues.
Sections du résumé
BACKGROUND
BACKGROUND
Time-resolved 4D cone beam-computed tomography (4D-CBCT) allows a daily assessment of patient anatomy and respiratory motion. However, 4D-CBCTs suffer from imaging artifacts that affect the CT number accuracy and prevent accurate proton dose calculations. Deep learning can be used to correct CT numbers and generate synthetic CTs (sCTs) that can enable CBCT-based proton dose calculations.
PURPOSE
OBJECTIVE
In this work, sparse view 4D-CBCTs were converted into 4D-sCT utilizing a deep convolutional neural network (DCNN). 4D-sCTs were evaluated in terms of image quality and dosimetric accuracy to determine if accurate proton dose calculations for adaptive proton therapy workflows of lung cancer patients are feasible.
METHODS
METHODS
A dataset of 45 thoracic cancer patients was utilized to train and evaluate a DCNN to generate 4D-sCTs, based on sparse view 4D-CBCTs reconstructed from projections acquired with a 3D acquisition protocol. Mean absolute error (MAE) and mean error were used as metrics to evaluate the image quality of single phases and average 4D-sCTs against 4D-CTs acquired on the same day. The dosimetric accuracy was checked globally (gamma analysis) and locally for target volumes and organs-at-risk (OARs) (lung, heart, and esophagus). Furthermore, 4D-sCTs were also compared to 3D-sCTs. To evaluate CT number accuracy, proton radiography simulations in 4D-sCT and 4D-CTs were compared in terms of range errors. The clinical suitability of 4D-sCTs was demonstrated by performing a 4D dose reconstruction using patient specific treatment delivery log files and breathing signals.
RESULTS
RESULTS
4D-sCTs resulted in average MAEs of 48.1 ± 6.5 HU (single phase) and 37.7 ± 6.2 HU (average). The global dosimetric evaluation showed gamma pass ratios of 92.3% ± 3.2% (single phase) and 94.4% ± 2.1% (average). The clinical target volume showed high agreement in D
CONCLUSION
CONCLUSIONS
In this study, we have investigated the accuracy of deep learning-based 4D-sCTs for daily dose calculations in adaptive proton therapy. Despite image quality differences between 4D-sCTs and 3D-sCTs, comparable dosimetric accuracy was observed globally and locally. Further improvement of 3D and 4D lung sCTs could be achieved by increasing CT number accuracy in lung tissues.
Identifiants
pubmed: 35982630
doi: 10.1002/mp.15930
pmc: PMC10087352
doi:
Substances chimiques
Protons
0
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
6824-6839Subventions
Organisme : Dutch Cancer Society (KWF research project 11518)
Informations de copyright
© 2022 The Authors. Medical Physics published by Wiley Periodicals LLC on behalf of American Association of Physicists in Medicine.
Références
Med Phys. 2020 Nov;47(11):5619-5631
pubmed: 33063329
Med Phys. 2019 Sep;46(9):3799-3811
pubmed: 31247134
Med Phys. 2016 Dec;43(12):6405
pubmed: 27908151
Phys Med Biol. 2016 Sep 21;61(18):6856-6877
pubmed: 27588815
Phys Med Biol. 2019 Jan 24;64(3):035011
pubmed: 30523998
Phys Med Biol. 2019 Nov 15;64(22):225004
pubmed: 31610527
Cureus. 2018 Apr 29;10(4):e2548
pubmed: 29963342
Phys Med Biol. 2016 Jun 7;61(11):4078-87
pubmed: 27164479
Int J Radiat Oncol Biol Phys. 2017 Nov 15;99(4):994-1003
pubmed: 28916139
Radiother Oncol. 2020 Dec;153:197-204
pubmed: 32976877
Radiother Oncol. 2020 Jun;147:178-185
pubmed: 32380117
Radiother Oncol. 2018 Nov;129(2):249-256
pubmed: 30241789
Med Phys. 2016 Oct;43(10):5635
pubmed: 27782706
Semin Radiat Oncol. 2019 Jul;29(3):245-257
pubmed: 31027642
Phys Imaging Radiat Oncol. 2018 Mar 08;5:69-75
pubmed: 33458372
Med Phys. 2020 Dec;47(12):6381-6387
pubmed: 33011990
Int J Radiat Oncol Biol Phys. 2019 Nov 1;105(3):495-503
pubmed: 31271823
Radiother Oncol. 2020 Nov;152:117-125
pubmed: 31547943
Z Med Phys. 2019 Aug;29(3):249-261
pubmed: 30448049
Med Phys. 2021 Aug;48(8):4498-4505
pubmed: 34077554
Med Phys. 2015 Mar;42(3):1354-66
pubmed: 25735290
Phys Imaging Radiat Oncol. 2020 May 25;14:24-31
pubmed: 33458310
Med Phys. 2019 Sep;46(9):3998-4009
pubmed: 31206709
Med Phys. 2021 Dec;48(12):7673-7684
pubmed: 34725829
Phys Med Biol. 2020 Apr 28;65(9):095002
pubmed: 32143207
Med Phys. 2018 Nov;45(11):e1086-e1095
pubmed: 30421805
Med Phys. 2021 Mar;48(3):1372-1380
pubmed: 33428795
Med Phys. 2021 Nov;48(11):6537-6566
pubmed: 34407209
Br J Radiol. 2020 Mar;93(1107):20190594
pubmed: 31647313
Med Phys. 2015 Aug;42(8):4449-59
pubmed: 26233175
Phys Med Biol. 2020 Nov 20;65(23):
pubmed: 33049722
Phys Med Biol. 2020 Dec 05;65(23):235036
pubmed: 33179874
Dentomaxillofac Radiol. 2011 Jul;40(5):265-73
pubmed: 21697151
Z Med Phys. 2022 Feb;32(1):74-84
pubmed: 33248812
Phys Med. 2020 Aug;76:243-276
pubmed: 32736286
Phys Med Biol. 2021 May 10;66(10):
pubmed: 33862616
Phys Imaging Radiat Oncol. 2020 Oct 27;16:89-94
pubmed: 33458349
Int J Radiat Oncol Biol Phys. 2016 May 1;95(1):549-559
pubmed: 27084664
Acta Oncol. 2013 Oct;52(7):1477-83
pubmed: 23879648
Med Phys. 2022 Nov;49(11):6824-6839
pubmed: 35982630
Phys Med Biol. 2015 Apr 21;60(8):R155-209
pubmed: 25803097
Phys Med Biol. 2019 Jun 10;64(12):125002
pubmed: 31108465
Acta Oncol. 2014 May;53(5):605-12
pubmed: 23957623
Med Phys. 2019 Mar;46(3):1140-1149
pubmed: 30609061
Phys Med Biol. 2021 Nov 15;66(22):
pubmed: 34710858
Phys Med. 2017 Jan;33:68-76
pubmed: 27998666
Phys Med Biol. 2021 Aug 30;66(17):
pubmed: 34293737