Investigating the use of machine learning to generate ventilation images from CT scans.
Galligas PET
functional lung imaging
machine learning
ventilation
Journal
Medical physics
ISSN: 2473-4209
Titre abrégé: Med Phys
Pays: United States
ID NLM: 0425746
Informations de publication
Date de publication:
Aug 2022
Aug 2022
Historique:
revised:
15
03
2022
received:
05
11
2021
accepted:
25
04
2022
pubmed:
4
5
2022
medline:
19
8
2022
entrez:
3
5
2022
Statut:
ppublish
Résumé
Radiotherapy treatment planning incorporating ventilation imaging can reduce the incidence of radiation-induced lung injury. The gold-standard of ventilation imaging, using nuclear medicine, has limitations with respect to availability and cost. An alternative type of ventilation imaging to nuclear medicine uses 4DCT (or breath-hold CT [BHCT] pair) with deformable image registration (DIR) and a ventilation metric to produce a CT ventilation image (CTVI). The purpose of this study is to investigate the application of machine learning as an alternative to DIR-based methods when producing CTVIs. A patient dataset of 15 inhale and exhale BHCTs and Galligas PET ventilation images were used to train and test a 2D U-Net style convolutional neural network. The neural network established relationships between axial input BHCT image pairs and axial labeled Galligas PET images and was evaluated using eightfold cross-validation. Once trained, the neural network could produce a CTVI from an input BHCT image pair. The CTVIs produced by the neural network were qualitatively assessed visually and quantitatively compared to a Galligas PET ventilation image using a Spearman correlation and Dice similarity coefficient (DSC). The DSC measured the spatial overlap between three segmented equal lung volumes by ventilation (high, medium, and low functioning lung [LFL]). The mean Spearman correlation between the CTVIs and the Galligas PET ventilation images was 0.58 ± 0.14. The mean DSC over high, medium, and LFL between the CTVIs and Galligas PET ventilation images was 0.55 ± 0.06. Visually, a systematic overprediction of ventilation within the lung was observed in the CTVIs with respect to the Galligas PET ventilation images, with jagged regions of ventilation in the sagittal and coronal planes. A convolutional neural network was developed that could produce a CTVI from a BHCT image pair, which was then compared with a Galligas PET ventilation image. The performance of this machine learning method was comparable to previous benchmark studies investigating a DIR-based CTVI, warranting future development, and investigation of applying machine learning to a CTVI.
Sections du résumé
BACKGROUND
BACKGROUND
Radiotherapy treatment planning incorporating ventilation imaging can reduce the incidence of radiation-induced lung injury. The gold-standard of ventilation imaging, using nuclear medicine, has limitations with respect to availability and cost.
PURPOSE
OBJECTIVE
An alternative type of ventilation imaging to nuclear medicine uses 4DCT (or breath-hold CT [BHCT] pair) with deformable image registration (DIR) and a ventilation metric to produce a CT ventilation image (CTVI). The purpose of this study is to investigate the application of machine learning as an alternative to DIR-based methods when producing CTVIs.
METHODS
METHODS
A patient dataset of 15 inhale and exhale BHCTs and Galligas PET ventilation images were used to train and test a 2D U-Net style convolutional neural network. The neural network established relationships between axial input BHCT image pairs and axial labeled Galligas PET images and was evaluated using eightfold cross-validation. Once trained, the neural network could produce a CTVI from an input BHCT image pair. The CTVIs produced by the neural network were qualitatively assessed visually and quantitatively compared to a Galligas PET ventilation image using a Spearman correlation and Dice similarity coefficient (DSC). The DSC measured the spatial overlap between three segmented equal lung volumes by ventilation (high, medium, and low functioning lung [LFL]).
RESULTS
RESULTS
The mean Spearman correlation between the CTVIs and the Galligas PET ventilation images was 0.58 ± 0.14. The mean DSC over high, medium, and LFL between the CTVIs and Galligas PET ventilation images was 0.55 ± 0.06. Visually, a systematic overprediction of ventilation within the lung was observed in the CTVIs with respect to the Galligas PET ventilation images, with jagged regions of ventilation in the sagittal and coronal planes.
CONCLUSIONS
CONCLUSIONS
A convolutional neural network was developed that could produce a CTVI from a BHCT image pair, which was then compared with a Galligas PET ventilation image. The performance of this machine learning method was comparable to previous benchmark studies investigating a DIR-based CTVI, warranting future development, and investigation of applying machine learning to a CTVI.
Identifiants
pubmed: 35502763
doi: 10.1002/mp.15688
pmc: PMC9545612
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
5258-5267Subventions
Organisme : Cancer Institute NSW Translational Program Grant
Organisme : Australian Government (NHMRC) Investigator Grant
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. 2022 Aug;49(8):5258-5267
pubmed: 35502763
Med Phys. 2014 Jan;41(1):011910
pubmed: 24387518
Semin Nucl Med. 2016 Sep;46(5):428-35
pubmed: 27553468
Clin Oncol (R Coll Radiol). 2016 Nov;28(11):695-707
pubmed: 27637724
Med Phys. 2016 Jan;43(1):33
pubmed: 26745897
Med Phys. 2019 Mar;46(3):1198-1217
pubmed: 30575051
Br J Radiol. 2022 Apr 1;95(1132):20201107
pubmed: 33877878
J Med Imaging Radiat Oncol. 2014 Apr;58(2):214-22
pubmed: 24373453
Int J Radiat Oncol Biol Phys. 2007 May 1;68(1):273-81
pubmed: 17448880
Phys Med Biol. 2021 Aug 23;66(17):
pubmed: 34293726
Med Dosim. 2010 Winter;35(4):297-303
pubmed: 19944585
J Digit Imaging. 2018 Jun;31(3):290-303
pubmed: 29181613
Front Oncol. 2019 Sep 06;9:877
pubmed: 31555602
Med Phys. 2020 Mar;47(3):1249-1257
pubmed: 31883382
Radiother Oncol. 2018 Nov;129(2):196-208
pubmed: 30082143
BMJ Open. 2020 Dec 10;10(12):e042465
pubmed: 33303468
Med Phys. 2010 Apr;37(4):1863-72
pubmed: 20443508
Int J Radiat Oncol Biol Phys. 2018 Nov 15;102(4):1276-1286
pubmed: 30355463
Radiother Oncol. 2019 Aug;137:175-185
pubmed: 31129503
Int J Radiat Oncol Biol Phys. 2017 Sep 1;99(1):202-209
pubmed: 28816147
Int J Radiat Oncol Biol Phys. 2013 Feb 1;85(2):444-50
pubmed: 22682812
Radiother Oncol. 2018 May;127(2):267-273
pubmed: 29290405
PeerJ. 2014 Jun 19;2:e453
pubmed: 25024921
Acad Radiol. 2019 Jan;26(1):38-49
pubmed: 29606339
Med Phys. 2019 May;46(5):2323-2329
pubmed: 30714159
Acad Radiol. 2012 Dec;19(12):1546-53
pubmed: 22999648