Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Volumetric modulated arc therapy dose prediction and deliverable treatment plan generation for prostate cancer patients using a densely connected deep learning model

Lempart, Michael LU ; Benedek, Hunor LU ; Nilsson, Mikael LU ; Eliasson, Niklas ; Bäck, Sven LU ; Munck af Rosenschöld, Per LU orcid ; Olsson, Lars E. LU orcid and Jamtheim Gustafsson, Christian LU (2021) In Physics and imaging in radiation oncology 19. p.112-119
Abstract

Background and purpose: Radiation therapy treatment planning is a manual, time-consuming task that might be accelerated using machine learning algorithms. In this study, we aimed to evaluate if a triplet-based deep learning model can predict volumetric modulated arc therapy (VMAT) dose distributions for prostate cancer patients. Materials and methods: A modified U-Net was trained on triplets, a combination of three consecutive image slices and corresponding segmentations, from 160 patients, and compared to a baseline U-Net. Dose predictions from 17 test patients were transformed into deliverable treatment plans using a novel planning workflow. Results: The model achieved a mean absolute dose error of 1.3%, 1.9%, 1.0% and ≤ 2.6% for... (More)

Background and purpose: Radiation therapy treatment planning is a manual, time-consuming task that might be accelerated using machine learning algorithms. In this study, we aimed to evaluate if a triplet-based deep learning model can predict volumetric modulated arc therapy (VMAT) dose distributions for prostate cancer patients. Materials and methods: A modified U-Net was trained on triplets, a combination of three consecutive image slices and corresponding segmentations, from 160 patients, and compared to a baseline U-Net. Dose predictions from 17 test patients were transformed into deliverable treatment plans using a novel planning workflow. Results: The model achieved a mean absolute dose error of 1.3%, 1.9%, 1.0% and ≤ 2.6% for clinical target volume (CTV) CTV_D100%, planning target volume (PTV) PTV_D98%, PTV_D95% and organs at risk (OAR) respectively, when compared to the clinical ground truth (GT) dose distributions. All predicted distributions were successfully transformed into deliverable treatment plans and tested on a phantom, resulting in a passing rate of 100% (global gamma, 3%, 2 mm, 15% cutoff). The dose difference between deliverable treatment plans and GT dose distributions was within 4.4%. The difference between the baseline model and our improved model was statistically significant (p < 0.05) for CVT_D100%, PTV_D98% and PTV_D95%. Conclusion: Triplet-based training improved VMAT dose distribution predictions when compared to 2D. Dose predictions were successfully transformed into deliverable treatment plans using our proposed treatment planning procedure. Our method may automate parts of the workflow for external beam prostate radiation therapy and improve the overall treatment speed and plan quality.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Deep learning, Deliverable treatment plans, Dose prediction, Machine learning, Radiotherapy, Volumetric modulated arc therapy
in
Physics and imaging in radiation oncology
volume
19
pages
8 pages
publisher
Elsevier
external identifiers
  • scopus:85113150519
  • pmid:34401537
ISSN
2405-6316
DOI
10.1016/j.phro.2021.07.008
language
English
LU publication?
yes
id
e20e3c93-c26a-440c-ab02-35d429eb83af
date added to LUP
2021-09-06 16:04:56
date last changed
2024-04-20 10:45:02
@article{e20e3c93-c26a-440c-ab02-35d429eb83af,
  abstract     = {{<p>Background and purpose: Radiation therapy treatment planning is a manual, time-consuming task that might be accelerated using machine learning algorithms. In this study, we aimed to evaluate if a triplet-based deep learning model can predict volumetric modulated arc therapy (VMAT) dose distributions for prostate cancer patients. Materials and methods: A modified U-Net was trained on triplets, a combination of three consecutive image slices and corresponding segmentations, from 160 patients, and compared to a baseline U-Net. Dose predictions from 17 test patients were transformed into deliverable treatment plans using a novel planning workflow. Results: The model achieved a mean absolute dose error of 1.3%, 1.9%, 1.0% and ≤ 2.6% for clinical target volume (CTV) CTV_D<sub>100%</sub>, planning target volume (PTV) PTV_D<sub>98%</sub>, PTV<sub>_</sub>D<sub>95%</sub> and organs at risk (OAR) respectively, when compared to the clinical ground truth (GT) dose distributions. All predicted distributions were successfully transformed into deliverable treatment plans and tested on a phantom, resulting in a passing rate of 100% (global gamma, 3%, 2 mm, 15% cutoff). The dose difference between deliverable treatment plans and GT dose distributions was within 4.4%. The difference between the baseline model and our improved model was statistically significant (p &lt; 0.05) for CVT_D<sub>100%</sub>, PTV_D<sub>98%</sub> and PTV_D<sub>95%</sub>. Conclusion: Triplet-based training improved VMAT dose distribution predictions when compared to 2D. Dose predictions were successfully transformed into deliverable treatment plans using our proposed treatment planning procedure. Our method may automate parts of the workflow for external beam prostate radiation therapy and improve the overall treatment speed and plan quality.</p>}},
  author       = {{Lempart, Michael and Benedek, Hunor and Nilsson, Mikael and Eliasson, Niklas and Bäck, Sven and Munck af Rosenschöld, Per and Olsson, Lars E. and Jamtheim Gustafsson, Christian}},
  issn         = {{2405-6316}},
  keywords     = {{Deep learning; Deliverable treatment plans; Dose prediction; Machine learning; Radiotherapy; Volumetric modulated arc therapy}},
  language     = {{eng}},
  pages        = {{112--119}},
  publisher    = {{Elsevier}},
  series       = {{Physics and imaging in radiation oncology}},
  title        = {{Volumetric modulated arc therapy dose prediction and deliverable treatment plan generation for prostate cancer patients using a densely connected deep learning model}},
  url          = {{http://dx.doi.org/10.1016/j.phro.2021.07.008}},
  doi          = {{10.1016/j.phro.2021.07.008}},
  volume       = {{19}},
  year         = {{2021}},
}