Journal
BRACHYTHERAPY
Volume 21, Issue 4, Pages 532-542Publisher
ELSEVIER SCIENCE INC
DOI: 10.1016/j.brachy.2022.03.002
Keywords
Deep learning; Convolutional neural network; Tandem and ovoids; Cervical cancer; Knowledge -based plan
Funding
- Pedal the Cause and the Agency for Healthcare Research and Quality [AHRQ R01HS025440]
Ask authors/readers for more resources
The purpose of this study was to develop a knowledge-based dose prediction system using a convolution neural network (CNN) for cervical brachytherapy treatments. The results showed that this knowledge-based dose prediction system can be used for treatment plan quality control and data-driven plan guidance.
PURPOSE: The purpose of this work was to develop a knowledge-based dose prediction system using a convolution neural network (CNN) for cervical brachytherapy treatments with a tandem -and-ovoid applicator. METHODS: A 3D U-NET CNN was utilized to make voxel-wise dose predictions based on organ-at-risk (OAR), high-risk clinical target volume (HRCTV), and possible source location geometry. The model comprised 395 previously treated cases: training (273), validation (61), test (61). To assess voxel prediction accuracy, we evaluated dose differences in all cohorts across the dose range of 20-130% of prescription, mean (SD) and standard deviation (sigma ), as well as iso-dose dice similarity coefficients for clinical and/or predicted dose distributions. We examined discrete Dose-Volume Histogram (DVH) metrics utilized for brachytherapy plan quality assessment (HRCTV D90%; bladder, rectum, and sigmoid D2cc) with ADx = D-x,D-actual - D(x,predicted )mean, standard deviation, and Pearson correlation coefficient further quantifying model performance. RESULTS: Ranges of voxel-wise dose difference accuracy (delta D +/- sigma ) for 20-130% dose interval in training (test) sets ranged from [-0.5% +/- 2.0% to + 2.0% +/- 14.0%] ([-0.1% +/- 4.0% to + 4.0% +/- 26.0%]) in all voxels, [-1.7% +/- 5.1% to-3.5% +/- 12.8%] ([-2.9% +/- 4.8% to-2.6% +/- 18.9%]) in HRCTV, [-0.02% +/- 2.40% to + 3.2% +/- 12.0%] ([-2.5% +/- 3.6% to + 0.8% +/- 12.7%]) in bladder, [-0.7% +/- 2.4% to + 15.5% +/- 11.0%] ([-0.9% +/- 3.2% to + 27.8% +/- 11.6%]) in rectum, and [-0.7% +/- 2.3% to + 10.7% +/- 15.0%] ([-0.4% +/- 3.0% to + 18.4% +/- 11.4%]) in sigmoid. Isodose dice similarity coefficients ranged from [0.96,0.91] for training and [0.94,0.87] for test cohorts. Relative DVH metric prediction in the training (test) set were HRCTV delta D-90 +/- sigma delta D =-0.19 +/- 0.55Gy (-0.09 +/- 0.67 Gy), bladder AD2(cc) +/- sigma delta D =-0.06 +/- 0.54Gy (-0.17 +/- 0.67 Gy), rectum delta D2(cc) +/- sigma AD =-0.03 +/- 0.36Gy (-0.04 +/- 0.46 Gy), and sigmoid delta D2(cc) +/- sigma AD =-0.01 +/- 0.34Gy (0.00 +/- 0.44 Gy). CONCLUSIONS: A 3D knowledge-based dose predictions provide voxel-level and DVH metric estimates that could be used for treatment plan quality control and data-driven plan guidance. (C) 2022 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available