Improved robotic stereotactic body radiation therapy plan quality and planning efficacy for organ-confined prostate cancer utilizing overlap-volume histogram-driven planning methodology

Wu, Binbin; Pang, Dalong; Lei, Siyuan; Gatti, John W.; Tong, Michael C. F.; McNutt, Todd; Kole, Thomas P.; Dritschilo, Anatoly; Collins, Sean P.

doi:10.1016/j.radonc.2014.07.009

Cited by 50 publications

(48 citation statements)

References 20 publications

(34 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With a 3D dose distribution, one can fully reconstruct the DVH, and with the DVH, the dose constraints can then be calculated. Many studies focused on either predicting dose constraints or the DVH, eventually forming the backbone of knowledge‐based planning (KBP) . Knowledge‐based planning used machine learning techniques and models to predict clinically acceptable dosimetric criteria, utilizing a large pool of historical patient plans and information to draw its knowledge from.…”

Section: Introductionmentioning

confidence: 99%

“…Many studies focused on either predicting dose constraints or the DVH, eventually forming the backbone of knowledge-based planning (KBP). [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41][42] Knowledge-based planning used machine learning techniques and models to predict clinically acceptable dosimetric criteria, utilizing a large pool of historical patient plans and information to draw its knowledge from. Before the era of deep neural networks, KBP's efficacy was heavily reliant on not only the patient data size and diversity, but also on the careful selection of features extracted from the data to be used in the model.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Incorporating human and learned domain knowledge into training deep neural networks: A differentiable dose‐volume histogram and adversarial inspired framework for generating Pareto optimal dose distributions in radiation therapy

et al. 2019

View full text Add to dashboard Cite

Purpose We propose a novel domain‐specific loss, which is a differentiable loss function based on the dose‐volume histogram (DVH), and combine it with an adversarial loss for the training of deep neural networks. In this study, we trained a neural network for generating Pareto optimal dose distributions, and evaluate the effects of the domain‐specific loss on the model performance. Methods In this study, three loss functions — mean squared error (MSE) loss, DVH loss, and adversarial (ADV) loss — were used to train and compare four instances of the neural network model: (a) MSE, (b) MSE + ADV, (c) MSE + DVH, and (d) MSE + DVH+ADV. The data for 70 prostate patients, including the planning target volume (PTV), and the organs at risk (OAR) were acquired as 96 × 96 × 24 dimension arrays at 5 mm3 voxel size. The dose influence arrays were calculated for 70 prostate patients, using a 7 equidistant coplanar beam setup. Using a scalarized multicriteria optimization for intensity‐modulated radiation therapy, 1200 Pareto surface plans per patient were generated by pseudo‐randomizing the PTV and OAR tradeoff weights. With 70 patients, the total number of plans generated was 84 000 plans. We divided the data into 54 training, 6 validation, and 10 testing patients. Each model was trained for a total of 100,000 iterations, with a batch size of 2. All models used the Adam optimizer, with a learning rate of 1 × 10−3. Results Training for 100 000 iterations took 1.5 days (MSE), 3.5 days (MSE+ADV), 2.3 days (MSE+DVH), and 3.8 days (MSE+DVH+ADV). After training, the prediction time of each model is 0.052 s. Quantitatively, the MSE+DVH+ADV model had the lowest prediction error of 0.038 (conformation), 0.026 (homogeneity), 0.298 (R50), 1.65% (D95), 2.14% (D98), and 2.43% (D99). The MSE model had the worst prediction error of 0.134 (conformation), 0.041 (homogeneity), 0.520 (R50), 3.91% (D95), 4.33% (D98), and 4.60% (D99). For both the mean dose PTV error and the max dose PTV, Body, Bladder and rectum error, the MSE+DVH+ADV outperformed all other models. Regardless of model, all predictions have an average mean and max dose error <2.8% and 4.2%, respectively. Conclusion The MSE+DVH+ADV model performed the best in these categories, illustrating the importance of both human and learned domain knowledge. Expert human domain‐specific knowledge can be the largest driver in the performance improvement, and adversarial learning can be used to further capture nuanced attributes in the data. The real‐time prediction capabilities allow for a physician to quickly navigate the tradeoff space for a patient, and produce a dose distribution as a tangible endpoint for the dosimetrist to use for planning. This is expected to considerably reduce the treatment planning time, allowing for clinicians to focus their efforts on the difficult and demanding cases.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Incorporating human and learned domain knowledge into training deep neural networks: A differentiable dose‐volume histogram and adversarial inspired framework for generating Pareto optimal dose distributions in radiation therapy

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Therefore, achieved OAR doses can be highly institution‐, planner‐ and patient dependent. Proposed solutions are knowledge‐based (KB) treatment plan QA or KB (semi‐)automated treatment planning . With KB treatment plan QA, treatment plans of prior patients are used to predict the achievable doses for a new patient.…”

Section: Introductionmentioning

confidence: 99%

“…Proposed solutions are knowledge-based (KB) treatment plan QA [1][2][3][4][5][6][7] or KB (semi-)automated treatment planning. [8][9][10][11][12][13] With KB treatment plan QA, treatment plans of prior patients are used to predict the achievable doses for a new patient. The predictions are then compared to the achieved treatment plans, to detect suboptimal sparing of OARs.…”

Section: Introductionmentioning

confidence: 99%

Knowledge‐based dose prediction models for head and neck cancer are strongly affected by interorgan dependency and dataset inconsistency

2018

View full text Add to dashboard Cite

Purpose The goal of this study was to generate a large treatment plan database for head and neck (H&N) cancer patients that can be considered as the gold standard to train and validate models for knowledge‐based (KB) treatment planning and QA. With this dataset, the intrinsic prediction performance, the effect of interorgan dependency, and the impact of dataset inconsistency was investigated for an existing treatment planning QA model. Methods The CT scans of 108 previously treated oropharyngeal patients were used to establish the plan database. For each patient, 15 Pareto optimal treatment plans with different planning priorities for the parotid glands were generated with fully automatic multicriterial treatment planning (1620 plans in total). For each of the 15 sets of plans in the database, a KB model was trained with 54 patients and validated on the other 54 by comparing the predictions with the achieved doses. The dose prediction accuracy (predicted—achieved) of the KB models was assessed and compared among the different models to characterize the intrinsic performance and effect of interorgan dependency. In addition, the effect of dataset inconsistency with respect to planning prioritizations was investigated by mixing plans with different prioritizations, for the training, the validation dataset, and for both combined. Results In the case of a high planning priority, the mean ± SD of the prediction error for the mean dose of the parotid glands was only 0.2 ± 2.2 Gy, but this increased to 1.0 ± 5.0 Gy in the case that the parotid glands had a low planning priority. Dataset inconsistency (in planning priority) led to a large increase in prediction error for the parotid glands (mean ± SD) from 0.2 ± 2.2 Gy to 2.8 ± 3.3 Gy, −3.2 ± 5.0 Gy or −0.6 ± 5.4 Gy, depending on the way the datasets were mixed. Conclusions The generated plan database can be used to validate and characterize KB prediction models for H&N cancer and will be made available upon request. The investigated KB model performed well in case the parotid glands had a high planning priority (little dependence on lower priority OARs), but poorly for organs for which the dose strongly depends on other higher priority OARs. To improve the performance of KB prediction models for H&N cancer, interorgan dependency should be modeled and accounted for. Dataset inconsistency has a large negative impact on the prediction errors of KB models and should be avoided as much as possible.

show abstract

“…Available normal tissue dose constraints have largely been extrapolated from HDR brachytherapy series and are based on biologically equivalent dose comparisons to conventionally fractionated limits [16, 17]. Recent attempts have also been made to determine objective prostate SBRT plan evaluation criteria based upon databases of achievable dose distributions in relation to anatomic geometry [18, 19], however, no specific toxicity related dose-volume criteria were determined. In a limited series of patients treated at a single institution, Elias et al recently reported correlates of QOL in patients treated with prostate SBRT, however, no associated dosimetric predictors of long-term urinary toxicity were found[20].…”

Section: Introductionmentioning

confidence: 99%

Late urinary toxicity modeling after stereotactic body radiotherapy (SBRT) in the definitive treatment of localized prostate cancer

Kole

Tong

et al. 2015

Acta Oncologica

Self Cite

View full text Add to dashboard Cite

Background Late urinary symptom flare has been shown to occur in a small subset of men treated with ultra-hypofractionated SBRT for prostate cancer. The purpose of this study was to use normal tissue complication probability modeling in an effort to derive SBRT specific dosimetric predictor’s of late urinary flare. Methods and Materials Two hundred and sixteen men were treated for localized prostate cancer using ultra-hypofractionated SBRT. A dose of 35 to 36.25 Gy in 5 fractions was delivered to the prostate and proximal seminal vesicles. Functional surveys were conducted before and after treatment to assess late toxicity. Phenomenologic NTCP models were fit to bladder DVHs and late urinary flare outcomes using maximum likelihood estimation. Results Twenty nine patients experienced late urinary flare within 2 years of completion of treatment. Fitting of bladder DVH data to a Lyman NTCP model resulted in parameter estimates of m, TD50, and n of 0.19 (0 to 0.47), 38.7Gy (31.1 to 46.4), and 0.13 (−0.14 to 0.41), respectively. Subsequent fit to a hottest volume probit model revealed a significant association of late urinary flare with dose to the hottest 12.7% of bladder volume. Multivariate analysis resulted in a final model that included patient age and hottest volume probit model predictions. Kaplan-Meier analysis demonstrated a 2 year urinary flare free survival of 95.7% in patients 65 years or older with a bladder D12.7% of 33.5 Gy or less, compared to 74.5% in patients meeting none of these criteria. Conclusions NTCP modeling of late urinary flare after ultra-hypofractionated prostate SBRT demonstrates a relatively small volume effect for dose to the bladder, suggesting that reduction of volume receiving elevated dose will result in decreased incidence of late urinary toxicity. Future studies will be needed to examine the impact of dose to other potential sources of late genitourinary toxicity.

show abstract

Improved robotic stereotactic body radiation therapy plan quality and planning efficacy for organ-confined prostate cancer utilizing overlap-volume histogram-driven planning methodology

Cited by 50 publications

References 20 publications

Incorporating human and learned domain knowledge into training deep neural networks: A differentiable dose‐volume histogram and adversarial inspired framework for generating Pareto optimal dose distributions in radiation therapy

Incorporating human and learned domain knowledge into training deep neural networks: A differentiable dose‐volume histogram and adversarial inspired framework for generating Pareto optimal dose distributions in radiation therapy

Knowledge‐based dose prediction models for head and neck cancer are strongly affected by interorgan dependency and dataset inconsistency

Late urinary toxicity modeling after stereotactic body radiotherapy (SBRT) in the definitive treatment of localized prostate cancer

Contact Info

Product

Resources

About