Neural network dose prediction for cervical brachytherapy: Overcoming data scarcity for applicator-specific models.

BACKGROUND: 3D neural network dose predictions are useful for automating brachytherapy (BT) treatment planning for cervical cancer. Cervical BT can be delivered with numerous applicators, which necessitates developing models that generalize to multiple applicator types. The variability and scarcity of data for any given applicator type poses challenges for deep learning. PURPOSE: The goal of this work was to compare three methods of neural network training-a single model trained on all applicator data, fine-tuning the combined model to each applicator, and individual (IDV) applicator models-to determine the optimal method for dose prediction. METHODS: Models were produced for four applicator types-tandem-and-ovoid (T&O), T&O with 1-7 needles (T&ON), tandem-and-ring (T&R) and T&R with 1-4 needles (T&RN). First, the combined model was trained on 859 treatment plans from 266 cervical cancer patients treated from 2010 onwards. The train/validation/test split was 70%/16%/14%, with approximately 49%/10%/19%/22% T&O/T&ON/T&R/T&RN in each dataset. Inputs included four channels for anatomical masks (high-risk clinical target volume [HRCTV], bladder, rectum, and sigmoid), a mask indicating dwell position locations, and applicator channels for each applicator component. Applicator channels were created by mapping the 3D dose for a single dwell position to each dwell position and summing over each applicator component with uniform dwell time weighting. A 3D Cascade U-Net, which consists of two U-Nets in sequence, and mean squared error loss function were used. The combined model was then fine-tuned to produce four applicator-specific models by freezing the first U-Net and encoding layers of the second and resuming training on applicator-specific data. Finally, four IDV models were trained using only data from each applicator type. Performance of these three model types was compared using the following metrics for the test set: mean error (ME, representing model bias) and mean absolute error (MAE) over all dose voxels and ME of clinical metrics (HRCTV D90% and D2cc of bladder, rectum, and sigmoid), averaged over all patients. A positive ME indicates the clinical dose was higher than predicted. 3D global gamma analysis with the prescription dose as reference value was performed. Dice similarity coefficients (DSC) were computed for each isodose volume. RESULTS: Fine-tuned and combined models showed better performance than IDV applicator training. Fine-tuning resulted in modest improvements in about half the metrics, compared to the combined model, while the remainder were mostly unchanged. Fine-tuned MAE = 3.98%/2.69%/5.36%/3.80% for T&O/T&R/T&ON/T&RN, and ME over all voxels = -0.08%/-0.89%/-0.59%/1.42%. ME D2cc were bladder = -0.77%/1.00%/-0.66%/-1.53%, rectum = 1.11%/-0.22%/-0.29%/-3.37%, sigmoid = -0.47%/-0.06%/-2.37%/-1.40%, and ME D90 = 2.6%/-4.4%/4.8%/0.0%. Gamma pass rates (3%/3 mm) were 86%/91%/83%/89%. Mean DSCs were 0.92%/0.92%/0.88%/0.91% for isodoses ≤ 150% of prescription. CONCLUSIONS: 3D BT dose was accurately predicted for all applicator types, as indicated by the low MAE and MEs, high gamma scores and high DSCs. Training on all treatment data overcomes challenges with data scarcity in each applicator type, resulting in superior performance than can be achieved by training on IDV applicators alone. This could presumably be explained by the fact that the larger, more diverse dataset allows the neural network to learn underlying trends and characteristics in dose that are common to all treatment applicators. Accurate, applicator-specific dose predictions could enable automated, knowledge-based planning for any cervical brachytherapy treatment.

A feasibility study of adaptive radiation therapy for postprostatectomy prostate cancer.

Postoperative prostate radiotherapy requires large planning target volume (PTV) margins to account for motion and deformation of the prostate bed. Adaptive radiation therapy (ART) can incorporate image-guidance data to personalize PTVs that maintain coverage while reducing toxicity. We present feasibility and dosimetry results of a prospective study of postprostatectomy ART. Twenty-one patients were treated with single-adaptation ART. Conventional treatments were delivered for fractions 1 to 6 and adapted plans for the remaining 27 fractions. Clinical target volumes (CTVs) and small bowel delineated on fraction 1 to 4 CBCT were used to generate adapted PTVs and planning organ-at-risk (OAR) volumes for adapted plans. PTV volume and OAR dose were compared between ART and conventional using Wilcoxon signed-rank tests. Weekly CBCT were used to assess the fraction of CTV covered by PTV, CTV D99, and small bowel D1cc. Clinical metrics were compared using a Student's t-test (p < 0.05 significant). Offline adaptive planning required 1.9 ± 0.4 days (mean ± SD). ART decreased mean adapted PTV volume 61 ± 37 cc and bladder wall D50 compared with conventional treatment (p < 0.01). The CTV was fully covered for 96% (97%) of fractions with ART (conventional). Reconstructing dose on weekly CBCT, a nonsignificant reduction in CTV D99 was observed with ART (94%) compared to conventional (96%). Reduced CTV D99 with ART was significantly correlated with large anterior-posterior rectal diameter on simulation CT. ART reduced the number of fractions exceeding our institution's small bowel D1c limit from 14% to 7%. This study has demonstrated the feasibility of offline ART for post-prostatectomy cancer. ART facilitates PTV volume reduction while maintaining reasonable CTV coverage and can reduce the dose to adjacent normal tissues.

Automated treatment planning framework for brachytherapy of cervical cancer using 3D dose predictions.

Objective. To lay the foundation for automated knowledge-based brachytherapy treatment planning using 3D dose estimations, we describe an optimization framework to convert brachytherapy dose distributions directly into dwell times (DTs).Approach. A dose rate kerneld(r,θ,φ)was produced by exporting 3D dose for one dwell position from the treatment planning system and normalizing by DT. By translating and rotating this kernel to each dwell position, scaling by DT and summing over all dwell positions, dose was computed (Dcalc). We used a Python-coded COBYLA optimizer to iteratively determine the DTs that minimize the mean squared error betweenDcalcand reference doseDref, computed using voxels withDref80%-120% of prescription. As validation of the optimization, we showed that the optimizer replicates clinical plans whenDref= clinical dose in 40 patients treated with tandem-and-ovoid (T&O) or tandem-and-ring (T&R) and 0-3 needles. Then we demonstrated automated planning in 10 T&O usingDref= dose predicted from a convolutional neural network developed in past work. Validation and automated plans were compared to clinical plans using mean absolute differences (MAD=1N∑n=1Nabsxn-xn') over all voxels (xn= Dose,N= #voxels) and DTs (xn= DT,N= #dwell positions), mean differences (MD) in organD2ccand high-risk CTV D90 over all patients (where positive indicates higher clinical dose), and mean Dice similarity coefficients (DSC) for 100% isodose contours.Main results. Validation plans agreed well with clinical plans (MADdose= 1.1%, MADDT= 4 s or 0.8% of total plan time,D2ccMD = -0.2% to 0.2% and D90 MD = -0.6%, DSC = 0.99). For automated plans, MADdose= 6.5% and MADDT= 10.3 s (2.1%). The slightly higher clinical metrics in automated plans (D2ccMD = -3.8% to 1.3% and D90 MD = -5.1%) were due to higher neural network dose predictions. The overall shape of the automated dose distributions were similar to clinical doses (DSC = 0.91).Significance. Automated planning with 3D dose predictions could provide significant time savings and standardize treatment planning across practitioners, regardless of experience.

A standardized workflow for respiratory-gated motion management decision-making.

PURPOSE: Motion management of tumors within the lung and abdomen is challenging because it requires balancing tissue sparing with accuracy of hitting the target, while considering treatment delivery efficiency. Physicists can play an important role in analyzing four-dimensional computed tomography (4DCT) data to recommend the optimal respiratory gating parameters for a patient. The goal of this work was to develop a standardized procedure for making recommendations regarding gating parameters and planning margins for lung and gastrointestinal stereotactic body radiotherapy (SBRT) treatments. In doing so, we hoped to simplify decision-making and analysis, and provide a tool for troubleshooting complex cases. METHODS: Factors that impact gating decisions and planning target volume (PTV) margins were identified. The gating options included gating on exhale with approximately a 50% duty cycle (Gate3070), exhale gating with a reduced duty cycle (Gate4060), and treating for most of respiration, excluding only extreme inhales and exhales (Gate100). A standard operating procedure was developed, as well as a physics consult document to communicate motion management recommendations to other members of the treatment team. This procedure was implemented clinically for 1 year and results are reported below. RESULTS: Identified factors that impact motion management included the magnitude of motion observed on 4DCT, the regularity of breathing and quality of 4DCT data, and ability to observe the target on fluoroscopy. These were collated into two decision tables-one specific to lung tumors and another for gastrointestinal tumors-such that a physicist could answer a series of questions to determine the optimal gating and PTV margin. The procedure was used clinically for 252 sites from 213 patients treated with respiratory-gated SBRT and standardized practice across our 12-member physics team. CONCLUSION: Implementation of a standardized procedure for respiratory gating had a positive impact in our clinic, improving efficiency and ease of 4DCT analysis and standardizing gating decision-making amongst physicists.

Knowledge-based three-dimensional dose prediction for tandem-and-ovoid brachytherapy.

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 (σ), as well as isodose 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 ΔDx=Dx,actual-Dx,predicted mean, standard deviation, and Pearson correlation coefficient further quantifying model performance. RESULTS: Ranges of voxel-wise dose difference accuracy (δD¯±σ) 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 ΔD¯90±σΔD = -0.19 ± 0.55Gy (-0.09 ± 0.67 Gy), bladder ΔD¯2cc±σΔD = -0.06 ± 0.54Gy (-0.17 ± 0.67 Gy), rectum ΔD¯2cc±σΔD= -0.03 ± 0.36Gy (-0.04 ± 0.46 Gy), and sigmoid ΔD¯2cc±σΔD = -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.

Evaluation of dose differences between intracavitary applicators for cervical brachytherapy using knowledge-based models.

PURPOSE: Currently, there is a lack of patient-specific tools to guide brachytherapy planning and applicator choice for cervical cancer. The purpose of this study is to evaluate the accuracy of organ-at-risk (OAR) dose predictions using knowledge-based intracavitary models, and the use of these models and clinical data to determine the dosimetric differences of tandem-and-ring (T&R) and tandem-and-ovoids (T&O) applicators. MATERIALS AND METHODS: Knowledge-based models, which predict organ D2cc, were trained on 77/75 cases and validated on 32/38 for T&R/T&O applicators. Model performance was quantified using ΔD2cc=D2cc,actual-D2cc,predicted, with standard deviation (σ(ΔD2cc)) representing precision. Model-predicted applicator dose differences were determined by applying T&O models to T&R cases, and vice versa, and compared to clinically-achieved D2cc differences. Applicator differences were assessed using a Student's t-test (p < 0.05 significant). RESULTS: Validation T&O/T&R model precision was 0.65/0.55 Gy, 0.55/0.38 Gy, and 0.43/0.60 Gy for bladder, rectum and sigmoid, respectively, and similar to training. When applying T&O/T&R models to T&R/T&O cases, bladder, rectum and sigmoid D2cc values in EQD2 were on average 5.69/2.62 Gy, 7.31/6.15 Gy and 3.65/0.69 Gy lower for T&R, with similar HRCTV volume and coverage. Clinical data also showed lower T&R OAR doses, with mean EQD2 D2cc deviations of 0.61 Gy, 7.96 Gy (p < 0.01) and 5.86 Gy (p < 0.01) for bladder, rectum and sigmoid. CONCLUSIONS: Accurate knowledge-based dose prediction models were developed for two common intracavitary applicators. These models could be beneficial for standardizing and improving the quality of brachytherapy plans. Both models and clinical data suggest that significant OAR sparing can be achieved with T&R over T&O applicators, particularly for the rectum.

Knowledge-based dose prediction models to inform gynecologic brachytherapy needle supplementation for locally advanced cervical cancer.

PURPOSE: The use of interstitial needles, combined with intracavitary applicators, enables customized dose distributions and is beneficial for complex cases, but increases procedure time. Overall, applicator selection is not standardized and depends on physician expertise and preference. The purpose of this study is to determine whether dose prediction models can guide needle supplementation decision-making for cervical cancer. MATERIALS AND METHODS: Intracavitary knowledge-based models for organ-at-risk (OAR) dose estimation were trained and validated for tandem-and-ring/ovoids (T&R/T&O) implants. Models were applied to hybrid cases with 1-3 implanted needles to predict OAR dose without needles. As a reference, 70/67 hybrid T&R/T&O cases were replanned without needles, following a standardized procedure guided by dose predictions. If a replanned dose exceeded the dose objective, the case was categorized as requiring needles. Receiver operating characteristic (ROC) curves of needle classification accuracy were generated. Optimal classification thresholds were determined from the Youden Index. RESULTS: Needle supplementation reduced dose to OARs. However, 67%/39% of replans for T&R/T&O met all dose constraints without needles. The ROC for T&R/T&O models had an area-under-curve of 0.89/0.86, proving high classification accuracy. The optimal threshold of 99%/101% of the dose limit for T&R/T&O resulted in classification sensitivity and specificity of 78%/86% and 85%/78%. CONCLUSIONS: Needle supplementation reduced OAR dose for most cases but was not always required to meet standard dose objectives, particularly for T&R cases. Our knowledge-based dose prediction model accurately identified cases that could have met constraints without needle supplementation, suggesting that such models may be beneficial for applicator selection.

Knowledge-Based Planning for Intact Cervical Cancer.

Cervical cancer radiotherapy is often complicated by significant variability in the quality and consistency of treatment plans. Knowledge-based planning (KBP), which utilizes prior patient data to correlated achievable optimal dosimetry with patient-specific anatomy, has demonstrated promise as a quality control tool for controlling this variability, with consequences for patient outcomes, as well as for the reliability of data from multi-institutional clinical trials. In this article we highlight the application of KBP-based quality control to cervical cancer radiotherapy. We discuss the potential impact of KBP on multi-institutional clinical trials to standardize cervical cancer treatment planning across diverse clinics, and discuss challenges and progress in the implementation of KBP for brachytherapy treatment planning. Additionally, we briefly discuss secondary applications of KBP for cervical cancer. The emerging picture from these studies indicates several exciting opportunities for increasing the utilization of KBP in day-to-day cervical cancer radiotherapy.

A knowledge-based organ dose prediction tool for brachytherapy treatment planning of patients with cervical cancer.

PURPOSE: The purpose of this study is to explore knowledge-based organ-at-risk dose estimation for intracavitary brachytherapy planning for cervical cancer. Using established external-beam knowledge-based dose-volume histogram (DVH) estimation methods, we sought to predict bladder, rectum, and sigmoid D2cc for tandem and ovoid treatments. METHODS AND MATERIALS: A total of 136 patients with loco-regionally advanced cervical cancer treated with 456 (356:100 training:validation ratio) CT-based tandem and ovoid brachytherapy fractions were analyzed. Single fraction prescription doses were 5.5-8 Gy with dose criteria for the high-risk clinical target volume, bladder, rectum, and sigmoid. DVH estimations were obtained by subdividing training set organs-at-risk into high-risk clinical target volume boundary distance subvolumes and computing cohort-averaged differential DVHs. Full DVH estimation was then performed on the training and validation sets. Model performance was quantified by ΔD2cc = D2cc(actual)-D2cc(predicted) (mean and standard deviation). ΔD2cc between training and validation sets were compared with a Student's t test (p < 0.01 significant). Categorical variables (physician, fraction-number, total fractions, and case complexity) that might explain model variance were examined using an analysis of variance test (Bonferroni-corrected p < 0.01 threshold). RESULTS: Training set deviations were bladder ΔD2cc = -0.04 ± 0.61 Gy, rectum ΔD2cc = 0.02 ± 0.57 Gy, and sigmoid ΔD2cc = -0.05 ± 0.52 Gy. Model predictions on validation set did not statistically differ: bladder ΔD2cc = -0.02 ± 0.46 Gy (p = 0.80), rectum ΔD2cc = -0.007 ± 0.47 Gy (p = 0.53), and sigmoid ΔD2cc = -0.07 ± 0.47 Gy (p = 0.70). The only significant categorical variable was the attending physician for bladder and rectum ΔD2cc. CONCLUSION: A simple boundary distance-driven knowledge-based DVH estimation exhibited promising results in predicting critical brachytherapy dose metrics. Future work will examine the utility of these predictions for quality control and automated brachytherapy planning.

Evaluation of Elekta Agility multi-leaf collimator performance using statistical process control tools.

PURPOSE: To evaluate the performance and stability of Elekta Agility multi-leaf collimator (MLC) leaf positioning using a daily, automated quality control (QC) test based on megavoltage (MV) images in combination with statistical process control tools, and identify special causes of variations in performance. METHODS: Leaf positions were collected daily for 13 Elekta linear accelerators over 11-37 months using the automated QC test, which analyzes 23 MV images to determine the location of MLC leaves relative to radiation isocenter. Leaf positioning stability was assessed using individual and moving range control charts. Specification levels of ±0.5, ±1, and ±1.5 mm were tested to determine positional accuracy. The durations between out-of-control and out-of-specification events were determined. Peaks in out-of-control leaf positions were identified and correlated to servicing events recorded for the whole duration of data collection. RESULTS: Mean leaf position error was -0.01 mm (range -1.3-1.6). Data stayed within ±1 mm specification for 457 days on average (range 3-838) and within ±1.5 mm for the entire date range. Measurements stayed within ±0.5 mm for 1 day on average (range 0-17); however, our MLC leaves were not calibrated to this level of accuracy. Leaf position varied little over time, as confirmed by tight individual (mean ±0.19 mm, range 0.09-0.43) and moving range (mean 0.23 mm, range 0.10-0.53) control limits. Due to sporadic out-of-control events, the mean in-control duration was 2.8 days (range 1-28.5). A number of factors were found to contribute to leaf position errors and out-of-control behavior, including servicing events, beam spot motion, and image artifacts. CONCLUSIONS: The Elekta Agility MLC model was found to perform with high stability, as evidenced by the tight control limits. The in-specification durations support the current recommendation of monthly MLC QC tests with a ±1 mm tolerance. Future work is on-going to determine if performance can be optimized further using high-frequency QC test results to drive recalibration frequency.