A GPU-based multi-criteria optimization algorithm for HDR brachytherapy

Automatic plan selection using deep network-A prostate study

BACKGROUND

Recently, high-dose-rate (HDR) brachytherapy treatment plans generation was improved with the development of multicriteria optimization (MCO) algorithms that can generate thousands of pareto optimal plans within seconds. This brings a shift, from the objective of generating an acceptable plan to choosing the best plans out of thousands.

PURPOSE

In order to choose the best plans, new criteria beyond usual dosimetrics volumes histogram (DVH) metrics are introduced and a deep learning (DL) framework is added as an automatic plan selection algorithm.

METHODS

The new criteria are visual-like criteria implemented for the bladder, rectum, and urethra. One criterion also takes into account the cold spot in the prostate. Those criteria, along with commonly used DVH criteria, are used to form classes on which to train the algorithm. The algorithm is trained with an input of two 3D images, dose and mask of the anatomy, in order to rank and automatically select a plan. The confidence in the output is used for ranking and the automatic plan selection. The algorithm is trained on 835 previously treated prostate cancer patients and evaluated on a separated 20 patients cohort previously evaluated by two experts (clinical medical physicists) in an inter-observer MCO study.

RESULTS

The deep network takes 10 s to rank 2000 plans (vs. 5-10 min for experts to rank 4 preferred plans). A total of four different networks are trained which offer different trade-offs. The key trade-offs are the target coverage or the organs at risk (OAR) sparing. The algorithm with the best network achieves no statistical difference with the plans chosen by the two experts for 6 and 9 criteria, respectively, out of 13 criteria (paired t-test with p > $>$ 0.05) while the two experts have no statistical difference between them for 7 criteria.

CONCLUSIONS

The developed approach is flexible since it allows the modification or addition of criteria to obtain different trade-offs in plan quality, per the institution standard. The approach is fast and robust while adding negligible time to MCO planning. These results demonstrate potential for clinical use.

Impact of robust optimization on patient specific error thresholds for high dose rate prostate brachytherapy source tracking

PURPOSE

The purpose of this study was to compare the effect of catheter shift errors and determine patient specific error thresholds (PSETs) for different high dose rate prostate brachytherapy (HDRPBT) plans generated by different forms of inverse optimization.

METHODS

Three plans were generated for 50 HDRPBT patients and PSETs were determined for each of the 3 plans. Plan 1 was the original Oncentra Prostate (v4.2.2.4, Elekta Brachytherapy, Veenendaal, The Netherlands) plan, the second plan used the graphical processor unit multi-criteria optimization (gMCO) algorithm, and plan 3 used gMCO but had a robustness parameter as an additional optimization criterion (gMCOr). gMCO and gMCOr plans were selected from a pool of 2000 pareto optimal plans. gMCO plan selection involved increasing prostate V100% and reducing rectum Dmax/urethra D01.cc progressively until only 1 plan remained. The gMCOr plan was the most robust plan (using robustness parameter) that met the clinical DVH criteria (V100% ≥ 95%, rectum Dmax ≤ 80%, urethra D0.1cc ≤ 118%). PSETs were determined using catheter shift software.

RESULTS

The initial dose volume histogram (DVH) characteristics showed all 50 patient plans met a prostate V100% > 95% and resulted in significant reduction in rectum Dmax and urethra D0.1cc for gMCO and gMCOr plans. No single plan showed benefits in PSETs for all shift directions compared to the other plans, however gMCO and gMCOr plans exhibit the best initial DVH characteristics assuming no errors occur. The robustness parameter showed no significant impact when considered in plan optimization.

CONCLUSIONS

PSETs were found to be equivalent regardless of optimization method. Indicating, no single optimization method can significantly increase the patient specific thresholds.

Penalty weight tuning in high dose rate brachytherapy using multi-objective Bayesian optimization

Objective.Treatment plan optimization in high dose rate brachytherapy often requires manual fine-tuning of penalty weights for each objective, which can be time-consuming and dependent on the planner's experience. To automate this process, this study used a multi-criteria approach called multi-objective Bayesian optimization with q-noisy expected hypervolume improvement as its acquisition function (MOBO-qNEHVI).Approach.The treatment plans of 13 prostate cancer patients were retrospectively imported to a research treatment planning system, RapidBrachyMTPS, where fast mixed integer optimization (FMIO) performs dwell time optimization given a set of penalty weights to deliver 15 Gy to the target volume. MOBO-qNEHVI was used to find patient-specific Pareto optimal penalty weight vectors that yield clinically acceptable dose volume histogram metrics. The relationship between the number of MOBO-qNEHVI iterations and the number of clinically acceptable plans per patient (acceptance rate) was investigated. The performance time was obtained for various parameter configurations.Main results.MOBO-qNEHVI found clinically acceptable treatment plans for all patients. With increasing the number of MOBO-qNEHVI iterations, the acceptance rate grew logarithmically while the performance time grew exponentially. Fixing the penalty weight of the tumour volume to maximum value, adding the target dose as a parameter, initiating MOBO-qNEHVI with 25 parallel sampling of FMIO, and running 6 MOBO-qNEHVI iterations found solutions that delivered 15 Gy to the hottest 95% of the clinical target volume while respecting the dose constraints to the organs at risk. The average acceptance rate for each patient was 89.74% ± 8.11%, and performance time was 66.6 ± 12.6 s. The initiation took 22.47 ± 7.57 s, and each iteration took 7.35 ± 2.45 s to find one Pareto solution.Significance.MOBO-qNEHVI combined with FMIO can automatically explore the trade-offs between treatment plan objectives in a patient specific manner within a minute. This approach can reduce the dependency of plan quality on planner's experience and reduce dose to the organs at risk.

Aleatoric and epistemic uncertainty extraction of patient-specific deep learning-based dose predictions in LDR prostate brachytherapy

Objective.In brachytherapy, deep learning (DL) algorithms have shown the capability of predicting 3D dose volumes. The reliability and accuracy of such methodologies remain under scrutiny for prospective clinical applications. This study aims to establish fast DL-based predictive dose algorithms for low-dose rate (LDR) prostate brachytherapy and to evaluate their uncertainty and stability.Approach.Data from 200 prostate patients, treated with125I sources, was collected. The Monte Carlo (MC) ground truth dose volumes were calculated with TOPAS considering the interseed effects and an organ-based material assignment. Two 3D convolutional neural networks, UNet and ResUNet TSE, were trained using the patient geometry and the seed positions as the input data. The dataset was randomly split into training (150), validation (25) and test (25) sets. The aleatoric (associated with the input data) and epistemic (associated with the model) uncertainties of the DL models were assessed.Main results.For the full test set, with respect to the MC reference, the predicted prostateD90metric had mean differences of -0.64% and 0.08% for the UNet and ResUNet TSE models, respectively. In voxel-by-voxel comparisons, the average global dose difference ratio in the [-1%, 1%] range included 91.0% and 93.0% of voxels for the UNet and the ResUNet TSE, respectively. One forward pass or prediction took 4 ms for a 3D dose volume of 2.56 M voxels (128 × 160 × 128). The ResUNet TSE model closely encoded the well-known physics of the problem as seen in a set of uncertainty maps. The ResUNet TSE rectum D2cchad the largest uncertainty metric of 0.0042.Significance.The proposed DL models serve as rapid dose predictors that consider the patient anatomy and interseed attenuation effects. The derived uncertainty is interpretable, highlighting areas where DL models may struggle to provide accurate estimations. The uncertainty analysis offers a comprehensive evaluation tool for dose predictor model assessment.

Simultaneous catheter and multicriteria optimization for HDR cervical cancer brachytherapy with a complex intracavity/interstitial applicator

BACKGROUND

Complex intracavity and interstitial (IC/IS) applicators, such as the Venezia applicator, can improve the HR-CTV coverage while adequately protecting organs at risk in the treatment of cervical cancer with high-dose-rate (HDR) brachytherapy. Although the Venezia applicator offers more choice for catheter selection, commercially available catheter and dose optimization algorithms are still missing for complex applicators. Moreover, studies on catheter and dose optimization for IC/IS implants in the treatment of cervical cancer are still limited.

PURPOSE

This work aims to combine a GPU-based multi-criteria optimization (gMCO) algorithm with a sparse catheter (SC) optimization algorithm for the Venezia applicator.

METHODS

Fifty-eight cervical cancer patients who received 28 Gy in 4 fx of HDR brachytherapy with the Venezia applicator (combination to external beam radiation therapy) are retrospectively revisited. The modelization of the applicator is done by virtually reconstructing all the IS catheters passing through the ring. Template catheters are reconstructed using an in-house python script. To perform simultaneous MCO and SC optimization (SC+MCO), the objective function includes aggregated dose objectives in a weighted sum and a group sparsity term that individually penalizes the contribution of IS catheters. Plans generated with the SC+MCO algorithm are compared with plans generated with MCO using clinical catheters (CC+MCO) and the clinical plans (CP). The EMBRACE II soft constraints (planning aims) and hard constraints (limits for prescribed dose) are used as plan evaluation criteria.

RESULTS

CC+MCO gives the most important gain with an increase up to 20.7% in meeting all EMBRACE II soft constraints compared with CP. The SC+MCO algorithm (adding catheter optimization to MCO) provides a second order increase (up to 12.1% with total acceptance rate of 60.3% or 35/58) in the acceptance rate versus CC+MCO (total increase of 32.8% vs. CP). Acceptance rate in EMBRACE II hard constraints is 98.3% (57/58) for both CC+MCO and SC+MCO versus 91.4% (53/58) for CP. The median SC+MCO optimization time is 11 s to generate a total of 5000 Pareto-optimal plans with different catheter configurations (position and number) for each fraction.

CONCLUSIONS

Simultaneous catheter and MCO optimization is clinically feasible for HDR cervical cancer brachytherapy using the Venezia applicator. Clinical catheter configurations could be improved and/or the catheter number could be reduced without decreasing plan quality using SC+MCO compared with the CP.

Keeping your best options open with AI-based treatment planning in prostate and cervix brachytherapy

PURPOSE

Without a clear definition of an optimal treatment plan, no optimization model can be perfect. Therefore, instead of automatically finding a single "optimal" plan, finding multiple, yet different near-optimal plans, can be an insightful approach to support radiation oncologists in finding the plan they are looking for.

METHODS AND MATERIALS

BRIGHT is a flexible AI-based optimization method for brachytherapy treatment planning that has already been shown capable of finding high-quality plans that trade-off target volume coverage and healthy tissue sparing. We leverage the flexibility of BRIGHT to find plans with similar dose-volume criteria, yet different dose distributions. We further describe extensions that facilitate fast plan adaptation should planning aims need to be adjusted, and straightforwardly allow incorporating hospital-specific aims besides standard protocols.

RESULTS

Results are obtained for prostate (n = 12) and cervix brachytherapy (n = 36). We demonstrate the possible differences in dose distribution for optimized plans with equal dose-volume criteria. We furthermore demonstrate that adding hospital-specific aims enables adhering to hospital-specific practice while still being able to automatically create cervix plans that more often satisfy the EMBRACE-II protocol than clinical practice. Finally, we illustrate the feasibility of fast plan adaptation.

CONCLUSIONS

Methods such as BRIGHT enable new ways to construct high-quality treatment plans for brachytherapy while offering new insights by making explicit the options one has. In particular, it becomes possible to present to radiation oncologists a manageable set of alternative plans that, from an optimization perspective are equally good, yet differ in terms of coverage-sparing trade-offs and shape of the dose distribution.

Development of patient and catheter specific error thresholds for high dose rate prostate brachytherapy

BACKGROUND

In-vivo source tracking has been an active topic of research in the field of high-dose rate brachytherapy in recent years to verify accuracy in treatment delivery. Although detection systems for source tracking are being developed, the allowable threshold of treatment error is still unknown and is likely patient-specific due to anatomy and planning variation.

PURPOSE

The purpose of this study was to determine patient and catheter-specific shift error thresholds for in-vivo source tracking during high-dose-rate prostate brachytherapy (HDRPBT).

METHODS

A module was developed in the previously described graphical processor unit multi-criteria optimization (gMCO) algorithm. The module generates systematic catheter shift errors retrospectively into HDRPBT treatment plans, performed on 50 patients. The catheter shift model iterates through the number of catheters shifted in the plan (from 1 to all catheters), the direction of shift (superior, inferior, medial, lateral, cranial, and caudal), and the magnitude of catheter shift (1-6 mm). For each combination of these parameters, 200 error plans were generated, randomly selecting the catheters in the plan to shift. After shifts were applied, dose volume histogram (DVH) parameters were re-calculated. Catheter shift thresholds were then derived based on plans where DVH parameters were clinically unacceptable (prostate V100 < 95%, urethra D0.1cc > 118%, and rectum Dmax > 80%). Catheter thresholds were also Pearson correlated to catheter robustness values.

RESULTS

Patient-specific thresholds varied between 1 to 6 mm for all organs, in all shift directions. Overall, patient-specific thresholds typically decrease with an increasing number of catheters shifted. Anterior and inferior directions were less sensitive than other directions. Pearson's correlation test showed a strong correlation between catheter robustness and catheter thresholds for the rectum and urethra, with correlation values of -0.81 and -0.74, respectively (p < 0.01), but no correlation was found for the prostate.

CONCLUSIONS

It was possible to determine thresholds for each patient, with thresholds showing dependence on shift direction, and number of catheters shifted. Not every catheter combination is explorable, however, this study shows the feasibility to determine patient-specific thresholds for clinical application. The correlation of patient-specific thresholds with the equivalent robustness value indicated the need for robustness consideration during plan optimization and treatment planning.

Robust stochastic optimization of needle configurations for robotic HDR prostate brachytherapy

BACKGROUND

Ideally, inverse planning for HDR brachytherapy (BT) should include the pose of the needles which define the trajectory of the source. This would be particularly interesting when considering the additional freedom and accuracy in needle pose which robotic needle placement enables. However, needle insertion typically leads to tissue deformation, resulting in uncertainty regarding the actual pose of the needles with respect to the tissue.

PURPOSE

To efficiently address uncertainty during inverse planning for HDR BT in order to robustly optimize the pose of the needles before insertion, that is, to facilitate path planning for robotic needle placement.

METHODS

We use a form of stochastic linear programming to model the inverse treatment planning problem. To account for uncertainty, we consider random tissue displacements at the needle tip to simulate tissue deformation. Conventionally for stochastic linear programming, each simulated deformation is reflected by an addition to the linear programming problem which increases problem size and computational complexity substantially and leads to impractical runtime. We propose two efficient approaches for stochastic linear programming. First, we consider averaging dose coefficients to reduce the problem size. Second, we study weighting of the slack variables of an adjusted linear problem to approximate the full stochastic linear program. We compare different approaches to optimize the needle configurations and evaluate their robustness with respect to different amounts of tissue deformation.

RESULTS

Our results illustrate that stochastic planning can improve the robustness of the treatment with respect to deformation. The proposed approaches approximating stochastic linear programming better conform to the tissue deformation compared to conventional linear programming. They show good correlation with the plans computed after deformation while reducing the runtime by two orders of magnitude compared to the complete stochastic linear program. Robust optimization of needle configurations takes on average 59.42 s. Skew needle configurations lead to mean coverage improvements compared to parallel needles from 0.39 to 2.94 percentage points, when 8 mm tissue deformation is considered. Considering tissue deformations from 4  to 10 mm during planning with weighted stochastic optimization and skew needles generally results in improved mean coverage from 1.77 to 4.21 percentage points.

CONCLUSIONS

We show that efficient stochastic optimization allows selecting needle configurations which are more robust with respect to potentially negative effects of target deformation and displacement on the achievable prescription dose coverage. The approach facilitates robust path planning for robotic needle placement.

TrDosePred: A deep learning dose prediction algorithm based on transformers for head and neck cancer radiotherapy

BACKGROUND

Intensity-Modulated Radiation Therapy (IMRT) has been the standard of care for many types of tumors. However, treatment planning for IMRT is a time-consuming and labor-intensive process.

PURPOSE

To alleviate this tedious planning process, a novel deep learning based dose prediction algorithm (TrDosePred) was developed for head and neck cancers.

METHODS

The proposed TrDosePred, which generated the dose distribution from a contoured CT image, was a U-shape network constructed with a convolutional patch embedding and several local self-attention based transformers. Data augmentation and ensemble approach were used for further improvement. It was trained based on the dataset from Open Knowledge-Based Planning Challenge (OpenKBP). The performance of TrDosePred was evaluated with two mean absolute error (MAE) based scores utilized by OpenKBP challenge (i.e., Dose score and DVH score) and compared to the top three approaches of the challenge. In addition, several state-of-the-art methods were implemented and compared to TrDosePred.

RESULTS

The TrDosePred ensemble achieved the dose score of 2.426 Gy and the DVH score of 1.592 Gy on the test dataset, ranking at 3rd and 9th respectively in the leaderboard on CodaLab as of writing. In terms of DVH metrics, on average, the relative MAE against the clinical plans was 2.25% for targets and 2.17% for organs at risk.

CONCLUSIONS

A transformer-based framework TrDosePred was developed for dose prediction. The results showed a comparable or superior performance as compared to the previous state-of-the-art approaches, demonstrating the potential of transformer to boost the treatment planning procedures.

Modern Tools for Modern Brachytherapy

This review aims to showcase the brachytherapy tools and technologies that have emerged during the last 10 years. Soft-tissue contrast using magnetic resonance and ultrasound imaging has seen enormous growth in use to plan all forms of brachytherapy. The era of image-guided brachytherapy has encouraged the development of advanced applicators and given rise to the growth of individualised 3D printing to achieve reproducible and predictable implants. These advances increase the quality of implants to better direct radiation to target volumes while sparing normal tissue. Applicator reconstruction has moved beyond manual digitising, to drag and drop of three-dimensional applicator models with embedded pre-defined source pathways, ready for auto-recognition and automation. The simplified TG-43 dose calculation formalism directly linked to reference air kerma rate of high-energy sources in the medium water remains clinically robust. Model-based dose calculation algorithms accounting for tissue heterogeneity and applicator material will advance the field of brachytherapy dosimetry to become more clinically accurate. Improved dose-optimising toolkits contribute to the real-time and adaptive planning portfolio that harmonises and expedites the entire image-guided brachytherapy process. Traditional planning strategies remain relevant to validate emerging technologies and should continue to be incorporated in practice, particularly for cervical cancer. Overall, technological developments need commissioning and validation to make the best use of the advanced features by understanding their strengths and limitations. Brachytherapy has become high-tech and modern by respecting tradition and remaining accessible to all.

Applying the column generation method to the intensity modulated high dose rate brachytherapy inverse planning problem

Objective.Intensity modulated high dose rate brachytherapy (IMBT) is a rapidly developing application of brachytherapy where anisotropic dose distributions can be produced at each source dwell position. This technique is made possible by placing rotating metallic shields inside brachytherapy needles or catheters. By dynamically directing the radiation towards the tumours and away from the healthy tissues, a more conformal dose distribution can be obtained. The resulting treatment planning involves optimizing dwell position and shield angle (DPSA). The aim of this study was to investigate the column generation method for IMBT treatment plan optimization.Approach.A column generation optimization algorithm was developed to optimize the dwell times and shield angles. A retrospective study was performed on 10 prostate cases using RapidBrachyMCTPS. At every iteration, the plan was optimized with the chosen DPSA which would best improve the cost function that was added to the plan. The optimization process was stopped when the remaining DPSAs would not add value to the plan to limit the plan complexity.Main results.The average number of DPSAs and voxels were 2270 and 7997, respectively. The column generation approach yielded near-optimal treatment plans by using only 11% of available DPSAs on average in ten prostate cases. The coverage and organs at risk constraints passed in all ten cases.Significance.The column generation method produced high-quality deliverable prostate IMBT plans. The treatment plan quality reached a plateau, where adding more DPSAs had a minimal effect on dose volume histogram parameters. The iterative nature of the column generation method allows early termination of the treatment plan creation process as soon as the dosimetric indices from dose volume histogram satisfy the clinical requirements or if their values stabilize.

Towards artificial intelligence-based automated treatment planning in clinical practice: A prospective study of the first clinical experiences in high-dose-rate prostate brachytherapy

PURPOSE

This prospective study evaluates our first clinical experiences with the novel ``BRachytherapy via artificial Intelligent GOMEA-Heuristic based Treatment planning'' (BRIGHT) applied to high-dose-rate prostate brachytherapy.

METHODS AND MATERIALS

Between March 2020 and October 2021, 14 prostate cancer patients were treated in our center with a 15Gy HDR-brachytherapy boost. BRIGHT was used for bi-objective treatment plan optimization and selection of the most desirable plans from a coverage-sparing trade-off curve. Selected BRIGHT plans were imported into the commercial treatment planning system Oncentra Brachy . In Oncentra Brachy a dose distribution comparison was performed for clinical plan choice, followed by manual fine-tuning of the preferred BRIGHT plan when deemed necessary. The reasons for plan selection, clinical plan choice, and fine-tuning, as well as process speed were monitored. For each patient, the dose-volume parameters of the (fine-tuned) clinical plan were evaluated.

RESULTS

In all patients, BRIGHT provided solutions satisfying all protocol values for coverage and sparing. In four patients not all dose-volume criteria of the clinical plan were satisfied after manual fine-tuning. Detailed information on tumour coverage, dose-distribution, dwell time pattern, and insight provided by the patient-specific trade-off curve, were used for clinical plan choice. Median time spent on treatment planning was 42 min, consisting of 16 min plan optimization and selection, and 26 min undesirable process steps.

CONCLUSIONS

BRIGHT is implemented in our clinic and provides automated prostate high-dose-rate brachytherapy planning with trade-off based plan selection. Based on our experience, additional optimization aims need to be implemented to further improve direct clinical applicability of treatment plans and process efficiency.

Brachytherapy evolution as seen today

While brachytherapy is the oldest form of radiation therapy, it is also a very exciting field from both physics and clinical perspectives. From the physics standpoint, brachytherapy dosimetry is largely being governed by the inverse-square law, leading to an unparalleled dose deposition kernel (dose emitted by a seed or single dwell position), even compared to proton or heavy-ion beamlets. There is slightly more dose beyond the central deposition point, but comparatively very little prior to it, that is, little or no entrance dose! It is easy to sum multiple dwell positions that cover a tumor, and the intensity can be modulated quite effectively using dwell times. From a clinical perspective, what sets brachytherapy apart from other intraoperative modalities (e.g., laser, radiofrequency, cryogenic) is our ability to precisely calculate the energy deposited across the relevant patient geometry, anticipate the effect from known dose-outcome relationships, and deliver that energy with exquisite control and selectively to the target volume while sparing organs at risks. This targeting ability has improved substantially over the last two decades. It is built upon key foundational elements, many of which stem from the research and development within our medical physics community. This article provides an overview of these elements that combine to make brachytherapy a successful and developing radiotherapy modality.

Catheters and dose optimization using a modified CVT algorithm and multi-criteria optimization in prostate HDR brachytherapy

Currently, in HDR brachytherapy planning, the catheter's positions are often selected by the planner which involves the planner's experience. The catheters are then inserted using a template which helps to guide the catheters. For certain applications, it is of interest to choose the optimal location and number of catheters needed for dose coverage and potential decrease of the treatment's toxicity. Hence, it is of great importance to develop patient-specific algorithms for catheters and dose optimization. A modified Centroidal Voronoi tessellation (CVT) algorithm is implemented and merged with a GPU-based multi-criteria optimization algorithm (gMCO). The CVT algorithm optimizes the catheters' positions, and the gMCO algorithm optimizes the dwell times and dwell positions. The CVT algorithm can be used simultaneously for insertion with or without a template. Some improvements to the CVT algorithm are presented such as a new way of considering the area that needs to be covered. One hundred and eight previously treated prostates HDR cases using real-time ultrasound (US) are used to evaluate the different optimization procedures. The plan robustness is evaluated using two types of errors; deviations (random) in the insertion and deviation (systematic) in the reconstruction of the catheters. Using gMCO on clinically inserted catheter increases the acceptance rate by 37% for RTOG criteria. Our results show that all the patients respect RTOG criteria with 11 catheters using CVT+gMCO with a template of 5 mm. The number of catheters needed for all patients to respect RTOG criteria with the freehand technique is 10 catheters using CVT+gMCO. When deviations are introduced, using a template, the acceptance rate goes to 85% with 3 mm deviations using 11 catheters. This decrease is less significant when the number of catheters is higher, decreasing by less than 5% with a 3 mm deviation using 13 catheters or more. In conclusion, it is feasible to decrease the number of catheters needed to treat most patients. Some cases still need a high number of catheters to reach the plan's criteria. Using gMCO allows an increase in the plan quality while using CVT reduces the number of catheters. A higher number of catheters equates to plans that are more robust to deviations. This article is protected by copyright. All rights reserved.

Emerging technologies in brachytherapy

Brachytherapy is a mature treatment modality. The literature is abundant in terms of review articles and comprehensive books on the latest established as well as evolving clinical practices. The intent of this article is to part ways and look beyond the current state-of-the-art and review emerging technologies that are noteworthy and perhaps may drive the future innovations in the field. There are plenty of candidate topics that deserve a deeper look, of course, but with practical limits in this communicative platform, we explore four topics that perhaps is worthwhile to review in detail at this time. First, intensity modulated brachytherapy (IMBT) is reviewed. The IMBT takes advantage ofanisotropicradiation profile generated through intelligent high-density shielding designs incorporated onto sources and applicators such to achieve high quality plans. Second, emerging applications of 3D printing (i.e. additive manufacturing) in brachytherapy are reviewed. With the advent of 3D printing, interest in this technology in brachytherapy has been immense and translation swift due to their potential to tailor applicators and treatments customizable to each individual patient. This is followed by, in third, innovations in treatment planning concerning catheter placement and dwell times where new modelling approaches, solution algorithms, and technological advances are reviewed. And, fourth and lastly, applications of a new machine learning technique, called deep learning, which has the potential to improve and automate all aspects of brachytherapy workflow, are reviewed. We do not expect that all ideas and innovations reviewed in this article will ultimately reach clinic but, nonetheless, this review provides a decent glimpse of what is to come. It would be exciting to monitor as IMBT, 3D printing, novel optimization algorithms, and deep learning technologies evolve over time and translate into pilot testing and sensibly phased clinical trials, and ultimately make a difference for cancer patients. Today's fancy is tomorrow's reality. The future is bright for brachytherapy.

Optimization in treatment planning of high dose-rate brachytherapy - Review and analysis of mathematical models

Treatment planning in high dose-rate brachytherapy has traditionally been conducted with manual forward planning, but inverse planning is today increasingly used in clinical practice. There is a large variety of proposed optimization models and algorithms to model and solve the treatment planning problem. Two major parts of inverse treatment planning for which mathematical optimization can be used are the decisions about catheter placement and dwell time distributions. Both these problems as well as integrated approaches are included in this review. The proposed models include linear penalty models, dose-volume models, mean-tail dose models, quadratic penalty models, radiobiological models, and multiobjective models. The aim of this survey is twofold: (i) to give a broad overview over mathematical optimization models used for treatment planning of brachytherapy and (ii) to provide mathematical analyses and comparisons between models. New technologies for brachytherapy treatments and methods for treatment planning are also discussed. Of particular interest for future research is a thorough comparison between optimization models and algorithms on the same dataset, and clinical validation of proposed optimization approaches with respect to patient outcome.

Monte Carlo simulation of the effect of magnetic fields on brachytherapy dose distributions in lung tissue material

The aim of this work was to use TOPAS Monte Carlo simulations to model the effect of magnetic fields on dose distributions in brachytherapy lung treatments, under ideal and clinical conditions. Idealistic studies were modeled consisting of either a monoenergetic electron source of 432 keV, or a polyenergetic electron source using the spectrum of secondary electrons produced by 192Ir gamma-ray irradiation. The electron source was positioned in the center of a homogeneous, lung tissue phantom (ρ = 0.26 g/cm3). Conversely, the clinical study was simulated using the VariSource VS2000 192Ir source in a patient with a lung tumor. Three contoured volumes were considered: the tumor, the planning tumor volume (PTV), and the lung. In all studies, dose distributions were calculated in the presence or absence of a constant magnetic field of 3T. Also, TG-43 parameters were calculated for the VariSource and compared with published data from EGS-brachy (EGSnrc) and PENELOPE. The magnetic field affected the dose distributions in the idealistic studies. For the monoenergetic and poly-energetic studies, the radial distance of the 10% iso-dose line was reduced in the presence of the magnetic field by 64.9% and 24.6%, respectively. For the clinical study, the magnetic field caused differences of 10% on average in the patient dose distributions. Nevertheless, differences in dose-volume histograms were below 2%. Finally, for TG-43 parameters, the dose-rate constant from TOPAS differed by 0.09% ± 0.33% and 0.18% ± 0.33% with respect to EGS-brachy and PENELOPE, respectively. The geometry and anisotropy functions differed within 1.2% ± 1.1%, and within 0.0% ± 0.3%, respectively. The Lorentz forces inside a 3T magnetic resonance machine during 192Ir brachytherapy treatment of the lung are not large enough to affect the tumor dose distributions significantly, as expected. Nevertheless, large local differences were found in the lung tissue. Applications of this effect are therefore limited by the fact that meaningful differences appeared only in regions containing air, which is not abundant inside the human.

Bi-objective optimization of catheter positions for high-dose-rate prostate brachytherapy

Purpose

Bi‐objective simultaneous optimization of catheter positions and dwell times for high‐dose‐rate (HDR) prostate brachytherapy, based directly on dose‐volume indices, has shown promising results. However, optimization with the state‐of‐the‐art evolutionary algorithm MO‐RV‐GOMEA so far required several hours of runtime, and resulting catheter positions were not always clinically feasible. The aim of this study is to extend the optimization model and apply GPU parallelization to achieve clinically acceptable computation times. The resulting optimization procedure is compared with a previously introduced method based solely on geometric criteria, the adapted Centroidal Voronoi Tessellations (CVT) algorithm.

Methods

Bi‐objective simultaneous optimization was performed with a GPU‐parallelized version of MO‐RV‐GOMEA. This optimization of catheter positions and dwell times was retrospectively applied to the data of 26 patients previously treated with HDR prostate brachytherapy for 8–16 catheters (steps of 2). Optimization of catheter positions using CVT was performed in seconds, after which optimization of only the dwell times using MO‐RV‐GOMEA was performed in 1 min.

Results

Simultaneous optimization of catheter positions and dwell times using MO‐RV‐GOMEA was performed in 5 min. For 16 down to 8 catheters (steps of 2), MO‐RV‐GOMEA found plans satisfying the planning‐aims for 20, 20, 18, 14, and 11 out of the 26 patients, respectively. CVT achieved this for 19, 17, 13, 9, and 2 patients, respectively. The P‐value for the difference between MO‐RV‐GOMEA and CVT was 0.023 for 16 catheters, 0.005 for 14 catheters, and <0.001 for 12, 10, and 8 catheters.

Conclusions

With bi‐objective simultaneous optimization on a GPU, high‐quality catheter positions can now be obtained within 5 min, which is clinically acceptable, but slower than CVT. For 16 catheters, the difference between MO‐RV‐GOMEA and CVT is clinically irrelevant. For 14 catheters and less, MO‐RV‐GOMEA outperforms CVT in finding plans satisfying all planning‐aims.

Evaluating the impact of real-time multicriteria optimizers integrated with interactive plan navigation tools for HDR brachytherapy

PURPOSE

Currently in high-dose-rate (HDR) brachytherapy planning, manual fine-tuning of an objective function is a common practice. Furthermore, automated planning approaches such as multicriteria optimization (MCO) are still limited to the automatic generation of a single treatment plan. This study aims to quantify planning efficiency gains when using a graphics processing unit-based MCO (gMCO) algorithm combined with a novel graphical user interface (gMCO-GUI) that integrates efficient automated and interactive plan navigation tools.

METHODS AND MATERIALS

The gMCO algorithm was used to generate 1000 Pareto optimal plans per case for 379 prostate cases. gMCO-GUI was developed to allow plan navigation through all plans. gMCO-GUI integrates interactive parameter selection tools directly with the optimization algorithm to allow plan navigation. The quality of each plan was evaluated based on the Radiation Treatment Oncology Group 0924 protocol and a more stringent institutional protocol (INSTp). gMCO-GUI allows real-time time display of the dose-volume histogram indices, the dose-volume histogram curves, and the isodose lines during the plan navigation.

RESULTS

Over the 379 cases, the fraction of Radiation Treatment Oncology Group 0924 protocol valid plans with target coverage greater than 95% was 90.8%, compared with 66.0% for clinical plans. The fraction of INSTp valid plans with target coverage greater than 95% was 81.8%, compared with 62.3% for clinical plans. The average time to compute 1000 deliverable plans with gMCO was 12.5 s, including the full computation of the 3D dose distributions.

CONCLUSIONS

Combining the gMCO algorithm with automated and interactive plan navigation tools resulted in simultaneous gains in both plan quality and planning efficiency.

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 D_2cc 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 ΔD_2cc = D_2cc(actual)-D_2cc(predicted) (mean and standard deviation). ΔD_2cc 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 ΔD_2cc = -0.04 ± 0.61 Gy, rectum ΔD_2cc = 0.02 ± 0.57 Gy, and sigmoid ΔD_2cc = -0.05 ± 0.52 Gy. Model predictions on validation set did not statistically differ: bladder ΔD_2cc = -0.02 ± 0.46 Gy (p = 0.80), rectum ΔD_2cc = -0.007 ± 0.47 Gy (p = 0.53), and sigmoid ΔD_2cc = -0.07 ± 0.47 Gy (p = 0.70). The only significant categorical variable was the attending physician for bladder and rectum ΔD_2cc. 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.

A Pareto-based beam orientation optimization method for spot scanning intensity-modulated proton therapy

PURPOSE

To provide a proof of principle of a Pareto-based method to automatically generate optimal intensity-modulated proton therapy (IMPT) plans for various noncoplanar beam orientations.

METHODS

A novel multicriteria beam orientation optimization (MCBOO) method was developed to generate Pareto database of optimal plans. The MCBOO method automatically explores the beam orientations and the scalarization parameters of the IMPT plans simultaneously. The MCBOO method is based on multicriteria bilevel optimization (i.e., hierarchical optimization with two nested levels, named the upper and lower level optimization). In MCBOO, the upper level optimization explores the noncoplanar beam orientation space, while the lower level explores the scalarization parameters for a given beam orientation. Differential evolution method was used in both levels, and the Pareto optimal plans were aggregated from the bilevel optimizations to construct the Pareto database. The MCBOO method was implemented on a multinode multi-GPU cluster, and it was tested on three brain tumor patient cases. The Pareto database of the three patients was generated for a set of DVH-based objectives. A statistical analysis was performed between a selected set of MCBOO plans and the manual plan (plan with manually selected beam orientation based on the clinical experience and optimized with the same single plan iterative optimizer used in the MCBOO). The selected set of MCBOO plans consisted of plans that matched the performance of the manual plan [i.e., MCBOO plans that have the same target coverage (within 2%) as the manual plan or better and achieved the same dose (within 2%) or lower to all of the organs at risks (OARs) but one OAR]. Additionally, a dosimetric comparison between of one of the selected MCBOO plans vs the manual plan was conducted.

RESULTS

The multicriteria beam orientation optimization algorithm automatically generated Pareto plans for the three noncoplanar brain tumor cases. The MCBOO plans provided an alternative objective trade-offs to the manual plan. The selected MCBOO plans showed a reduction in dose to multiple organs at risk vs the manual plan with a maximum value which ranged between 10.8 and 12.9 Gy for the three patients. The trade-off of the OAR dose reduction resulted in higher dose to no more than one OAR for each of the selected MCBOO plans vs the manual plan. The maximum dose increase in the MCBOO plans over the manual plan ranged from 7.8 to 11.8 Gy.

CONCLUSIONS

A novel multicriteria beam orientation optimization method was developed and tested on three IMPT patient cases. The method automatically generates Pareto plans database by exploring the noncoplanar beam orientations. The method was able to identify beam orientations with Pareto optimal plans that are comparable to the manually created plans with varying objective trade-offs.