Online Adaptive Proton Therapy Research Articles

First clinical implementation of a highly efficient daily online adapted proton therapy (DAPT) workflow

Objective.This study presents the first clinical implementation of an efficient online daily adaptive proton therapy workflow (DAPT).Approach.The DAPT workflow includes apre-treatment phase,where atemplateand afallback planare optimized on the planning computed tomography (CT). In theonline phase, theadapted planis re-optimized on daily images from an in-room CT. Daily structures are rigidly propagated from the planning CT. Automated Quality Assurance (QA) involves geometric, sanity checks and an independent dose calculation from the machine files. Differences from the template plan are analyzed field-by-field, and clinical plan is assessed by reviewing the achieved clinical goals using a traffic light protocol. If the daily adapted plan fails any QA or clinical goals, the fallback plan is used. In theoffline phasethe delivered dose is recalculated from log-files onto the daily CT, and a gamma analysis is performed (3%/3 mm). The DAPT workflow has been applied to selected adult patients treated in rigid anatomy for the last serie of the treatment between October 2023 and April 2024.Main Results.DAPT treatment sessions averaged around 23 min [range: 15-30 min] and did not exceed the typical 30 minute time slot. Treatment adaptation, including QA and clinical plan assessment, averaged just under 7 min [range: 3:30-16 min] per fraction. All plans passed the online QAs steps. In the offline phase a good agreement with the log-files reconstructed dose was achieved (minimum gamma pass rate of 97.5%). The online adapted plan was delivered for >85% of the fractions. In 92% of total fractions, adapted plans exhibited improved individual dose metrics to the targets and/or organs at risk.Significance.This study demonstrates the successful implementation of an online daily DAPT workflow. Notably, the duration of a DAPT session did not exceed the time slot typically allocated for non-DAPT treatment. As far as we are aware, this is a first clinical implementation of daily online adaptive proton therapy.

A comparative study on automatic treatment planning for online adaptive proton therapy of esophageal cancer: which combination of deformable registration and deep learning planning tools performs the best?

Objective.To demonstrate the feasibility of integrating fully-automated online adaptive proton therapy strategies (OAPT) within a commercially available treatment planning system and underscore what limits their clinical implementation. These strategies leverage existing deformable image registration (DIR) algorithms and state-of-the-art deep learning (DL) networks for organ segmentation and proton dose prediction.Approach.Four OAPT strategies featuring automatic segmentation and robust optimization were evaluated on a cohort of 17 patients, each undergoing a repeat CT scan. (1) DEF-INIT combines deformably registered contours with template-based optimization. (2) DL-INIT, (3) DL-DEF, and (4) DL-DL employ a nnU-Net DL network for organ segmentation and a controlling ROIs-guided DIR algorithm for internal clinical target volume (iCTV) segmentation. DL-INIT uses this segmentation alongside template-based optimization, DL-DEF integrates it with a dose-mimicking (DM) step using a reference deformed dose, and DL-DL merges it with DM on a reference DL-predicted dose. All strategies were evaluated on manual contours and contours used for optimization and compared with manually adapted plans. Key dose volume metrics like iCTV D98% are reported.Main results.iCTV D98% was comparable in manually adapted plans and for all strategies in nominal cases but dropped to 20 Gy in worst-case scenarios for a few patients per strategy, highlighting the need to correct segmentation errors in the target volume. Evaluations on optimization contours showed minimal relative error, with some outliers, particularly in template-based strategies (DEF-INIT and DL-INIT). DL-DEF achieves a good trade-off between speed and dosimetric quality, showing a passing rate (iCTV D98% > 94%) of 90% when evaluated against 2, 4 and 5 mm setup error and of 88% when evaluated against 7 mm setup error. While template-based methods are more rigid, DL-DEF and DL-DL have potential for further enhancements with proper DM algorithm tuning.Significance.Among investigated strategies, DL-DEF and DL-DL demonstrated promising within 10 min OAPT implementation results and significant potential for improvements.

831: Online-adaptive proton therapy: Feasibility of prompt-gamma verification for CBCT-based adapted plan

831: Online-adaptive proton therapy: Feasibility of prompt-gamma verification for CBCT-based adapted plan

Multi-institutional experimental validation of online adaptive proton therapy workflows

Objective. To experimentally validate two online adaptive proton therapy (APT) workflows using Gafchromic EBT3 films and optically stimulated luminescent dosimeters (OSLDs) in an anthropomorphic head-and-neck phantom. Approach. A three-field proton plan was optimized on the planning CT of the head-and-neck phantom with 2.0 Gy(RBE) per fraction prescribed to the clinical target volume. Four fractions were simulated by varying the internal anatomy of the phantom. Three distinct methods were delivered: daily APT researched by the Paul Scherrer Institute (DAPTPSI), online adaptation researched by the Massachusetts General Hospital (OAMGH), and a non-adaptive (NA) workflow. All methods were implemented and measured at PSI. DAPTPSI performed full online replanning based on analytical dose calculation, optimizing to the same objectives as the initial treatment plan. OAMGH performed Monte-Carlo-based online plan adaptation by only changing the fluences of a subset of proton beamlets, mimicking the planned dose distribution. NA delivered the initial plan with a couch-shift correction based on in-room imaging. For all 12 deliveries, two films and two sets of OSLDs were placed at different locations in the phantom. Main results. Both adaptive methods showed improved dosimetric results compared to NA. For film measurements in the presence of anatomical variations, the [min-max] gamma pass rates (3%/3 mm) between measured and clinically approved doses were [91.5%–96.1%], [94.0%–95.8%], and [67.2%–93.1%] for DAPTPSI, OAMGH, and NA, respectively. The OSLDs confirmed the dose calculations in terms of absolute dosimetry. Between the two adaptive workflows, OAMGH showed improved target coverage, while DAPTPSI showed improved normal tissue sparing, particularly relevant for the brainstem. Significance. This is the first multi-institutional study to experimentally validate two different concepts with respect to online APT workflows. It highlights their respective dosimetric advantages, particularly in managing interfractional variations in patient anatomy that cannot be addressed by non-adaptive methods, such as internal anatomy changes.

Yet anOther Dose Algorithm (YODA) for independent computations of dose and dose changes due to anatomical changes

Objective. To assess the viability of a physics-based, deterministic and adjoint-capable algorithm for performing treatment planning system independent dose calculations and for computing dosimetric differences caused by anatomical changes. Approach. A semi-numerical approach is employed to solve two partial differential equations for the proton phase-space density which determines the deposited dose. Lateral hetereogeneities are accounted for by an optimized (Gaussian) beam splitting scheme. Adjoint theory is applied to approximate the change in the deposited dose caused by a new underlying patient anatomy. Main results. The dose engine’s accuracy was benchmarked through three-dimensional gamma index comparisons against Monte Carlo simulations done in TOPAS. For a lung test case, the worst passing rate with (1 mm, 1%, 10% dose cut-off) criteria is 94.55%. The effect of delivering treatment plans on repeat CTs was also tested. For non-robustly optimized plans the adjoint component was accurate to 5.7% while for a robustly optimized plan it was accurate to 4.8%. Significance. Yet anOther Dose Algorithm is capable of accurate dose computations in both single and multi spot irradiations when compared to TOPAS. Moreover, it is able to compute dosimetric differences due to anatomical changes with small to moderate errors thereby facilitating its use for patient-specific quality assurance in online adaptive proton therapy.

The partial adaptation strategy for online-adaptive proton therapy: A proof of concept study in head and neck cancer patients.

The accuracy of intensity-modulated proton therapy (IMPT) is greatly affected by anatomy variations that might occur during the treatment course. Online plan adaptations have been proposed as a solution to intervene promptly during a treatment session once the anatomy changes are detected. The implementation of online-adaptive proton therapy (OAPT) is still hindered by time-consuming tasks in the workflow. The study introduces the novel concept of partial adaptation and aims at investigating its feasibility as a potential solution to parallelize tasks during an OAPT workflow for saving valuable in-room time. The proof-of-principle simulation study includes datasets from six head and neck cancer (HNC) patients, each consisting of one planning CT (pCT) and three contoured control CTs (cCTs). Robust 3-field normo-fractionated initial IMPT plans were generated on the pCTs with a standardized field configuration, delivering 66 Gy and 54Gy to the high-risk and low-risk clinical target volume (CTVHigh and CTVLow), respectively.For each cCT, a dose-mimicking-based partial adaptation was applied: two fields were adapted on the current anatomy taking into account the background dose of the first non-adapted field supposedly delivered in the meantime. Fraction doses on the cCTs resulting from partially adapted plans with different first (non-adapted) field assignments were compared against those from non-adapted and fully adapted plans regarding target coverage and organs at risk (OARs) sparing. The robustness of partially adapted plans was also evaluated. Partially adapted plans showed comparable results to fully adapted plans and were superior to non-adapted plans for both target coverage and OAR sparing. Target coverage degradation in the non-adapted plans (median D98%: 95.9% and 97.5% for CTVLow and CTVHigh, respectively) was recovered by both partial (98.0% and 98.5%) and full adaptation (98.2% and 98.7%) in comparison to the initial plans (98.7% and 98.8%). The initial hotspot dose for the CTVHigh (median D2%: 101.8%) increased in the non-adapted plans (102.9%) and was recovered by the adaptive strategies (partial: 102.5%, full: 101.9%). The near-maximum dose (D0.01cc) to brainstem and spinal cord was within clinical constraints for all investigated dose distributions, but clearly increased for no adaptation and improved in the (both partially and fully) adapted plans with respect to the non-adapted ones. The parotids' median doses (D50) were mainly patient-specific depending on the proximity to the target region, but anyway lower for the partially and fully adapted plans compared to the non-adapted ones. OAR sparing was furthermore improved for the partially adapted plans in comparison to full adaptation. Robustness of the target dose metrics was preserved in all evaluated scenarios. For OAPT of HNC patients, partial adaptation is able to generate plans of superior conformity to non-adapted plans and of comparable conformity as fully adapted plans, while having the potential to speed up the online-adaptive workflows. Thus, partial adaptation represents an intermediate approach until fast online adaptation workflows become available. Furthermore, it can be applied in workflows where online treatment verification stops the delivery and triggers an online adaptation for the remaining fraction.

A reference-point-method-based online proton treatment plan re-optimization strategy and a novel solution to planning constraint infeasibility problem

Objective. Propose a highly automated treatment plan re-optimization strategy suitable for online adaptive proton therapy. The strategy includes a rapid re-optimization method that generates quality replans and a novel solution that efficiently addresses the planning constraint infeasibility issue that can significantly prolong the re-optimization process. Approach. We propose a systematic reference point method (RPM) model that minimizes the l-infinity norm from the initial treatment plan in the daily objective space for online re-optimization. This model minimizes the largest objective value deviation among the objectives of the daily replan from their reference values, leading to a daily replan similar to the initial plan. Whether a set of planning constraints is feasible with respect to the daily anatomy cannot be known before solving the corresponding optimization problem. The conventional trial-and-error-based relaxation process can cost a significant amount of time. To that end, we propose an optimization problem that first estimates the magnitude of daily violation of each planning constraint. Guided by the violation magnitude and clinical importance of the constraints, the constraints are then iteratively converted into objectives based on their priority until the infeasibility issue is solved. Main results. The proposed RPM-based strategy generated replans similar to the offline manual replans within the online time requirement for six head and neck and four breast patients. The average target D 95 and relevant organ at risk sparing parameter differences between the RPM replans and clinical offline replans were −0.23, −1.62 Gy for head and neck cases and 0.29, −0.39 Gy for breast cases. The proposed constraint relaxation solution made the RPM problem feasible after one round of relaxation for all four patients who encountered the infeasibility issue. Significance. We proposed a novel RPM-based re-optimization strategy and demonstrated its effectiveness on complex cases, regardless of whether constraint infeasibility is encountered.

A retrospective study on the investigation of potential dosimetric benefits of online adaptive proton therapy for head and neck cancer.

Proton therapy is sensitive to anatomical changes, often occurring in head-and-neck (HN) cancer patients. Although multiple studies have proposed online adaptive proton therapy (APT), there is still a concern in the radiotherapy community about the necessity of online APT. We have performed a retrospective study to investigate the potential dosimetric benefits of online APT for HN patients relative to the current offline APT. Our retrospective study has a patient cohort of 10 cases. To mimic online APT, we re-evaluated the dose of the in-use treatment plan on patients' actual treatment anatomy captured by cone-beam CT (CBCT) for each fraction and performed a templated-based automatic replanning if needed, assuming that these were performed online before treatment delivery. Cumulative dose of the simulated online APT course was calculated and compared with that of the actual offline APT course and the designed plan dose of the initial treatment plan (referred to as nominal plan). The ProKnow scoring system was employed and adapted for our study to quantify the actual quality of both courses against our planning goals. The average score of the nominal plans over the 10 cases is 41.0, while those of the actual offline APT course and our simulated online course is 25.8 and 37.5, respectively. Compared to the offline APT course, our online course improved dose quality for all cases, with the score improvement ranging from 0.4 to 26.9 and an average improvement of 11.7. The results of our retrospective study have demonstrated that online APT can better address anatomical changes for HN cancer patients than the current offline replanning practice. The advanced artificial intelligence based automatic replanning technology presents a promising avenue for extending potential benefits of online APT.

Generation and evaluation of anatomy-preserving virtual CT for online adaptive proton therapy.

Daily CTs generated by CBCT correction are required for daily replanning in online-adaptive proton therapy (APT) to effectively deal with inter-fractional changes. Out of the currently available methods, the suitability of a daily CT generation method for proton dose calculation also depends on the anatomical site. We propose an anatomy-preserving virtual CT (APvCT) method as a hybrid method of CBCT correction, which is especially suitable for large anatomy deformations. The accuracy of the hybrid method was assessed by comparison with the corrected CBCT (cCBCT) and virtual CT (vCT) methods in the context of online APT. Seventy-one daily CBCTs of four prostate cancer patients treated with intensity modulated proton therapy (IMPT) were converted to daily CTs using cCBCT, vCT, and the newly proposed APvCT method. In APvCT, planning CT (pCT) were mapped to CBCT geometry using deformable image registration with boundary conditions on controlling regions of interest (ROIs) created with deep learning segmentation on cCBCT. The relative frequency distribution (RFD) of HU, mass density and stopping power ratio (SPR) values were assessed and compared with the pCT. The ROIs in the APvCT and vCT were compared with cCBCT in terms of Dice similarity coefficient (DSC) and mean distance-to-agreement (mDTA). For each patient, a robustly optimized IMPT plan was created on the pCT and subsequent daily adaptive plans on daily CTs. For dose distribution comparison on the same anatomy, the daily adaptive plans on cCBCT and vCT were recalculated on the corresponding APvCT. The dose distributions were compared in terms of isodose volumes and 3D global gamma-index passing rate (GPR) at γ(2%, 2mm) criterion. For all patients, no noticeable difference in RFDs was observed amongst APvCT, vCT, and pCT except in cCBCT, which showed a noticeable difference. The minimum DSC value was 0.96 and 0.39 for contours in APvCT and vCT respectively. The average value of mDTA for APvCT was 0.01cm for clinical target volume and ≤0.01cm for organs at risk, which increased to 0.18cm and ≤0.52cm for vCT. The mean GPR value was 90.9%, 64.5%, and 67.0% for APvCT versus cCBCT, vCT versus cCBCT, and APvCT versus vCT, respectively. When recalculated on APvCT, the adaptive cCBCT and vCT plans resulted in mean GPRs of 89.5±5.1% and 65.9±19.1%, respectively. The mean DSC values for 80.0%, 90.0%, 95.0%, 98.0%, and 100.0% isodose volumes were 0.97, 0.97, 0.97, 0.95, and 0.91 for recalculated cCBCT plans, and 0.89, 0.88, 0.87, 0.85, and 0.81 for recalculated vCT plans. Hausdorff distance for the 100.0% isodose volume in some cases of recalculated cCBCT plans on APvCT exceeded 1.00cm. APvCT contours showed good agreement with reference contours of cCBCT which indicates anatomy preservation in APvCT. A vCT with erroneous anatomy can result in an incorrect adaptive plan. Further, slightly lower values of GPR between the APvCT and cCBCT-based adaptive plans can be explained by the difference in the cCBCT's SPR RFD from the pCT.

Optically stimulated luminescence dosimeters for simultaneous measurement of point dose and dose-weighted LET in an adaptive proton therapy workflow.

To demonstrate the suitability of optically stimulated luminescence detectors (OSLDs) for accurate simultaneous measurement of the absolute point dose and dose-weighted linear energy transfer (LETD) in an anthropomorphic phantom for experimental validation of daily adaptive proton therapy. A clinically realistic intensity-modulated proton therapy (IMPT) treatment plan was created based on a CT of an anthropomorphic head-and-neck phantom made of tissue-equivalent material. The IMPT plan was optimized with three fields to deliver a uniform dose to the target volume covering the OSLDs. Different scenarios representing inter-fractional anatomical changes were created by modifying the phantom. An online adaptive proton therapy workflow was used to recover the daily dose distribution and account for the applied geometry changes. To validate the adaptive workflow, measurements were performed by irradiating Al2O3:C OSLDs inside the phantom. In addition to the measurements, retrospective Monte Carlo simulations were performed to compare the absolute dose and dose-averaged LET (LETD) delivered to the OSLDs. The online adaptive proton therapy workflow was shown to recover significant degradation in dose conformity resulting from large anatomical and positioning deviations from the reference plan. The Monte Carlo simulations were in close agreement with the OSLD measurements, with an average relative error of 1.4% for doses and 3.2% for LETD. The use of OSLDs for LET determination allowed for a correction for the ionization quenched response. The OSLDs appear to be an excellent detector for simultaneously assessing dose and LET distributions in proton irradiation of an anthropomorphic phantom. The OSLDs can be cut to almost any size and shape, making them ideal for in-phantom measurements to probe the radiation quality and dose in a predefined region of interest. Although we have presented the results obtained in the experimental validation of an adaptive proton therapy workflow, the same approach can be generalized and used for a variety of clinical innovations and workflow developments that require accurate assessment of point dose and/or average LET.

Estimating Potential Benefits of Online Adaptive Proton Therapy for Head-and-Neck Cancer: A Retrospective Study

Proton therapy is highly sensitive to anatomical changes and setup variations in head-and-neck (HN) treatments. To address this issue, proton centers often acquire patient CT images weekly to monitor patient anatomical changes during the treatment course and perform offline plan adaptation when needed. However, offline adaptation cannot fully account for daily setup variations or the anatomical changes occurring with high frequency. There are a few groups endeavoring to develop advanced technologies to enable online adaptive proton therapy (APT). However, the necessity of online APT remains controversial, as it is unknown that whether online APT will significantly improve treatment quality and outcomes compared to offline APT. The purpose of this study is to estimate the clinical potential of online APT in the management of HN cancers in relation to the current offline APT. Our retrospective study was conducted with four HN patients (35 fractions per patient), who had been treated with intensity modulated proton therapy and had offline adaptation once or twice during their treatment courses. Synthetic CT (sCT) images were generated from 140 daily CBCT images for us to recalculate the dose of the treatment plan in patient's actual treatment anatomy for each treatment fraction and adapt the plan when warranted. These adaptations were assumed to be performed online before treatment delivery to mimic an online APT course. Accumulative doses were calculated for both courses using the CBCT-based sCT images of every fraction for us to compare the target coverage, organ at risk (OAR) sparing, tumor control probability (TCP) and normal tissue complication probability (NTCP). An in-house script was developed to semi-automate this process in a commercial treatment planning system to facilitate our study. All patients would benefit from online APT to different extents. For the first patient, with OAR doses comparable to the actual offline course, the retrospective online APT course improved dose coverages of the three CTVs from 95.2%, 98.64% and 89.53% to 98.88%, 99.81%, 98.97%, which would lead to a 4.52% improvement in TCP. Similarly, online APT would yield a 2.66% improvement in TCP for the second patient. For the third patient, with comparable CTV dose coverages, the mean doses of right parotid and oral cavity were decreased from 29.52 Gy relative biological effectiveness (RBE) and 41.89 Gy RBE to 22.16 Gy RBE and 34.61 Gy RBE, leading to a reduce of 1.67% and 3.40% in NTCP. The mean dose of right parotid was decreased from 21.71 Gy RBE to 19.37 Gy RBE for the last patient, leading to a reduce of 0.73% in NTCP. Our results showed that online APT could better maintain the treatment plan quality than offline APT for all the four patients, despite their significant anatomical changes. Future investigation will focus on collecting more patient data to obtain statistically significant results and help identify the patients to whom the online APT will be of most benefit.

Beam mask and sliding window-facilitated deep learning-based accurate and efficient dose prediction for pencil beam scanning proton therapy.

Accurate and efficient dose calculation is essential for on-line adaptive planning in proton therapy. Deep learning (DL) has shown promising dose prediction results in photon therapy. However, there is a scarcity of DL-based dose prediction methods specifically designed for proton therapy. Successful dose prediction method for proton therapy should account for more challenging dose prediction problems in pencil beam scanning proton therapy (PBSPT) due to its sensitivity to heterogeneities. To develop a DL-based PBSPT dose prediction workflow with high accuracy and balanced complexity to support on-line adaptive proton therapy clinical decision and subsequent replanning. PBSPT plans of 103 prostate cancer patients (93 for training and the other 10 for independent testing) and 83 lung cancer patients (73 for training and the other 10 for independent testing) previously treated at our institution were included in the study, each with computed tomography scans (CTs), structure sets, and plan doses calculated by the in-house developed Monte-Carlo dose engine (considered as the ground truth in the model training and testing). For the ablation study, we designed three experiments corresponding to the following three methods: (1) Experiment 1, the conventional region of interest (ROI) (composed of targets and organs-at-risk [OARs]) method. (2) Experiment 2, the beam mask (generated by raytracing of proton beams) method to improve proton dose prediction. (3) Experiment 3, the sliding window method for the model to focus on local details to further improve proton dose prediction. A fully connected 3D-Unet was adopted as the backbone. Dose volume histogram (DVH) indices, 3D Gamma passing rates with a criterion of 3%/3mm/10%, and dice coefficients for the structures enclosed by the iso-dose lines between the predicted and the ground truth doses were used as the evaluation metrics. The calculation time for each proton dose prediction was recorded to evaluate the method's efficiency. Compared to the conventional ROI method, the beam mask method improved the agreement of DVH indices for both targets and OARs and the sliding window method further improved the agreement of the DVH indices (for lung cancer, CTV D98 absolute deviation: 0.74 ± 0.18 vs. 0.57 ± 0.21 vs. 0.54 ± 0.15 Gy[RBE], ROI vs. beam mask vs. sliding window methods, respectively). For the 3D Gamma passing rates in the target, OARs, and BODY (outside target and OARs), the beam mask method improved the passing rates in these regions and the sliding window method further improved them (for prostate cancer, targets: 96.93% ± 0.53% vs. 98.88% ± 0.49% vs. 99.97% ± 0.07%, BODY: 86.88% ± 0.74% vs. 93.21% ± 0.56% vs. 95.17% ± 0.59%). A similar trend was also observed for the dice coefficients. This trend was especially remarkable for relatively low prescription isodose lines (for lung cancer, 10% isodose line dice: 0.871 ± 0.027 vs. 0.911 ± 0.023 vs. 0.927 ± 0.017). The dose predictions for all the testing cases were completed within 0.25s. An accurate and efficient deep learning-augmented proton dose prediction framework has been developed for PBSPT, which can predict accurate dose distributions not only inside but also outside ROI efficiently. The framework can potentially further reduce the initial planning and adaptive replanning workload in PBSPT.

Dosimetric comparison of autocontouring techniques for online adaptive proton therapy

Objective. Anatomical and daily set-up uncertainties impede high precision delivery of proton therapy. With online adaptation, the daily plan is reoptimized on an image taken shortly before the treatment, reducing these uncertainties and, hence, allowing a more accurate delivery. This reoptimization requires target and organs-at-risk (OAR) contours on the daily image, which need to be delineated automatically since manual contouring is too slow. Whereas multiple methods for autocontouring exist, none of them are fully accurate, which affects the daily dose. This work aims to quantify the magnitude of this dosimetric effect for four contouring techniques. Approach. Plans reoptimized on automatic contours are compared with plans reoptimized on manual contours. The methods include rigid and deformable registration (DIR), deep-learning based segmentation and patient-specific segmentation. Main results. It was found that independently of the contouring method, the dosimetric influence of using automatic OAR contours is small (<5% prescribed dose in most cases), with DIR yielding the best results. Contrarily, the dosimetric effect of using the automatic target contour was larger (>5% prescribed dose in most cases), indicating that manual verification of that contour remains necessary. However, when compared to non-adaptive therapy, the dose differences caused by automatically contouring the target were small and target coverage was improved, especially for DIR. Significance. The results show that manual adjustment of OARs is rarely necessary and that several autocontouring techniques are directly usable. Contrarily, manual adjustment of the target is important. This allows prioritizing tasks during time-critical online adaptive proton therapy and therefore supports its further clinical implementation.

Detectability of Anatomical Changes With Prompt-Gamma Imaging: First Systematic Evaluation of Clinical Application During Prostate-Cancer Proton Therapy

The development of online-adaptive proton therapy (PT) is essential to overcome limitations encountered by day-to-day variations of the patient's anatomy. Range verification could play an essential role in an online feedback loop for the detection of treatment deviations such as anatomical changes. Here, we present the results of the first systematic patient study regarding the detectability of anatomical changes by a prompt-gamma imaging (PGI) slit-camera system. For 15 patients with prostate cancer, PGI measurements were performed during 105 fractions (201 fields) with in-room control computed tomography (CT)acquisitions. Field-wise doses on control CT scans were manually classified as whether showing relevant or non-relevant anatomical changes. This manual classification of the treatment fields was then used to establish an automatic field-wise ground truth based on spot-wise dosimetric range shifts, which were retrieved from integrated depth-dose (IDD) profiles. To determine the detection capability of anatomical changes with PGI, spot-wise PGI-based range shifts were initially compared with corresponding dosimetric IDD range shifts. As final endpoint, the agreement of a developed field-wise PGI classification model with the field-wise ground truth was determined. Therefore, the PGI model was optimized and tested for a subcohort of 131 and 70 treatment fields, respectively. The correlation between PGI and IDD range shifts was high, ρpearson=0.67 (p < 0.01). Field-wise binary PGI classification resulted in an area under the curve of 0.72 and 0.80 for training and test cohorts, respectively. The model detected relevant anatomical changes in the independent test cohort, with a sensitivity and specificity of 74% and 79%, respectively. For the first time, evidence of the detection capability of anatomical changes in prostate-cancer PT from clinically acquired PGI data is shown. This emphasizes the benefit of PGI-based range verification and demonstrates its potential for online-adaptive PT.

Dose mimicking based strategies for online adaptive proton therapy of head and neck cancer

Objective. To compare a not adapted (NA) robust planning strategy with three fully automated online adaptive proton therapy (OAPT) workflows based on the same optimization method: dose mimicking (DM). The added clinical value and limitations of the OAPT methods are investigated for head and neck cancer (HNC) patients. Approach. The three OAPT strategies aimed at compensating for inter-fractional anatomical changes by mimiking different dose distributions on corrected cone beam CT images (corrCBCTs). Order by complexity, the OAPTs were: (1) online adaptive dose restoration (OADR) where the approved clinical dose on the planning-CT (pCT) was mimicked, (2) online adaptation using DM of the deformed clinical dose from the pCT to corrCBCTs (OADEF), and (3) online adaptation applying DM to a predicted dose on corrCBCTs (OAML). Adaptation was only applied in fractions where the target coverage criteria were not met (D98% < 95% of the prescribed dose). For 10 HNC patients, the accumulated dose distributions over the 35 fractions were calculated for NA, OADR, OADEF, and OAML. Main results. Higher target coverage was observed for all OAPT strategies compared to no adaptation. OADEF and OAML outperformed both NA and OADR and were comparable in terms of target coverage to initial clinical plans. However, only OAML provided comparable NTCP values to those from the clinical dose without statistically significant differences. When the NA initial plan was evaluated on corrCBCTs, 51% of fractions needed adaptation. The adaptation rate decreased significantly to 25% when the last adapted plan with OADR was selected for delivery, to 16% with OADEF, and to 21% with OAML. The reduction was even greater when the best plan among previously generated adapted plans (instead of the last one) was selected. Significance. The implemented OAPT strategies provided superior target coverage compared to no adaptation, higher OAR sparing, and fewer required adaptations.

OC-0117 A deterministic algorithm for patient-specific quality assurance in online adaptive proton therapy

OC-0117 A deterministic algorithm for patient-specific quality assurance in online adaptive proton therapy

Comparison of MR-guided radiotherapy accumulated doses for central lung tumors with non-adaptive and online adaptive proton therapy.

Stereotactic body radiation therapy (SBRT) of central lung tumors with photon or proton therapy has a risk of increased toxicity. Treatment planning studies comparing accumulated doses for state-of-the-art treatment techniques, such as MR-guided radiotherapy (MRgRT) and intensity modulated proton therapy (IMPT), are currently lacking. We conducted a comparison of accumulated doses for MRgRT, robustly optimized non-adaptive IMPT, and online adaptive IMPT for central lung tumors. A special focus was set on analyzing the accumulated doses to the bronchial tree, a parameter linked to high-grade toxicities. Data of 18 early-stage central lung tumor patients, treated at a 0.35T MR-linac in eight or five fractions, were analyzed. Three gated treatment scenarios were compared: (S1) online adaptive MRgRT, (S2) non-adaptive IMPT, and (S3) online adaptive IMPT. The treatment plans were recalculated or reoptimized on the daily imaging data acquired during MRgRT, and accumulated over all treatment fractions. Accumulated dose-volume histogram (DVH) parameters of the gross tumor volume (GTV), lung, heart, and organs-at-risk (OARs) within 2cm of the planning target volume (PTV) were extracted for each scenario and compared in Wilcoxon signed-rank tests between S1 & S2, and S1 & S3. The accumulated GTV D98% was above the prescribed dose for all patients and scenarios. Significant reductions (p < 0.05) of the mean ipsilateral lung dose (S2: -8%; S3: -23%) and mean heart dose (S2: -79%; S3: -83%) were observed for both proton scenarios compared to S1. The bronchial tree D0.1cc was significantly lower for S3 (S1: 48.1 Gy; S3: 39.2 Gy; p = 0.005), but not significantly different for S2 (S2: 45.0 Gy; p = 0.094), compared to S1. The D0.1cc for S2 and S3 compared to S1 was significantly (p < 0.05) smaller for OARs within 1-2cm of the PTV (S1: 30.2 Gy; S2: 24.6 Gy; S3: 23.1 Gy), but not significantlydifferent for OARs within 1cm of the PTV. A significant dose sparing potential of non-adaptive and online adaptive proton therapy compared to MRgRT for OARs in close, but not direct proximity of central lung tumors was identified. The near-maximum dose to the bronchial tree was not significantly different for MRgRT and non-adaptive IMPT. Online adaptive IMPT achieved significantly lower doses to the bronchial tree compared to MRgRT.

A Deterministic Adjoint-Based Semi-Analytical Algorithm for Fast Response Change Computations in Proton Therapy

In this paper we propose a solution to the need for a fast particle transport algorithm in Online Adaptive Proton Therapy capable of cheaply, but accurately computing the changes in patient dose metrics as a result of changes in the system parameters. We obtain the proton phase-space density through the product of the numerical solution to the one-dimensional Fokker-Planck equation and the analytical solution to the Fermi-Eyges equation. Moreover, a corresponding adjoint system was derived and solved for the adjoint flux. The proton phase-space density together with the adjoint flux and the metric (chosen as the energy deposited by the beam in a variable region of interest) allowed assessing the accuracy of our algorithm to different perturbation ranges in the system parameters and regions of interest. The algorithm achieved negligible errors () for small Hounsfield unit (HU) perturbation ranges (–40 HU to 40 HU) and small to moderate errors (3% to 17%) – in line with the well-known limitation of adjoint approaches – for large perturbation ranges (–400 HU to 400 HU) in the case of most clinical interest where the region of interest surrounds the Bragg peak. Given these results coupled with the capability of further improving the timing performance it can be concluded that our algorithm presents a viable solution for the specific purpose of Online Adaptive Proton Therapy.

SWFT-Net: a deep learning framework for efficient fine-tuning spot weights towards adaptive proton therapy

Objective. One critical task for adaptive proton therapy is how to perform spot weight re-tuning and reoptimize plan, both of which are time-consuming and labor intensive. We proposed a deep learning framework (SWFT-Net) to speed up such a task, a starting point for us to move towards online adaptive proton therapy. Approach. For a H&N patient case, a reference intensity modulated proton therapy plan was generated. For data augmentation, spot weights were modified to generate three datasets (DS10, DS30, DS50), corresponding to different levels of weight adjustment. For each dataset, the samples were split into the training and testing groups at a ratio of 8:2 (6400 for training, 1706 for testing). To ease the difficulty of machine learning, the residuals of dose maps and spot weights (i.e. difference relative to a reference) were used as inputs and outputs, respectively. Quantitative analyses were performed in terms of normalized root mean square error (NRMSE) of spot weights, Gamma passing rate and dose difference within the PTV. Main results. The SWFT-Net is able to generate an adapted plan in less than a second with a NVIDIA GeForce RTX 3090 GPU. For the 1706 samples in the testing dataset, the NRMSE is 0.41% (DS10), 1.05% (DS30) and 2.04% (DS50), respectively. Cold/hot spots in the dose maps after adaptation are observed. The mean relative dose difference is 0.64% (DS10), 0.92% (DS30) and 0.88% (DS50), respectively. For all three datasets, the mean Gamma passing rate is consistently over 95% for both 1 mm/1% and 3 mm/3% settings. Significance. The proposed SWFT-Net is a promising tool to help realize adaptive proton therapy. It can be used as an alternative tool to other spot fine-tuning optimization algorithms, likely demonstrating superior performance in terms of speed, accuracy, robustness and minimum human interaction. This study lays down a foundation for us to move further incorporating other factors such as daily anatomical changes and propagated PTVs, and develop a truly online adaptive workflow in proton therapy.

Automatic detection and classification of treatment deviations in proton therapy from realistically simulated prompt gamma imaging data.

A clinical study regarding the potential of range verification in proton therapy (PT) by prompt gamma imaging (PGI) is carried out at our institution. Manual interpretation of the detected spot-wise range shift information is time-consuming, highly complex, and therefore not feasible in a broad routine application. Here, we present an approach to automatically detect and classify treatment deviations in realistically simulated PGI data for head-and-neck cancer (HNC) treatments using convolutional neural networks (CNNs) and conventional machine learning (ML) approaches. For 12 HNC patients and 1 anthropomorphic head phantom (n=13), pencil beam scanning (PBS) treatment plans were generated, and 1 field per plan was assumed to be monitored with a PGI slit camera system. In total, 386 scenarios resembling different relevant or non-relevant treatment deviations were simulated on planning and control CTs and manually classified into 7 classes: non-relevant changes (NR) and relevant changes (RE) triggering treatment intervention due to range prediction errors (±RP), setup errors in beam direction (±SE), anatomical changes (AC), or a combination of such errors (CB). PBS spots with reliable PGI information were considered with their nominal Bragg peak position for the generation of two 3D spatial maps of 16× 16×16 voxels containing PGI-determined range shift and proton number information. Three complexity levels of simulated PGI data were investigated: (I) optimal PGI data, (II) realistic PGI data with simulated Poisson noise based on the locally delivered proton number, and (III) realistic PGI data with an additional positioning uncertainty of the slit camera following an experimentally determined distribution. For each complexity level, 3D-CNNs were trained on a data subset (n=9) using patient-wise leave-one-out cross-validation and tested on an independent test cohort (n=4). Both the binary task of detecting RE and the multi-class task of classifying the underlying error source were investigated. Similarly, four different conventional ML classifiers (logistic regression, multilayer perceptron, random forest, and support vector machine) were trained using five previously established handcrafted features extracted from the PGI data and used for performance comparison. On the test data, the CNN ensemble achieved a binary accuracy of 0.95, 0.96, and 0.93 and a multi-class accuracy of 0.83, 0.81, and 0.76 for the complexity levels (I), (II), and (III), respectively. In the case of binary classification, the CNN ensemble detected treatment deviations in the most realistic scenario with a sensitivity of 0.95 and a specificity of 0.88. The best performing ML classifiers showed a similar test performance. This study demonstrates that CNNs can reliably detect relevant changes in realistically simulated PGI data and classify most of the underlying sources of treatment deviations. The CNNs extracted meaningful features from the PGI data with a performance comparable to ML classifiers trained on previously established handcrafted features. These results highlight the potential of a reliable, automatic interpretation of PGI data for treatment verification, which is highly desired for a broad clinical application and a prerequisite for the inclusion of PGI in an automated feedback loop for online adaptive PT.