Automated Robust Planning for IMPT in Oropharyngeal Cancer Patients Using Machine Learning.

The aim of this study was to evaluate an automated treatment planning method for robustly optimized intensity modulated proton therapy (IMPT) plans for oropharyngeal carcinoma patients and to compare the results with manually optimized robust IMPT plans. An atlas regression forest-based machine learning (ML) model for dose prediction was trained on CT scans, contours, and dose distributions of robust IMPT plans of 88 oropharyngeal cancer (OPC) patients. The ML model was combined with a robust voxel and dose volume histogram-based dose mimicking optimization algorithm for 21 perturbed scenarios to generate a machine-deliverable plan from the predicted dose distribution. Machine learning optimization (MLO) configuration was performed using a cross-validation approach with 3 × 8 tuning patients and comprised of adjustments to the mimicking optimization, to generate higher-quality MLO plans. Independent testing of the MLO algorithm was performed with another 25 patients. Plan quality of clinical and MLO plans was evaluated by the clinical target volume (D98% voxel-wise minimum dose >94%), organ at risk (OAR) doses, and the normal tissue complication probability (NTCP) (sum (Σ) of grade-2 and grade-3 dysphagia and xerostomia). Adequate robust target coverage was achieved in 24/25 clinical plans and in 23/25 MLO plans in the primary clinical target volume (CTV). In the elective CTV, 22/25 clinical plans and 24/25 MLO plans passed the robust target coverage evaluation threshold. The MLO average Σgrade 2 and Σgrade 3 NTCPs were comparable to the clinical plans (Σgrade 2 NTCPs: clinical 47.5% vs MLO 48.4%, Σgrade 3 NTCPs: clinical 11.9% vs MLO 12.3%). Significant increases in OAR average doses in MLO plans were found in the pharynx constrictor muscles (163 cGy, P = .002) and cervical esophagus (265 cGy, P = .002). The MLO plans were created within 45 minutes. This study showed that automated MLO planning can generate robustly optimized MLO plans with quality comparable to clinical plans in OPC patients.

van Bruggen IG ，Huiskes M ，de Vette SPM ，Holmström M ，Langendijk JA ，Both S ，Kierkels RGJ ，Korevaar EW ... -  《-》

Validation of Fully Automated Robust Multicriterial Treatment Planning for Head and Neck Cancer IMPT.

Our purpose was to compare robust intensity modulated proton therapy (IMPT) plans, automatically generated with wish-list-based multicriterial optimization as implemented in Erasmus-iCycle, with manually created robust clinical IMPT plans for patients with head and neck cancer. Thirty-three patients with head and neck cancer were retrospectively included. All patients were previously treated with a manually created IMPT plan with 7000 cGy dose prescription to the primary tumor (clinical target volume [CTV]7000) and 5425 cGy dose prescription to the bilateral elective volumes (CTV5425). Plans had a 4-beam field configuration and were generated with scenario-based robust optimization (21 scenarios, 3-mm setup error, and ±3% density uncertainty for the CTVs). Three clinical plans were used to configure the Erasmus-iCycle wish-list for automated generation of robust IMPT plans for the other 30 included patients, in line with clinical planning requirements. Automatically and manually generated IMPT plans were compared for (robust) target coverage, organ-at-risk (OAR) doses, and normal tissue complication probabilities (NTCP). No manual fine-tuning of automatically generated plans was performed. For all automatically generated plans, voxel-wise minimum D98% values for the CTVs were within clinical constraints and similar to manual plans. All investigated OAR parameters were favorable in the automatically generated plans (all P < .001). Median reductions in mean dose to OARs went up to 667 cGy for the inferior pharyngeal constrictor muscle, and median reductions in D0.03cm3 in serial OARs ranged up to 1795 cGy for the spinal cord surface. The observed lower mean dose in parallel OARs resulted in statistically significant lower NTCP for xerostomia (grade ≥2: 34.4% vs 38.0%; grade ≥3: 9.0% vs 10.2%) and dysphagia (grade ≥2: 11.8% vs 15.0%; grade ≥3: 1.8% vs 2.8%). Erasmus-iCycle was able to produce IMPT dose distributions fully automatically with similar (robust) target coverage and improved OAR doses and NTCPs compared with clinical manual planning, with negligible hands-on planning workload.

Huiskes M ，Kong W ，Oud M ，Crama K ，Rasch C ，Breedveld S ，Heijmen B ，Astreinidou E ... -  《-》

Spot scanning proton arc therapy reduces toxicity in oropharyngeal cancer patients.

Proton arc technology has recently shown dosimetric gains for various treatment indications. The increased number of beams and energy layers (ELs) in proton arc plans, increases the degrees of freedom in plan optimization and thereby flexibility to spare dose in organs at risk (OARs). A relationship exists between dosimetric plan quality, delivery efficiency, the number of ELs -and beams in a proton arc plan. This work aims to investigate the effect of the number of beams and ELs in a proton arc plan, on toxicity and delivery time for oropharyngeal cancer patients (OPC) selected for intensity modulated proton therapy (IMPT) based on the Dutch model-based approach. The EL reduction algorithm iteratively selects ELs from beams equidistantly spaced over a 360° arc. The beams in the final plan may contain multiple ELs, making them suited for static delivery on the studied treatment machine. The produced plans can therefore be called "step and shoot" proton arc plans. The number of beams and ELs were varied to determine the relationship with the planning cost function value, normal tissue complication probability (NTCP) and delivery time. Proton arc plans with robust target coverage and optimal energy layer reduction (ELR) settings to reduce NTCP, were generated for 10 OPC patients. Proton arc plans were compared to clinical volumetric modulated arc therapy (VMAT) and IMPT plans in terms of integral dose, OAR dose, NTCP for xerostomia and dysphagia and delivery time. Furthermore, dose-weighted average linear energy transfer (LETd ) distributions were compared between the IMPT and proton arc plans. A dry run delivery of a plan containing 20 beams and 360 ELs was performed to evaluate delivery time and accuracy. We found 360 ELs distributed over 30 beams generated proton arc plans with near minimal expected plan toxicity. Relative to corresponding IMPT and VMAT plans, an average reduction of 21 ± 3% and 58 ± 10% in integral dose was observed. D m e a n $_{mean}$ was reduced most in the pharyngeal constrictor muscle (PCM) medius structure, with on average 9.0 ± 4.2 Gy(RBE) (p = 0.0002) compared to the clinical IMPT plans. The average NTCP for grade≥2 and grade≥3 xerostomia at 6 months after treatment significantly decreased with 4.7 ± 1.8% (p = 0.002) and 1.7 ± 0.8% (p = 0.002), respectively, while the average NTCP for grade≥2 and grade≥3 dysphagia decreased with 4.4 ± 2.9% (p = 0.002) and 0.9 ± 0.4% (p = 0.002), respectively, increasing the benefit of protons relative to VMAT. For a "step and shoot" proton arc delivery with auto beam sequencing the estimated delivery time is 11 min, similar to the delivery time of a 6-field IMPT treatment. Gamma analysis between the planned and delivered dose distribution resulted in a 99.99% pass rate using 1mm/1% dose difference/distance to agreement criteria. "Step and shoot" proton arc demonstrates potential to further reduce toxicity compared to IMPT and VMAT in OPC treatment. By employing 360 ELs and 30 beams in the proposed ELR method, delivery time can reach clinically acceptable levels without compromising plan toxicity when automatic beam sequencing is available.

de Jong BA ，Battinelli C ，Free J ，Wagenaar D ，Engwall E ，Janssens G ，Langendijk JA ，Korevaar EW ，Both S ... -  《-》

Machine learning-based automatic proton therapy planning: Impact of post-processing and dose-mimicking in plan robustness.

Automated treatment planning strategies are being widely implemented in clinical routines to reduce inter-planner variability, speed up the optimization process, and improve plan quality. This study aims to evaluate the feasibility and quality of intensity-modulated proton therapy (IMPT) plans generated with four different knowledge-based planning (KBP) pipelines fully integrated into a commercial treatment planning system (TPS). A data set containing 60 oropharyngeal cancer patients was split into 11 folds, each containing 47 patients for training, five patients for validation, and five patients for testing. A dose prediction model was trained on each of the folds, resulting in a total of 11 models. Three patients were left out in order to assess if the differences introduced between models were significant. From voxel-based dose predictions, we analyze the two steps that follow the dose prediction: post-processing of the predicted dose and dose mimicking (DM). We focused on the effect of post-processing (PP) or no post-processing (NPP) combined with two different DM algorithms for optimization: the one available in the commercial TPS RayStation (RSM) and a simpler isodose-based mimicking (IBM). Using 55 test patients (five test patients for each model), we evaluated the quality and robustness of the plans generated by the four proposed KBP pipelines (PP-RSM, PP-IBM, NPP-RSM, NPP-IBM). After robust evaluation, dose-volume histogram (DVH) metrics in nominal and worst-case scenarios were compared to those of the manually generated plans. Nominal doses from the four KBP pipelines showed promising results achieving comparable target coverage and improved dose to organs at risk (OARs) compared to the manual plans. However, too optimistic post-processing applied to the dose prediction (i.e. important decrease of the dose to the organs) compromised the robustness of the plans. Even though RSM seemed to partially compensate for the lack of robustness in the PP plans, still 65% of the patients did not achieve the expected robustness levels. NPP-RSM plans seemed to achieve the best trade-off between robustness and OAR sparing. PP and DM strategies are crucial steps to generate acceptable robust and deliverable IMPT plans from ML-predicted doses. Before the clinical implementation of any KBP pipeline, the PP and DM parameters predefined by the commercial TPS need to be modified accordingly with a comprehensive feedback loop in which the robustness of the final dose calculations is evaluated. With the right choice of PP and DM parameters, KBP strategies have the potential to generate IMPT plans within clinically acceptable levels comparable to plans manually generated by dosimetrists.

Borderias-Villarroel E ，Huet Dastarac M ，Barragán-Montero AM ，Helander R ，Holmstrom M ，Geets X ，Sterpin E ... -  《-》

Automated Robust Proton Planning Using Dose-Volume Histogram-Based Mimicking of the Photon Reference Dose and Reducing Organ at Risk Dose Optimization.

Patient selection for proton therapy is increasingly based on proton to photon plan comparisons. To improve efficient decision making, we developed a dose mimicking and reducing (DMR) algorithm to automatically generate a robust proton plan from a reference photon dose, as well as target and organ at risk (OAR) delineations. The DMR algorithm was evaluated in 40 patients with head and neck cancer. The first step of the DMR algorithm comprises dose-volume histogram-based mimicking of the photon dose distribution in the clinical target volumes and OARs. Target robustness is included by mimicking the nominal photon dose in 21 perturbed scenarios. The second step of the optimization aims to reduce the OAR doses while retaining the robust target coverage as achieved in the first step. We evaluated each DMR plan against the manually robustly optimized reference proton plan in terms of plan robustness (voxel-wise minimum dose). Furthermore, the DMR plans were evaluated against the reference photon plan using normal tissue complication probability (NTCP) models of xerostomia, dysphagia, and tube feeding dependence. Consequently, ΔNTCPs were defined as the difference between the NTCPs of the photon and proton plans. The dose distributions of the DMR and reference proton plans were very similar in terms of target robustness and OAR dose values. Regardless of proton planning technique (ie, DMR or reference proton plan), the same treatment modality was selected in 80% (32 of 40) of cases based on the ∑ΔNTCPs. In 15% (6 of 40) of cases, a conflicting decision was made based on relatively small dose differences to the OARs (<2.0 Gy). The DMR algorithm automatically optimized robust proton plans from a photon reference dose that were comparable to the dosimetrist-optimized proton plans in patients with head and neck cancer. This algorithm has been successfully embedded into a framework to automatically select patients for proton therapy based on NTCPs.

Kierkels RGJ ，Fredriksson A ，Both S ，Langendijk JA ，Scandurra D ，Korevaar EW ... -  《-》