In presence of inter-fractional anatomical changes, clinical benefits are anticipated from image-guided adaptive radiotherapy. Nowadays, cone-beam CT (CBCT) imaging is mostly utilized during pre-treatment imaging for position verification. Due to various artifacts, image quality is typically not sufficient for photon or proton dose calculation, thus demanding accurate CBCT correction, as potentially provided by deep learning techniques. This work aimed at investigating the feasibility of utilizing a cycle-consistent generative adversarial network (cycleGAN) for prostate CBCT correction using unpaired training. Thirty-three patients were included. The network was trained to translate uncorrected, original CBCT images (CBCT _org ) into planning CT equivalent images (CBCT _cycleGAN ). HU accuracy was determined by comparison to a previously validated CBCT correction technique (CBCT _cor ). Dosimetric accuracy was inferred for volumetric-modulated arc photon therapy (VMAT) and opposing single-field uniform dose (OSFUD) proton plans, optimized on CBCT _cor and recalculated on CBCT _cycleGAN . Single-sided SFUD proton plans were utilized to assess proton range accuracy. The mean HU error of CBCT _cycleGAN with respect to CBCT _cor decreased from 24 HU for CBCT _org to −6 HU. Dose calculation accuracy was high for VMAT, with average pass-rates of 100%/89% for a 2%/1% dose difference criterion. For proton OSFUD plans, the average pass-rate for a 2% dose difference criterion was 80%. Using a (2%, 2 mm) gamma criterion, the pass-rate was 96%. 93% of all analyzed SFUD profiles had a range agreement better than 3 mm. CBCT correction time was reduced from 6–10 min for CBCT _cor to 10 s for CBCT _cycleGAN . Our study demonstrated the feasibility of utilizing a cycleGAN for CBCT correction, achieving high dose calculation accuracy for VMAT. For proton therapy, further improvements may be required. Due to unpaired training, the approach does not rely on anatomically consistent training data or potentially inaccurate deformable image registration. The substantial speed-up for CBCT correction renders the method particularly interesting for adaptive radiotherapy.

Artificial Intelligence (AI) models in radiation therapy are being developed with increasing pace. Despite this, the radiation therapy community has not widely adopted these models in clinical practice. A cohesive guideline on how to develop, report and clinically validate AI algorithms might help bridge this gap. A Delphi process with all co-authors was followed to determine which topics should be addressed in this comprehensive guideline. Separate sections of the guideline, including Statements, were written by subgroups of the authors and discussed with the whole group at several meetings. Statements were formulated and scored as highly recommended or recommended. The following topics were found most relevant: Decision making, image analysis, volume segmentation, treatment planning, patient specific quality assurance of treatment delivery, adaptive treatment, outcome prediction, training, validation and testing of AI model parameters, model availability for others to verify, model quality assurance/updates and upgrades, ethics. Key references were given together with an outlook on current hurdles and possibilities to overcome these. 19 Statements were formulated. A cohesive guideline has been written which addresses main topics regarding AI in radiation therapy. It will help to guide development, as well as transparent and consistent reporting and validation of new AI tools and facilitate adoption.

Image intensity correction is crucial to enable cone beam computed tomography (CBCT) based radiotherapy dose calculations. This study evaluated three different deep learning based correction methods using a U-shaped convolutional neural network architecture (Unet) in terms of their photon and proton dose calculation accuracy. CT and CBCT imaging data of 42 prostate cancer patients were included. For target ground truth data generation, a CBCT correction method based on CT to CBCT deformable image registration (DIR) was used. The method yields a deformed CT called (i) virtual CT (vCT) which is used to generate (ii) corrected CBCT projections allowing the reconstruction of (iii) a final corrected CBCT image. The single Unet architecture was trained using these three different datasets: (Unet1) raw and corrected CBCT projections, (Unet2) raw CBCT and vCT image slices and (Unet3) raw and reference corrected CBCT image slices. Volumetric arc therapy (VMAT) and proton pencil beam scanning (PBS) single field uniform dose (SFUD) plans were optimized on the reference corrected image and recalculated on the obtained Unet-corrected CBCT images. The mean error (ME) and mean absolute error (MAE) for Unet1/2/3 were Hounsfield units (HU) and HU. The 1% dose difference pass rates were better than 98.4% for VMAT for 8 test patients not seen during training, with little difference between Unets. Gamma evaluation results were even better. For protons a gamma evaluation was employed to account for small range shifts, and mm pass rates for Unet1/2/3 were better than and 91%. A 3 mm range difference threshold was established. Only for Unet3 the 5th and 95th percentiles of the range difference distributions over all fields, test patients and dose profiles were within this threshold. A single Unet architecture was successfully trained using both CBCT projections and CBCT image slices. Since the results of the other Unets were poorer than Unet3, we conclude that training using corrected CBCT image slices as target data is optimal for PBS SFUD proton dose calculations, while for VMAT all Unets provided sufficient accuracy.

Background: Hybrid magnetic resonance (MR)-Linac systems have recently been introduced into clinical practice. The systems allow online adaption of the treatment plan with the aim of compensating for interfractional anatomical changes. The aim of this study was to evaluate the dose volume histogram (DVH)-based dosimetric benefits of online adaptive MR-guided radiotherapy (oMRgRT) across different tumor entities and to investigate which subgroup of plans improved the most from adaption. Methods: Fifty patients treated with oMRgRT for five different tumor entities (liver, lung, multiple abdominal lymph nodes, pancreas, and prostate) were included in this retrospective analysis. Various target volume (gross tumor volume GTV, clinical target volume CTV, and planning target volume PTV) and organs at risk (OAR) related DVH parameters were compared between the dose distributions before and after plan adaption. Results: All subgroups clearly benefited from online plan adaption in terms of improved PTV coverage. For the liver, lung and abdominal lymph nodes cases, a consistent improvement in GTV coverage was found, while many fractions of the prostate subgroup showed acceptable CTV coverage even before plan adaption. The largest median improvements in GTV near-minimum dose (D_98%) were found for the liver (6.3%, p < 0.001), lung (3.9%, p < 0.001), and abdominal lymph nodes (6.8%, p < 0.001) subgroups. Regarding OAR sparing, the largest median OAR dose reduction during plan adaption was found for the pancreas subgroup (-87.0%). However, in the pancreas subgroup an optimal GTV coverage was not always achieved because sparing of OARs was prioritized. Conclusion: With online plan adaptation, it was possible to achieve significant improvements in target volume coverage and OAR sparing for various tumor entities and account for interfractional anatomical changes.

Intensity modulated proton therapy (IMPT) of head and neck (H&N) tumors may benefit from plan adaptation to correct for the dose perturbations caused by weight loss and tumor volume changes observed in these patients. As cone beam CT (CBCT) is increasingly considered in proton therapy, it may be possible to use available CBCT images following intensity correction for plan adaptation. This is the first study exploring IMPT plan adaptation on CBCT images corrected and delineated by deformable image registration of the planning CT (pCT) to the CBCT, yielding a virtual CT (vCT). A Morphons algorithm was used to deform the pCTs and corresponding delineations of 9 H&N cancer patients to a weekly CBCT acquired within ±3 days of a control replanning CT scan (rpCT). The IMPT treatment plans were adapted using the vCT and the adapted and original plans were recalculated on the rpCT for dose/volume parameter evaluation of the impact of adaptation. On the rpCT, the adapted plans were equivalent to the original plans in terms of target volumes D 95 and V 95, but showed a significant reduction of D 2 in these volumes. OAR doses were mostly equivalent or reduced. In particular, the adapted plans did not reduce parotid gland D mean, but the dose to the optical system. For three patients the spinal cord or brain stem received higher, though well below tolerance, maximum dose. Subsequent tightening of the treatment planning constraints for these OARs on new vCT-adapted plans did not degrade target coverage and yielded pCT equivalent plans on the vCT. An offline automated procedure to generate an adapted IMPT plan on CBCT images was developed and investigated. When evaluating the adapted plan on a control rpCT we observed reduced D 2 in target volumes as major improvement. OAR sparing was only partially improved by the procedure. Despite potential limitations in the accuracy of the vCT approach, an improved quality of the adapted plans could be achieved.