A population based statistical model for daily geometric variations in the thorax

Robust optimization incorporating weekly predicted anatomical CTs in IMPT of nasopharyngeal cancer

Objective.This study proposes a robust optimization (RO) strategy utilizing virtual CTs (vCTs) predicted by an anatomical model in intensity-modulated proton therapy (IMPT) for nasopharyngeal cancer (NPC).Methods and Materials.For ten NPC patients, vCTs capturing anatomical changes at different treatment weeks were generated using a population average anatomy model. Two RO strategies of a 6 beams IMPT with 3 mm setup uncertainty (SU) and 3% range uncertainty (RU) were compared: conventional robust optimization (cRO) based on a single planning CT (pCT), and anatomical RO incorporating 2 and 3 predicted anatomies (aRO2 and aRO3). The robustness of these plans was assessed by recalculating them on weekly CTs (week 2-7) and extracting the voxel wise-minimum and maximum doses with 1 mm SU and 3% RU (voxmin\voxmax1mm3%).Results.The aRO plans demonstrated improved robustness in high-risk CTV1 and low-risk CTV 2 coverage compared to cRO plans. The weekly evaluation showed a lower plan adaptation rate for aRO3 (40%) vs. cRO (70%). The weekly nominal and voxmax1mm3%doses to OARs, especially spinal cord, are better controlled relative to their baseline doses at week 1 with aRO plans. The accumulated dose analysis showed that CTV1&2 had adequate coverage and serial organs (spinal cord and brainstem) were within their dose tolerances in the voxmin\voxmax1mm3%, respectively.Conclusion.Incorporating predicted weekly CTs from a population based average anatomy model in RO improves week-to-week target dose coverage and reduces false plan adaptations without increasing normal tissue doses. This approach enhances IMPT plan robustness, potentially facilitating reduced SU and further lowering OAR doses.

Modelling systematic anatomical uncertainties of head and neck cancer patients during fractionated radiotherapy treatment

Objective.Head and neck cancer patients experience systematic as well as random day to day anatomical changes during fractionated radiotherapy treatment. Modelling the expected systematic anatomical changes could aid in creating treatment plans which are more robust against such changes.Approach.Inter- patient correspondence aligned all patients to a model space. Intra- patient correspondence between each planning CT scan and on treatment cone beam CT scans was obtained using diffeomorphic deformable image registration. The stationary velocity fields were then used to develop B-Spline based patient specific (SM) and population average (AM) models. The models were evaluated geometrically and dosimetrically. A leave-one-out method was used to compare the training and testing accuracy of the models.Main results.Both SMs and AMs were able to capture systematic changes. The average surface distance between the registration propagated contours and the contours generated by the SM was less than 2 mm, showing that the SM are able to capture the anatomical changes which a patient experiences during the course of radiotherapy. The testing accuracy was lower than the training accuracy of the SM, suggesting that the model overfits to the limited data available and therefore, also captures some of the random day to day changes. For most patients the AMs were a better estimate of the anatomical changes than assuming there were no changes, but the AMs could not capture the variability in the anatomical changes seen in all patients. No difference was seen in the training and testing accuracy of the AMs. These observations were highlighted in both the geometric and dosimetric evaluations and comparisons.Significance.In this work, a SM and AM are presented which are able to capture the systematic anatomical changes of some head and neck cancer patients over the course of radiotherapy treatment. The AM is able to capture the overall trend of the population, but there is large patient variability which highlights the need for more complex, capable population models.

DiffuseRT: predicting likely anatomical deformations of patients undergoing radiotherapy

Objective. Predicting potential deformations of patients can improve radiotherapy treatment planning. Here, we introduce new deep-learning models that predict likely anatomical changes during radiotherapy for head and neck cancer patients.Approach. Denoising diffusion probabilistic models (DDPMs) were developed to generate fraction-specific anatomical changes based on a reference cone-beam CT (CBCT), the fraction number and the dose distribution delivered. Three distinct DDPMs were developed: (1) theimage modelwas trained to directly generate likely future CBCTs, (2) the deformable vector field (DVF) model was trained to generate DVFs that deform a reference CBCT and (3) thehybrid modelwas trained similarly to the DVF model, but without relying on an external deformable registration algorithm. The models were trained on 9 patients with longitudinal CBCT images (224 CBCTs) and evaluated on 5 patients (152 CBCTs).Results. The generated images mainly exhibited random positioning shifts and small anatomical changes for early fractions. For later fractions, all models predicted weight losses in accordance with the training data. The distributions of volume and position changes of the body, esophagus, and parotids generated with the image and hybrid models were more similar to the ground truth distribution than the DVF model, evident from the lower Wasserstein distance achieved with the image (0.33) and hybrid model (0.30) compared to the DVF model (0.36). Generating several images for the same fraction did not yield the expected variability since the ground truth anatomical changes were only in 76% of the fractions within the 95% bounds predicted with the best model. Using the generated images for robust optimization of simplified proton therapy plans improved the worst-case clinical target volume V95 with 7% compared to optimizing with 3 mm set-up robustness while maintaining a similar integral dose.Significance. The newly developed DDPMs generate distributions similar to the real anatomical changes and have the potential to be used for robust anatomical optimization.

Probabilistic evaluation of plan quality for time-dependent anatomical deformations in head and neck cancer patients

PURPOSE

In addition to patient set-up uncertainties, anatomical deformations, e.g., weight loss, lead to time-dependent differences between the planned and delivered dose in a radiotherapy course that currently cannot easily be predicted. The aim of this study was to create time-varying prediction models to describe both the average and residual anatomical deformations.

METHODS

Weekly population-based principal component analysis models were generated from on-treatment cone-beam CT scans (CBCTs) of 30 head and neck cancer patients, with additional data of 35 patients used as a validation cohort. We simulated treatment courses accounting for a) anatomical deformations, b) set-up uncertainties and c) a combination of both. The dosimetric effects of the simulated deformations were compared to a direct dose accumulation based on deformable registration of the CBCT data.

RESULTS

Set-up uncertainties were seen to have a larger effect on the organ at risk (OAR) doses than anatomical deformations for all OARs except the larynx and the primary CTV. Distributions from simulation results were in good agreement with those of the accumulated dose.

CONCLUSIONS

We present a novel method of modelling time-varying organ deformations in head and neck cancer. The effect on the OAR doses from these deformations are smaller than the effect of set-up uncertainties for most OARs. These models can, for instance, be used to predict which patients could benefit from adaptive radiotherapy, prior to commencing treatment.

A probabilistic deep learning model of inter-fraction anatomical variations in radiotherapy

Objective. In radiotherapy, the internal movement of organs between treatment sessions causes errors in the final radiation dose delivery. To assess the need for adaptation, motion models can be used to simulate dominant motion patterns and assess anatomical robustness before delivery. Traditionally, such models are based on principal component analysis (PCA) and are either patient-specific (requiring several scans per patient) or population-based, applying the same set of deformations to all patients. We present a hybrid approach which, based on population data, allows to predict patient-specific inter-fraction variations for an individual patient.Approach. We propose a deep learning probabilistic framework that generates deformation vector fields warping a patient's planning computed tomography (CT) into possible patient-specific anatomies. This daily anatomy model (DAM) uses few random variables capturing groups of correlated movements. Given a new planning CT, DAM estimates the joint distribution over the variables, with each sample from the distribution corresponding to a different deformation. We train our model using dataset of 312 CT pairs with prostate, bladder, and rectum delineations from 38 prostate cancer patients. For 2 additional patients (22 CTs), we compute the contour overlap between real and generated images, and compare the sampled and 'ground truth' distributions of volume and center of mass changes.Results. With a DICE score of 0.86 ± 0.05 and a distance between prostate contours of 1.09 ± 0.93 mm, DAM matches and improves upon previously published PCA-based models, using as few as 8 latent variables. The overlap between distributions further indicates that DAM's sampled movements match the range and frequency of clinically observed daily changes on repeat CTs.Significance. Conditioned only on planning CT values and organ contours of a new patient without any pre-processing, DAM can accurately deformations seen during following treatment sessions, enabling anatomically robust treatment planning and robustness evaluation against inter-fraction anatomical changes.

An organ deformation model using Bayesian inference to combine population and patient-specific data

Objective.Organ deformation models have the potential to improve delivery and reduce toxicity of radiotherapy, but existing data-driven motion models are based on either patient-specific or population data. We propose to combine population and patient-specific data using a Bayesian framework. Our goal is to accurately predict individual motion patterns while using fewer scans than previous models.Approach.We have derived and evaluated two Bayesian deformation models. The models were applied retrospectively to the rectal wall from a cohort of prostate cancer patients. These patients had repeat CT scans evenly acquired throughout radiotherapy. Each model was used to create coverage probability matrices (CPMs). The spatial correlations between these estimated CPMs and the ground truth, derived from independent scans of the same patient, were calculated.Main results.Spatial correlation with ground truth were significantly higher for the Bayesian deformation models than both patient-specific and population-derived models with 1, 2 or 3 patient-specific scans as input. Statistical motion simulations indicate that this result will also hold for more than 3 scans.Significance.The improvement over previous models means that fewer scans per patient are needed to achieve accurate deformation predictions. The models have applications in robust radiotherapy planning and evaluation, among others.

Evaluating principal component analysis models for representing anatomical changes in head and neck radiotherapy

Background and purpose

Anatomical changes during radiotherapy pose a challenge to robustness of plans. Principal component analysis (PCA) is commonly used to model such changes. We propose a toolbox to evaluate how closely a given PCA model can represent actual deformations seen in the patient and highlight regions where the model struggles to capture these changes.

Materials and methods

We propose to calculate a residual error map from the difference between an actual displacement vector field (DVF) and the closest DVF that the PCA model can produce. This was done by taking the inner product of the DVF with the PCA components from the model. As a global measure of error, the 90th percentile of the residual errors (Mres90) across the whole scan was used. As proof of principle, we demonstrated this approach on both patient-specific cases and a population-based PCA in head and neck (H&N) cancer patients. These models were created using deformation data from deformable registrations between the planning computed tomography and cone-beam computed tomography (CBCTs), and were evaluated against DVFs from registrations of CBCTs not used to create the model.

Results

For our example cases, the oropharyngeal and the nasal cavity regions showed the largest local residual error, indicating the PCA models struggle to predict deformations seen in these regions. Mres90 ranged from 0.4 mm to 6.3 mm across the different models.

Conclusions

A method to quantitatively evaluate how well PCA models represent observed anatomical changes was proposed. We demonstrated our approach on H&N PCA models, but it can be applied to other sites.

A Patients-Based Statistical Model of Radiotherapy Dose Distribution in Nasopharyngeal Cancer

Purpose:

To develop a patients-based statistical model of dose distribution among patients with nasopharyngeal cancer (NPC).

Methods and Materials:

The dose distributions of 75 patients with NPC were acquired and preprocessed to generate a dose-template library. Subsequently, the dominant modes of dose distribution were extracted using principal component analysis (PCA). Leave-one-out cross-validation (LOOCV) was performed for evaluation. Residual reconstruction errors between the doses reconstructed using different dominating eigenvectors and the planned dose distribution were calculated to investigate the convergence characteristics. Three-dimensional Gamma analysis was performed to investigate the accuracy of dose reconstruction.

Results:

The first 29 components contained 90% of the variance in dose distribution, and 45 components accounted for more than 95% of the variance on average. The residual error of the LOOCV model for the cumulative sum of components over all patients decreased from 8.16 to 4.79 Gy when 1 to 74 components were included in the LOOCV model. The 3-dimensional Gamma analysis results implied that the PCA model was capable of dose distribution reconstruction, and the accuracy was especially satisfactory in the high-dose area.

Conclusions:

A PCA-based model of dose distribution variations in patients with NPC was developed, and its accuracy was determined. This model could serve as a predictor of 3-dimensional dose distribution.

A method for acquiring random range uncertainty probability distributions in proton therapy

In treatment planning we depend upon accurate knowledge of geometric and range uncertainties. If the uncertainty model is inaccurate then the plan will produce under-dosing of the target and/or overdosing of OAR. We aim to provide a method for which centre and site-specific population range uncertainty due to inter-fraction motion can be quantified to improve the uncertainty model in proton treatment planning. Daily volumetric MVCT data from previously treated radiotherapy patients has been used to investigate inter-fraction changes to water equivalent path-length (WEPL). Daily image-guidance scans were carried out for each patient and corrected for changes in CTV position (using rigid transformations). An effective depth algorithm was used to determine residual range changes, after corrections had been applied, throughout the treatment by comparing WEPL within the CTV at each fraction for several beam angles. As a proof of principle this method was used to quantify uncertainties for inter-fraction range changes for a sample of head and neck patients of [Formula: see text] mm, [Formula: see text] mm and overall [Formula: see text] mm. For prostate [Formula: see text] mm, [Formula: see text] mm and overall [Formula: see text] mm. The choice of beam angle for head and neck did not affect the inter-fraction range error significantly; however this was not the same for prostate. Greater range changes were seen using a lateral beam compared to an anterior beam for prostate due to relative motion of the prostate and femoral heads. A method has been developed to quantify population range changes due to inter-fraction motion that can be adapted for the clinic. The results of this work highlight the importance of robust planning and analysis in proton therapy. Such information could be used in robust optimisation algorithms or treatment plan robustness analysis. Such knowledge will aid in establishing beam start conditions at planning and for establishing adaptive planning protocols.