Deformable image registration for contour propagation from CT to cone-beam CT scans in radiotherapy of prostate cancer

Evaluation of the geometric and dosimetric accuracies of deformable image registration of targets and critical organs in prostate CBCT-guided adaptive radiotherapy

PURPOSE

Kilovoltage cone beam computed tomography (kVCBCT)-guided adaptive radiation therapy (ART) uses daily deformed CT (dCT), which is generated automatically through deformable registration methods. These registration methods may perform poorly in reproducing volumes of the target organ, rectum, and bladder during treatment. We analyzed the registration errors between the daily kVCBCTs and corresponding dCTs for these organs using the default optical flow algorithm and two registration procedures. We validated the effectiveness of these registration methods in replicating the geometry for dose calculation on kVCBCT for ART.

METHODS

We evaluated three deformable image registration (DIR) methods to assess their registration accuracy and dose calculation effeciency in mapping target and critical organs. The DIR methods include (1) default intensity-based deformable registration, (2) hybrid deformable registration, and (3) a two-step deformable registration process. Each technique was applied to a computerized imaging reference system (CIRS) phantom (Model 062 M) and to five patients who received volumetric modulated arc therapy to the prostate. Registration accuracy was assessed using the 95% Hausdorff distance (HD95) and Dice similarity coefficient (DSC), and each method was compared with the intensity-based registration method. The improvement in the dCT image quality of the CIRS phantom and five patients was assessed by comparing dCT with kVCBCT. Image quality quantitative metrics for the phantom included the signal-to-noise ratio (SNR), uniformity, and contrast-to-noise ratio (CNR), whereas those for the patients included the mean absolute error (MAE), mean error, peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM). To determine dose metric differences, we used a dose-volume histogram (DVH) and 3.0%/0.3 mm gamma analysis to compare planning computed tomography (pCT) and kVCBCT recalculations with restimulated CT images used as a reference.

RESULTS

The dCT images generated by the hybrid (dCTH) and two-step (dCTC) registration methods resulted in significant improvements compared to kVCBCT in the phantom model. Specifically, the SNR improved by 107% and 107.2%, the uniformity improved by 90% and 75%, and the CNR improved by 212.2% and 225.6 for dCTH and dCTC methods, respectively. For the patient images, the MAEs improved by 98% and 94%, the PSNRs improved by 16.3% and 22.9%, and the SSIMs improved by 1% and 1% in the dCTH and dCTC methods, respectively. For the geometric evaluation, only the two-step registration method improved registration accuracy. The dCTH method yielded an average HD95 of 12 mm and average DSC of 0.73, whereas dCTC yielded an average HD95 of 2.9 mm and average DSC of 0.902. The DVH showed that the dCTC-based dose calculations differed by <2% from the expected results for treatment targets and volumes of organs at risk. Additionally, gamma indices for dCTC-based treatment plans were >95% at all points, whereas they were <95% for kVCBCT-based treatment plans.

CONCLUSION

The two-step registration method outperforms the intensity-based and hybrid registration methods. While the hybrid and two-step-based methods improved the image quality of kVCBCT in a linear accelerator, only the two-step method improved the registration accuracy of the corresponding structures among the pCT and kVCBCT datasets. A two-step registration process is recommended for applying kVCBCT to ART, which achieves better registration accuracy for local and global image structures. This method appears to be beneficial for radiotherapy dose calculation in patients with pelvic cancer.

Clinical evaluation of a deep learning CBCT auto-segmentation software for prostate adaptive radiation therapy

Purpose

Aim of the present study is to characterize a deep learning-based auto-segmentation software (DL) for prostate cone beam computed tomography (CBCT) images and to evaluate its applicability in clinical adaptive radiation therapy routine.

Materials and methods

Ten patients, who received exclusive radiation therapy with definitive intent on the prostate gland and seminal vesicles, were selected. Femoral heads, bladder, rectum, prostate, and seminal vesicles were retrospectively contoured by four different expert radiation oncologists on patients CBCT, acquired during treatment. Consensus contours (CC) were generated starting from these data and compared with those created by DL with different algorithms, trained on CBCT (DL-CBCT) or computed tomography (DL-CT). Dice similarity coefficient (DSC), centre of mass (COM) shift and volume relative variation (VRV) were chosen as comparison metrics. Since no tolerance limit can be defined, results were also compared with the inter-operator variability (IOV), using the same metrics.

Results

The best agreement between DL and CC was observed for femoral heads (DSC of 0.96 for both DL-CBCT and DL-CT). Performance worsened for low-contrast soft tissue organs: the worst results were found for seminal vesicles (DSC of 0.70 and 0.59 for DL-CBCT and DL-CT, respectively). The analysis shows that it is appropriate to use algorithms trained on the specific imaging modality. Furthermore, the statistical analysis showed that, for almost all considered structures, there is no significant difference between DL-CBCT and human operator in terms of IOV.

Conclusions

The accuracy of DL-CBCT is in accordance with CC; its use in clinical practice is justified by the comparison with the inter-operator variability.

2D antiscatter grid and scatter sampling based CBCT method for online dose calculations during CBCT guided radiation therapy of pelvis

BACKGROUND

Online dose calculations before the delivery of radiation treatments have applications in dose delivery verification, online adaptation of treatment plans, and simulation-free treatment planning. While dose calculations by directly utilizing CBCT images are desired, dosimetric accuracy can be compromised due to relatively lower HU accuracy in CBCT images.

PURPOSE

In this work, we propose a novel CBCT imaging pipeline to enhance the accuracy of CBCT-based dose calculations in the pelvis region. Our approach aims to improve the HU accuracy in CBCT images, thereby improving the overall accuracy of CBCT-based dose calculations prior to radiation treatment delivery.

METHODS

An in-house developed quantitative CBCT pipeline was implemented to address the CBCT raw data contamination problem. The pipeline combines algorithmic data correction strategies and 2D antiscatter grid-based scatter rejection to achieve high CT number accuracy. To evaluate the effect of the quantitative CBCT pipeline on CBCT-based dose calculations, phantoms mimicking pelvis anatomy were scanned using a linac-mounted CBCT system, and a gold standard multidetector CT used for treatment planning (pCT). A total of 20 intensity-modulated treatment plans were generated for five targets, using 6 and 10 MV flattening filter-free beams, and utilizing small and large pelvis phantom images. For each treatment plan, four different dose calculations were performed using pCT images and three CBCT imaging configurations: quantitative CBCT, clinical CBCT protocol, and a high-performance 1D antiscatter grid (1D ASG). Subsequently, dosimetric accuracy was evaluated for both targets and organs at risk as a function of patient size, target location, beam energy, and CBCT imaging configuration.

RESULTS

When compared to the gold-standard pCT, dosimetric errors in quantitative CBCT-based dose calculations were not significant across all phantom sizes, beam energies, and treatment sites. The largest error observed was 0.6% among all dose volume histogram metrics and evaluated dose calculations. In contrast, dosimetric errors reached up to 7% and 97% in clinical CBCT and high-performance ASG CBCT-based treatment plans, respectively. The largest dosimetric errors were observed in bony targets in the large phantom treated with 6 MV beams. The trends of dosimetric errors in organs at risk were similar to those observed in the targets.

CONCLUSIONS

The proposed quantitative CBCT pipeline has the potential to provide comparable dose calculation accuracy to the gold-standard planning CT in photon radiation therapy for the abdomen and pelvis. These robust dose calculations could eliminate the need for density overrides in CBCT images and enable direct utilization of CBCT images for dose delivery monitoring or online treatment plan adaptations before the delivery of radiation treatments.

Artificial general intelligence for radiation oncology

The emergence of artificial general intelligence (AGI) is transforming radiation oncology. As prominent vanguards of AGI, large language models (LLMs) such as GPT-4 and PaLM 2 can process extensive texts and large vision models (LVMs) such as the Segment Anything Model (SAM) can process extensive imaging data to enhance the efficiency and precision of radiation therapy. This paper explores full-spectrum applications of AGI across radiation oncology including initial consultation, simulation, treatment planning, treatment delivery, treatment verification, and patient follow-up. The fusion of vision data with LLMs also creates powerful multimodal models that elucidate nuanced clinical patterns. Together, AGI promises to catalyze a shift towards data-driven, personalized radiation therapy. However, these models should complement human expertise and care. This paper provides an overview of how AGI can transform radiation oncology to elevate the standard of patient care in radiation oncology, with the key insight being AGI's ability to exploit multimodal clinical data at scale.

Deep-learning based fast and accurate 3D CT deformable image registration in lung cancer

BACKGROUND

Deformable Image Registration (DIR) is an essential technique required in many applications of radiation oncology. However, conventional DIR approaches typically take several minutes to register one pair of 3D CT images and the resulting deformable vector fields (DVFs) are only specific to the pair of images used, making it less appealing for clinical application.

PURPOSE

A deep-learning-based DIR method using CT images is proposed for lung cancer patients to address the common drawbacks of the conventional DIR approaches and in turn can accelerate the speed of related applications, such as contour propagation, dose deformation, adaptive radiotherapy (ART), etc. METHODS: A deep neural network based on VoxelMorph was developed to generate DVFs using CT images collected from 114 lung cancer patients. Two models were trained with the weighted mean absolute error (wMAE) loss and structural similarity index matrix (SSIM) loss (optional) (i.e., the MAE model and the M+S model). In total, 192 pairs of initial CT (iCT) and verification CT (vCT) were included as a training dataset and the other independent 10 pairs of CTs were included as a testing dataset. The vCTs usually were taken 2 weeks after the iCTs. The synthetic CTs (sCTs) were generated by warping the vCTs according to the DVFs generated by the pre-trained model. The image quality of the synthetic CTs was evaluated by measuring the similarity between the iCTs and the sCTs generated by the proposed methods and the conventional DIR approaches, respectively. Per-voxel absolute CT-number-difference volume histogram (CDVH) and MAE were used as the evaluation metrics. The time to generate the sCTs was also recorded and compared quantitatively. Contours were propagated using the derived DVFs and evaluated with SSIM. Forward dose calculations were done on the sCTs and the corresponding iCTs. Dose volume histograms (DVHs) were generated based on dose distributions on both iCTs and sCTs generated by two models, respectively. The clinically relevant DVH indices were derived for comparison. The resulted dose distributions were also compared using 3D Gamma analysis with thresholds of 3 mm/3%/10% and 2 mm/2%/10%, respectively.

RESULTS

The two models (wMAE and M+S) achieved a speed of 263.7±163 / 265.8±190 ms and a MAE of 13.15±3.8 / 17.52±5.8 HU for the testing dataset, respectively. The average SSIM scores of 0.987±0.006 and 0.988±0.004 were achieved by the two proposed models, respectively. For both models, CDVH of a typical patient showed that less than 5% of the voxels had a per-voxel absolute CT-number-difference larger than 55 HU. The dose distribution calculated based on a typical sCT showed differences of ≤2cGy[RBE] for clinical target volume (CTV) D95 and D5 , within ±0.06% for total lung V5 , ≤1.5cGy[RBE] for heart and esophagus Dmean , and ≤6cGy[RBE] for cord Dmax compared to the dose distribution calculated based on the iCT. The good average 3D Gamma passing rates (> 96% for 3 mm/3%/10% and > 94% for 2 mm/2%/10%, respectively) were also observed.

CONCLUSION

A deep neural network-based DIR approach was proposed and has been shown to be reasonably accurate and efficient to register the initial CTs and verification CTs in lung cancer.

Experimental validation of a novel method of dose accumulation for the rectum

BACKGROUND

Dose-surface maps (DSMs) are an increasingly popular tool to evaluate spatial dose-outcome relationships for the rectum. Recently, DSM addition has been proposed as an alternative method of dose accumulation from deformable registration-based techniques. In this study, we performed the first experimental investigation of the accuracy at which DSM accumulation can capture the total dose delivered to a rectum's surface in the presence of inter-fraction motion.

MATERIAL AND METHODS

A custom PVC rectum phantom capable of representing typical rectum inter-fraction motion and filling variations was constructed for this project. The phantom allowed for the placement of EBT3 film sheets on the representative rectum surface to measure rectum surface dose. A multi-fraction prostate VMAT treatment was designed and delivered to the phantom in a water tank for a variety of inter-fraction motion scenarios. DSMs for each fraction were calculated in two ways using CBCT images acquired during delivery and summed to produce accumulated DSMs. Accumulated DSMs were then compared to film measurements using gamma analysis (3%/2 mm criteria). Similarity of isodose clusters between films and DSMs was also investigated.

RESULTS

Baseline agreement between film measurements and accumulated DSMs for a stationary rectum was 95.6%. Agreement between film and accumulated DSMs in the presence of different types of inter.-fraction motion was ≥92%, and isodose cluster mean distance to agreement was within 1.5 mm for most scenarios. Overall, DSM accumulation performed the best when using DSMs that accounted for changes in rectum path orientation.

CONCLUSION

Dose accumulation performed with DSMs was found to accurately replicate total delivered dose to a rectum phantom in the presence of inter-fraction motion.

An Unsupervised Learning-Based Regional Deformable Model for Automated Multi-Organ Contour Propagation

The aim of this study is to evaluate a regional deformable model based on a deep unsupervised learning model for automatic contour propagation in breast cone-beam computed tomography-guided adaptive radiation therapy. A deep unsupervised learning model was introduced to map breast's tumor bed, clinical target volume, heart, left lung, right lung, and spinal cord from planning computed tomography to cone-beam CT. To improve the traditional image registration method's performance, we used a regional deformable framework based on the narrow-band mapping, which can mitigate the effect of the image artifacts on the cone-beam CT. We retrospectively selected 373 anonymized cone-beam CT volumes from 111 patients with breast cancer. The cone-beam CTs are divided into three sets. 311 / 20 / 42 cone-beam CT images were used for training, validating, and testing. The manual contour was used as reference for the testing set. We compared the results between the reference and the model prediction for evaluating the performance. The mean Dice between manual reference segmentations and the model predicted segmentations for breast tumor bed, clinical target volume, heart, left lung, right lung, and spinal cord were 0.78 ± 0.09, 0.90 ± 0.03, 0.88 ± 0.04, 0.94 ± 0.03, 0.95 ± 0.02, and 0.77 ± 0.07, respectively. The results demonstrated a good agreement between the reference and the proposed contours. The proposed deep learning-based regional deformable model technique can automatically propagate contours for breast cancer adaptive radiotherapy. Deep learning in contour propagation was promising, but further investigation was warranted.

Evaluating the accumulated dose distribution of organs at risk in combined radiotherapy for cervical carcinoma based on deformable image registration

OBJECTIVE

To evaluate the feasibility and value of deformable image registration (DIR) in calculating the cumulative doses of organs at risk (OARs) in the combined radiotherapy of cervical cancer.

PATIENTS AND METHODS

Thirty cervical cancer patients treated with external beam radiotherapy (EBRT) combined with intracavitary brachytherapy (ICBT) were reviewed. The simulation CT images of EBRT and ICBT were imported into Varian Velocity 4.1 for the DIR-based dose accumulation. Cumulative dose-volume parameters of D2cc for rectum and bladder were compared between the direct addition (DA) and DIR methods. The quantitative parameters were measured to evaluate the accuracy of DIR.

RESULTS

The three-dimensional cumulative dose distribution of the tumor and OARs were graphically well illustrated by composite isodose lines. In combined EBRT and ICBT, the mean cumulative bladder D2cc calculated by DIR and DA was 86.13 Gy and 86.27 Gy, respectively. The mean cumulative rectal D2cc calculated by DIR and DA was 72.97 Gy and 73.90 Gy, respectively. No significant differences were noted between these two methods (p > 0.05). As to the parameters used to evaluate the DIR accuracy, the mean DSC, Jacobian, MDA (mm) and Hausdorff distance (mm) were 0.79, 1.0, 3.84, and 22.01 respectively for the bladder and 0.53, 1.2, 7.31, and 29.58 respectively for the rectum. In this study, the DSC seemed to be slightly lower compared with previous studies.

CONCLUSION

Dose accumulation based on DIR might be an alternative method to illustrate and evaluate the cumulative doses of the OARs in combined radiotherapy for cervical cancer. However, DIR should be used with caution before overcoming the relevant limitations.

Evaluation of generalization ability for deep learning-based auto-segmentation accuracy in limited field of view CBCT of male pelvic region

PURPOSE

The aim of this study was to evaluate generalization ability of segmentation accuracy for limited FOV CBCT in the male pelvic region using a full-image CNN. Auto-segmentation accuracy was evaluated using various datasets with different intensity distributions and FOV sizes.

METHODS

A total of 171 CBCT datasets from patients with prostate cancer were enrolled. There were 151, 10, and 10 CBCT datasets acquired from Vero4DRT, TrueBeam STx, and Clinac-iX, respectively. The FOV for Vero4DRT, TrueBeam STx, and Clinac-iX was 20, 26, and 25 cm, respectively. The ROIs, including the bladder, prostate, rectum, and seminal vesicles, were manually delineated. The U² -Net CNN network architecture was used to train the segmentation model. A total of 131 limited FOV CBCT datasets from Vero4DRT were used for training (104 datasets) and validation (27 datasets); thereafter the rest were for testing. The training routine was set to save the best weight values when the DSC in the validation set was maximized. Segmentation accuracy was qualitatively and quantitatively evaluated between the ground truth and predicted ROIs in the different testing datasets.

RESULTS

The mean scores ± standard deviation of visual evaluation for bladder, prostate, rectum, and seminal vesicle in all treatment machines were 1.0 ± 0.7, 1.5 ± 0.6, 1.4 ± 0.6, and 2.1 ± 0.8 points, respectively. The median DSC values for all imaging devices were ≥0.94 for the bladder, 0.84-0.87 for the prostate and rectum, and 0.48-0.69 for the seminal vesicles. Although the DSC values for the bladder and seminal vesicles were significantly different among the three imaging devices, the DSC value of the bladder changed by less than 1% point. The median MSD values for all imaging devices were ≤1.2 mm for the bladder and 1.4-2.2 mm for the prostate, rectum, and seminal vesicles. The MSD values for the seminal vesicles were significantly different between the three imaging devices.

CONCLUSION

The proposed method is effective for testing datasets with different intensity distributions and FOV from training datasets.

Dosimetric impact of deformable image registration using radiophotoluminescent glass dosimeters with a physical geometric phantom

PURPOSE

To study the dosimetry impact of deformable image registration (DIR) using radiophotoluminescent glass dosimeter (RPLD) and custom developed phantom with various inserts.

METHODS

The phantom was developed to facilitate simultaneous evaluation of geometric and dosimetric accuracy of DIR. Four computed tomography (CT) images of the phantom were acquired with four different configurations. Four volumetric modulated arc therapy (VMAT) plans were computed for different phantom. Two different patterns were applied to combination of four phantom configurations. RPLD dose measurement was combined between corresponding two phantom configurations. DIR-based dose accumulation was calculated between corresponding two CT images with two commercial DIR software and various DIR parameter settings, and an open source software. Accumulated dose calculated using DIR was then compared with measured dose using RPLD.

RESULTS

The mean ± standard deviation (SD) of dose difference was 2.71 ± 0.23% (range, 2.22%-3.01%) for tumor-proxy and 3.74 ± 0.79% (range, 1.56%-4.83%) for rectum-proxy. The mean ± SD of target registration error (TRE) was 1.66 ± 1.36 mm (range, 0.03-4.43 mm) for tumor-proxy and 6.87 ± 5.49 mm (range, 0.54-17.47 mm) for rectum-proxy. These results suggested that DIR accuracy had wide range among DIR parameter setting.

CONCLUSIONS

The dose difference observed in our study was 3% for tumor-proxy and within 5% for rectum-proxy. The custom developed physical phantom with inserts showed potential for accurate evaluation of DIR-based dose accumulation. The prospect of simultaneous evaluation of geometric and dosimetric DIR accuracy in a single phantom may be useful for validation of DIR for clinical use.

Deep learning-based segmentation in prostate radiation therapy using Monte Carlo simulated cone-beam computed tomography

PURPOSE

Segmenting organs in cone-beam CT (CBCT) images would allow to adapt the radiotherapy based on the organ deformations that may occur between treatment fractions. However, this is a difficult task because of the relative lack of contrast in CBCT images, leading to high inter-observer variability. Deformable image registration (DIR) and deep-learning based automatic segmentation approaches have shown interesting results for this task in the past years. However, they are either sensitive to large organ deformations, or require to train a convolutional neural network (CNN) from a database of delineated CBCT images, which is difficult to do without improvement of image quality. In this work, we propose an alternative approach: to train a CNN (using a deep learning-based segmentation tool called nnU-Net) from a database of artificial CBCT images simulated from planning CT, for which it is easier to obtain the organ contours.

METHODS

Pseudo-CBCT (pCBCT) images were simulated from readily available segmented planning CT images, using the GATE Monte Carlo simulation. CT reference delineations were copied onto the pCBCT, resulting in a database of segmented images used to train the neural network. The studied segmentation contours were: bladder, rectum, and prostate contours. We trained multiple nnU-Net models using different training: (1) segmented real CBCT, (2) pCBCT, (3) segmented real CT and tested on pseudo-CT (pCT) generated from CBCT with cycleGAN, and (4) a combination of (2) and (3). The evaluation was performed on different datasets of segmented CBCT or pCT by comparing predicted segmentations with reference ones thanks to Dice similarity score and Hausdorff distance. A qualitative evaluation was also performed to compare DIR-based and nnU-Net-based segmentations.

RESULTS

Training with pCBCT was found to lead to comparable results to using real CBCT images. When evaluated on CBCT obtained from the same hospital as the CT images used in the simulation of the pCBCT, the model trained with pCBCT scored mean DSCs of 0.92 ± 0.05, 0.87 ± 0.02, and 0.85 ± 0.04 and mean Hausdorff distance 4.67 ± 3.01, 3.91 ± 0.98, and 5.00 ± 1.32 for the bladder, rectum, and prostate contours respectively, while the model trained with real CBCT scored mean DSCs of 0.91 ± 0.06, 0.83 ± 0.07, and 0.81 ± 0.05 and mean Hausdorff distance 5.62 ± 3.24, 6.43 ± 5.11, and 6.19 ± 1.14 for the bladder, rectum, and prostate contours, respectively. It was also found to outperform models using pCT or a combination of both, except for the prostate contour when tested on a dataset from a different hospital. Moreover, the resulting segmentations demonstrated a clinical acceptability, where 78% of bladder segmentations, 98% of rectum segmentations, and 93% of prostate segmentations required minor or no corrections, and for 76% of the patients, all structures of the patient required minor or no corrections.

CONCLUSION

We proposed to use simulated CBCT images to train a nnU-Net segmentation model, avoiding the need to gather complex and time-consuming reference delineations on CBCT images.

Evaluation of DIR algorithm performance in real patients for radiotherapy treatments: A systematic review of operator-dependent strategies

PURPOSE

The performance of deformable medical image registration (DIR) algorithms has become a major concern.

METHODS

We aimed to obtain updated information on DIR algorithm performance quantification through a literature review of articles published between 2010 and 2022. We focused only on studies using operator-based methods to treat real patients. The PubMed, Google Scholar and Embase databases were searched following PRISMA guidelines.

RESULTS

One hundred and seven articles were identified. The mean number of patients and registrations per publication was 20 and 63, respectively. We found 23 different geometric metrics appearing at least twice, and the dosimetric impact of DIR was quantified in 32 articles. Forty-eight different at-risk organs were described, and target volumes were studied in 43 publications. Prostate, head-and-neck and thoracic locations represented more than ¾ of the studied locations. We summarized the type of DIR and the images used, and other key elements. Intra/interobserver variability, threshold values and the correlation between metrics were also discussed.

CONCLUSIONS

This literature review covers the past decade and should facilitate the implementation of DIR algorithms in clinical practice by providing practical and pertinent information to quantify their performance on real patients.

Assessment of CT to CBCT contour mapping for radiomic feature analysis in prostate cancer

This study provides a quantitative assessment of the accuracy of a commercially available deformable image registration (DIR) algorithm to automatically generate prostate contours and additionally investigates the robustness of radiomic features to differing contours. Twenty-eight prostate cancer patients enrolled on an institutional review board (IRB) approved protocol were selected. Planning CTs (pCTs) were deformably registered to daily cone-beam CTs (CBCTs) to generate prostate contours (auto contours). The prostate contours were also manually drawn by a physician. Quantitative assessment of deformed versus manually drawn prostate contours on daily CBCT images was performed using Dice similarity coefficient (DSC), mean distance-to-agreement (MDA), difference in center-of-mass position (ΔCM) and difference in volume (ΔVol). Radiomic features from 6 classes were extracted from each contour. Lin’s concordance correlation coefficient (CCC) and mean absolute percent difference in radiomic feature-derived data (mean |%Δ|RF) between auto and manual contours were calculated. The mean (± SD) DSC, MDA, ΔCM and ΔVol between the auto and manual prostate contours were 0.90 ± 0.04, 1.81 ± 0.47 mm, 2.17 ± 1.26 mm and 5.1 ± 4.1% respectively. Of the 1,010 fractions under consideration, 94.8% of DIRs were within TG-132 recommended tolerance. 30 radiomic features had a CCC > 0.90 and 21 had a mean |%∆|RF < 5%. Auto-propagation of prostate contours resulted in nearly 95% of DIRs within tolerance recommendations of TG-132, leading to the majority of features being regarded as acceptably robust. The use of auto contours for radiomic feature analysis is promising but must be done with caution.

Møller D, Assenholt MS, Hansen R, Nyvang L, Ravkilde T, Thomsen MS, Hoffmann L. Density calibrated cone beam CT as a tool for adaptive radiotherapy

BACKGROUND

Visual inspections of anatomical changes observed on daily cone-beam CT (CBCT) images are often used as triggers for radiotherapy plan adaptation to avoid unacceptable dose levels to the target or OARs. Direct CBCT dose calculations would improve the ability to adapt only those plans where dosimetric changes are observed. This study investigates the accuracy of dose calculations on CBCTs.

MATERIALS AND METHODS

Calibration curves were obtained for CBCT imagers at nine identical accelerators. CBCT scans of a phantom with different density inserts were recorded for two scan modes (Head-Neck and Pelvis) and mean calibration curves were calculated. Subsequently, CBCT scans of the phantom with six different density inserts were recorded, the dose distributions on the CBCTs were calculated and compared to dose on the planning CT (pCT). The uncertainty was quantified by the dosimetric difference between the pCT and the CBCT. The two mean calibration curves were used to calculate the daily delivered CBCT dose for ten Head-Neck-, eleven Lung-, and ten pelvic patients. Additional patient calculations were performed using low-HU empirically corrected calibration curves. Patient doses were compared on target coverage and mean dose, and D1cc for OARs.

RESULTS

The dose differences between pCT and CBCT for phantom data were small for all DVH parameters, with mean deviations below ±0.6% for both CBCT modes. For patient data, it was found that low-HU corrected calibration curves performed the best. The mean deviations for the mean dose and coverage of the target were 0.2%±0.7% and 0.1%±0.6%, across all patient groups.

CONCLUSION

Dose calculation on CBCT images results in target coverage and mean dose with an accuracy of the order of 1%, which makes this acceptable for clinical use. The CBCT mode specific calibration curves can be used at all identical imaging devices and for all patient groups.

Analysis of the Influence of Peripheral Anatomical Changes for CBCT-Guided Prostate Cancer Radiotherapy

Purpose:

To analyze the influence of the bladder and rectum filling and the body contour changes on the prostate target dose.

Methods:

A total of 190 cone-beam CT (CBCT) image data sets from 16 patients with prostate cancer were used in this study. Dose reconstruction was performed on the virtual CT generated by the deformable planning CT. Then, the effects of the bladder filling, rectal filling, and the patient’s body contour changes of the PCTV1 (the prostate area, B1) and PCTV2 (the seminal vesicle area, B2) on the target dose were analyzed. Correlation analysis was performed for the ratio of bladder and rectal volume variation and the variation of the bladder and rectal dose.

Results:

The mean Dice coefficients of B1, B2, bladder, and rectum were 0.979, 0.975, 0.888 and 0.827, respectively, and the mean Hausdorff distances were 0.633, 1.505, 2.075, and 1.533, respectively. With the maximum volume variations of 142.04 ml for the bladder and 40.50 ml for the rectum, the changes of V₁₀₀, V₉₅, D₂, and D₉₈ were 1.739 ± 1.762 (%), 0.066 ± 0.169 (%), 0.562 ± 0.442 (%), and 0.496 ± 0.479 (%) in PCTV1 and 1.686 ± 1.051 (%), 0.240 ± 0.215 (%), 1.123 ± 0.925 (%), and 0.924 ± 0.662 (%) in PCTV2, respectively. With a 10% increase in the volume of the bladder and rectum, the V₇₅, V₇₀, and V₆₅ of rectum increased at 0.73 (%), 0.71 (%), and 1.18 (%), and the V₇₅, V₇₀, and V₆₅ of bladder changed at −0.21 (%), −0.32 (%), and −0.39 (%), respectively.

Conclusion:

Significant correlations were observed between the volume variation and the dose variation of the bladder and rectum. However, when a bladder and rectal filling protocol was adopted, the target dose coverage can be effectively ensured based on CBCT guidance to correct the prostate target position.

MR to ultrasound image registration with segmentation-based learning for HDR prostate brachytherapy

PURPOSE

Propagation of contours from high-quality magnetic resonance (MR) images to treatment planning ultrasound (US) images with severe needle artifacts is a challenging task, which can greatly aid the organ contouring in high dose rate (HDR) prostate brachytherapy. In this study, a deep learning approach was developed to automatize this registration procedure for HDR brachytherapy practice.

METHODS

Because of the lack of training labels and difficulty of accurate registration from inferior image quality, a new segmentation-based registration framework was proposed for this multi-modality image registration problem. The framework consisted of two segmentation networks and a deformable registration network, based on the weakly -supervised registration strategy. Specifically, two 3D V-Nets were trained for the prostate segmentation on the MR and US images separately, to generate the weak supervision labels for the registration network training. Besides the image pair, the corresponding prostate probability maps from the segmentation were further fed to the registration network to predict the deformation matrix, and an augmentation method was designed to randomly scale the input and label probability maps during the registration network training. The overlap between the deformed and fixed prostate contours was analyzed to evaluate the registration accuracy. Three datasets were collected from our institution for the MR and US image segmentation networks, and the registration network learning, which contained 121, 104, and 63 patient cases, respectively.

RESULTS

The mean Dice similarity coefficient (DSC) results of the two prostate segmentation networks are 0.86 ± 0.05 and 0.90 ± 0.03, for MR images and the US images after the needle insertion, respectively. The mean DSC, center-of-mass (COM) distance, Hausdorff distance (HD), and averaged symmetric surface distance (ASSD) results for the registration of manual prostate contours were 0.87 ± 0.05, 1.70 ± 0.89 mm, 7.21 ± 2.07 mm, 1.61 ± 0.64 mm, respectively. By providing the prostate probability map from the segmentation to the registration network, as well as applying the random map augmentation method, the evaluation results of the four metrics were all improved, such as an increase in DSC from 0.83 ± 0.08 to 0.86 ± 0.06 and from 0.86 ± 0.06 to 0.87 ± 0.05, respectively.

CONCLUSIONS

A novel segmentation-based registration framework was proposed to automatically register prostate MR images to the treatment planning US images with metal artifacts, which not only largely saved the labor work on the data preparation, but also improved the registration accuracy. The evaluation results showed the potential of this approach in HDR prostate brachytherapy practice.

Automated contour propagation of the prostate from pCT to CBCT images via deep unsupervised learning

PURPOSE

To develop and evaluate a deep unsupervised learning (DUL) framework based on a regional deformable model for automated prostate contour propagation from planning computed tomography (pCT) to cone-beam CT (CBCT).

METHODS

We introduce a DUL model to map the prostate contour from pCT to on-treatment CBCT. The DUL framework used a regional deformable model via narrow-band mapping to augment the conventional strategy. Two hundred and fifty-one anonymized CBCT images from prostate cancer patients were retrospectively selected and divided into three sets: 180 were used for training, 12 for validation, and 59 for testing. The testing dataset was divided into two groups. Group 1 contained 50 CBCT volumes, with one physician-generated prostate contour on CBCT image. Group 2 contained nine CBCT images, each including prostate contours delineated by four independent physicians and a consensus contour generated using the STAPLE method. Results were compared between the proposed DUL and physician-generated contours through the Dice similarity coefficients (DSCs), the Hausdorff distances, and the distances of the center-of-mass.

RESULTS

The average DSCs between DUL-based prostate contours and reference contours for test data in group 1 and group 2 consensus were 0.83 ± 0.04, and 0.85 ± 0.04, respectively. Correspondingly, the mean center-of-mass distances were 3.52 mm ± 1.15 mm, and 2.98 mm ± 1.42 mm, respectively.

CONCLUSIONS

This novel DUL technique can automatically propagate the contour of the prostate from pCT to CBCT. The proposed method shows that highly accurate contour propagation for CBCT-guided adaptive radiotherapy is achievable via the deep learning technique.

Domain adversarial networks and intensity-based data augmentation for male pelvic organ segmentation in cone beam CT

In radiation therapy, a CT image is used to manually delineate the organs and plan the treatment. During the treatment, a cone beam CT (CBCT) is often acquired to monitor the anatomical modifications. For this purpose, automatic organ segmentation on CBCT is a crucial step. However, manual segmentations on CBCT are scarce, and models trained with CT data do not generalize well to CBCT images. We investigate adversarial networks and intensity-based data augmentation, two strategies leveraging large databases of annotated CTs to train neural networks for segmentation on CBCT. Adversarial networks consist of a 3D U-Net segmenter and a domain classifier. The proposed framework is aimed at encouraging the learning of filters producing more accurate segmentations on CBCT. Intensity-based data augmentation consists in modifying the training CT images to reduce the gap between CT and CBCT distributions. The proposed adversarial networks reach DSCs of 0.787, 0.447, and 0.660 for the bladder, rectum, and prostate respectively, which is an improvement over the DSCs of 0.749, 0.179, and 0.629 for "source only" training. Our brightness-based data augmentation reaches DSCs of 0.837, 0.701, and 0.734, which outperforms the morphons registration algorithms for the bladder (0.813) and rectum (0.653), while performing similarly on the prostate (0.731). The proposed adversarial training framework can be used for any segmentation application where training and test distributions differ. Our intensity-based data augmentation can be used for CBCT segmentation to help achieve the prescribed dose on target and lower the dose delivered to healthy organs.

Deep learning-based medical image segmentation with limited labels

Deep learning (DL)-based auto-segmentation has the potential for accurate organ delineation in radiotherapy applications but requires large amounts of clean labeled data to train a robust model. However, annotating medical images is extremely time-consuming and requires clinical expertise, especially for segmentation that demands voxel-wise labels. On the other hand, medical images without annotations are abundant and highly accessible. To alleviate the influence of the limited number of clean labels, we propose a weakly supervised DL training approach using deformable image registration (DIR)-based annotations, leveraging the abundance of unlabeled data. We generate pseudo-contours by utilizing DIR to propagate atlas contours onto abundant unlabeled images and train a robust DL-based segmentation model. With 10 labeled TCIA dataset and 50 unlabeled CT scans from our institution, our model achieved Dice similarity coefficient of 87.9%, 73.4%, 73.4%, 63.2% and 61.0% on mandible, left & right parotid glands and left & right submandibular glands of TCIA test set and competitive performance on our institutional clinical dataset and a third party (PDDCA) dataset. Experimental results demonstrated the proposed method outperformed traditional multi-atlas DIR methods and fully supervised limited data training and is promising for DL-based medical image segmentation application with limited annotated data.

Evaluation of an a priori scatter correction algorithm for cone-beam computed tomography based range and dose calculations in proton therapy

BACKGROUND AND PURPOSE

Scatter correction of cone-beam computed tomography (CBCT) projections may enable accurate online dose-delivery estimations in photon and proton-based radiotherapy. This study aimed to evaluate the impact of scatter correction in CBCT-based proton range/dose calculations, in scans acquired in both proton and photon gantries.

MATERIAL AND METHODS

CBCT projections of a Catphan and an Alderson phantom were acquired on both a proton and a photon gantry. The scatter corrected CBCTs (corrCBCTs) and the clinical reconstructions (stdCBCTs) were compared against CTs rigidly registered to the CBCTs (rigidCTs). The CBCTs of the Catphan phantom were segmented by materials for CT number analysis. Water equivalent path length (WEPL) maps were calculated through the Alderson phantom while proton plans optimized on the rigidCT and recalculated on all CBCTs were compared in a gamma analysis.

RESULTS

In medium and high-density materials, the corrCBCT CT numbers were much closer to those of the rigidCT than the stdCBCTs. E.g. in the 50% bone segmentations the differences were reduced from above 300 HU (with stdCBCT) to around 60-70 HU (with corrCBCT). Differences in WEPL from the rigidCT were typically well below 5 mm for the corrCBCTs, compared to well above 10 mm for the stdCBCTs with the largest deviations in the head and thorax regions. Gamma pass rates (2%/2mm) when comparing CBCT-based dose re-calculations to rigidCT calculations were improved from around 80% (with stdCBCT) to mostly above 90% (with corrCBCT).

CONCLUSION

Scatter correction leads to substantial artefact reductions, improving accuracy of CBCT-based proton range/dose calculations.

Computed Tomography/Magnetic Resonance Imaging (CT/MRI) Image Registration and Fusion Assessment for Accurate Glioblastoma Radiotherapy Treatment Planning

Background: In this study, computed tomography/magnetic resonance imaging (CT/MRI) image registration and fusion in the 3D conformal radiotherapy treatment planning of Glioblastoma brain tumor was investigated. Good CT/MRI image registration and fusion made a great impact on dose calculation and treatment planning accuracy. Indeed, the uncertainly associated with the registration and fusion methods must be well verified and communicated. Unfortunately, there is no standard procedure or mathematical formalism to perform this verification due to noise, distortion, and complicated anatomical situations. Objectives: This study aimed at assessing the effective contribution of MRI in Glioma radiotherapy treatment by improving the localization of target volumes and organs at risk (OARs). It is also a question to provide clinicians with some suitable metrics to evaluate the CT/MRI image registration and fusion results. Methods: Quantitative image registration and fusion evaluation were used in this study to compare Eclipse TPS tools and Elastix CT/MRI image registration fusion. Thus, Dice score coefficient (DSC), Jaccard similarity coefficient (JSC), and Hausdorff distance (HD) were found to be suitable metrics for the evaluation and comparison of the image registration and fusion methods of Eclipse TPS and Elastix. Results: The programmed tumor’s volumes (PTV) delineated on CT slices were approximately 1.38 times smaller than those delineated on CT/MRI fused images. Large differences were observed for the edema and the brainstem. It was also found that MRI considerably optimized the dose to be delivered to the optic nerve and brainstem. Conclusions: Image registration and fusion is a fundamental step for suitable and efficient Glioma treatment planning in 3D conformal radiotherapy that ensure accurate dose delivery and unnecessary OAR irradiation. MRI can provide accurate localization of targeted volumes leading to better irradiation control of Glioma tumor.

Pelvic multi-organ segmentation on cone-beam CT for prostate adaptive radiotherapy

BACKGROUND AND PURPOSE

The purpose of this study is to develop a deep learning-based approach to simultaneously segment five pelvic organs including prostate, bladder, rectum, left and right femoral heads on cone-beam CT (CBCT), as required elements for prostate adaptive radiotherapy planning.

MATERIALS AND METHODS

We propose to utilize both CBCT and CBCT-based synthetic MRI (sMRI) for the segmentation of soft tissue and bony structures, as they provide complementary information for pelvic organ segmentation. CBCT images have superior bony structure contrast and sMRIs have superior soft tissue contrast. Prior to segmentation, sMRI was generated using a cycle-consistent adversarial networks (CycleGAN), which was trained using paired CBCT-MR images. To combine the advantages of both CBCT and sMRI, we developed a cross-modality attention pyramid network with late feature fusion. Our method processes CBCT and sMRI inputs separately to extract CBCT-specific and sMRI-specific features prior to combining them in a late-fusion network for final segmentation. The network was trained and tested using 100 patients' datasets, with each dataset including the CBCT and manual physician contours. For comparison, we trained another two networks with different network inputs and architectures. The segmentation results were compared to manual contours for evaluations.

RESULTS

For the proposed method, dice similarity coefficients and mean surface distances between the segmentation results and the ground truth were 0.96 ± 0.03, 0.65 ± 0.67 mm; 0.91 ± 0.08, 0.93 ± 0.96 mm; 0.93 ± 0.04, 0.72 ± 0.61 mm; 0.95 ± 0.05, 1.05 ± 1.40 mm; and 0.95 ± 0.05, 1.08 ± 1.48 mm for bladder, prostate, rectum, left and right femoral heads, respectively. As compared to the other two competing methods, our method has shown superior performance in terms of the segmentation accuracy.

CONCLUSION

We developed a deep learning-based segmentation method to rapidly and accurately segment five pelvic organs simultaneously from daily CBCTs. The proposed method could be used in the clinic to support rapid target and organs-at-risk contouring for prostate adaptive radiation therapy.

Dosimetric impact of organ at risk daily variation during prostate stereotactic ablative radiotherapy

OBJECTIVE

Prostate stereotactic ablative radiotherapy (SABR) delivers large doses using a fast dose rate. This amplifies the effect geometric uncertainties have on normal tissue dose. The aim of this study was to determine whether the treatment dose-volume histogram (DVH) agrees with the planned dose to organs at risk (OAR).

METHODS

41 low-intermediate risk prostate cancer patients were treated with SABR using a linac based technique. Dose prescribed was 35 Gy in five fractions delivered on alternate days, planned using volumetric modulated arc therapy (VMAT) with 10X flattening filter free (FFF). On treatment, prostate was matched to fiducial markers on cone beam CT (CBCT). OAR were retrospectively delineated on 205 pre-treatment CBCT images. Daily CBCT contours were overlaid on the planning CT for dosimetric analysis. Verification plan used to evaluate the daily DVH for each structure. The daily doses received by OAR were recorded using the D%.

RESULTS

The median rectum and bladder volumes at planning were 67.1 cm3 (interquartile range 56.4-78.2) and 164.4 cm3 (interquartile range 120.3-213.4) respectively. There was no statistically significant difference in median rectal volume at each of the five treatment scans compared to the planning scan (p = 0.99). This was also the case for median bladder volume (p = 0.79). The median dose received by rectum and bladder at each fraction was higher than planned, at the majority of dose levels. For rectum the increase ranged from 0.78-1.64Gy and for bladder 0.14-1.07Gy. The percentage of patients failing for rectum D35% < 18 Gy (p = 0.016), D10% < 28 Gy (p = 0.004), D5% < 32 Gy (p = 0.0001), D1% < 35 Gy (p = 0.0001) and bladder D1% < 35 Gy (p = 0.001) at treatment were all statistically significant.

CONCLUSION

In this cohort of prostate SABR patients, we estimate the OAR treatment DVH was higher than planned. This was due to rectal and bladder organ variation.

ADVANCES IN KNOWLEDGE

OAR variation in prostate SABR using a FFF technique, may cause the treatment DVH to be higher than planned.

Cross-Domain Data Augmentation for Deep-Learning-Based Male Pelvic Organ Segmentation in Cone Beam CT

For prostate cancer patients, large organ deformations occurring between radiotherapy treatment sessions create uncertainty about the doses delivered to the tumor and surrounding healthy organs. Segmenting those regions on cone beam CT (CBCT) scans acquired on treatment day would reduce such uncertainties. In this work, a 3D U-net deep-learning architecture was trained to segment bladder, rectum, and prostate on CBCT scans. Due to the scarcity of contoured CBCT scans, the training set was augmented with CT scans already contoured in the current clinical workflow. Our network was then tested on 63 CBCT scans. The Dice similarity coefficient (DSC) increased significantly with the number of CBCT and CT scans in the training set, reaching 0.874 ± 0.096 , 0.814 ± 0.055 , and 0.758 ± 0.101 for bladder, rectum, and prostate, respectively. This was about 10% better than conventional approaches based on deformable image registration between planning CT and treatment CBCT scans, except for prostate. Interestingly, adding 74 CT scans to the CBCT training set allowed maintaining high DSCs, while halving the number of CBCT scans. Hence, our work showed that although CBCT scans included artifacts, cross-domain augmentation of the training set was effective and could rely on large datasets available for planning CT scans.

Accuracy of automatic deformable structure propagation for high-field MRI guided prostate radiotherapy

BACKGROUND

In this study we have evaluated the accuracy of automatic, deformable structure propagation from planning CT and MR scans for daily online plan adaptation for MR linac (MRL) treatment, which is an important element to minimize re-planning time and reduce the risk of misrepresenting the target due to this time pressure.

METHODS

For 12 high-risk prostate cancer patients treated to the prostate and pelvic lymph nodes, target structures and organs at risk were delineated on both planning MR and CT scans and propagated using deformable registration to three T2 weighted MR scans acquired during the treatment course. Generated structures were evaluated against manual delineations on the repeated scans using intra-observer variation obtained on the planning MR as ground truth.

RESULTS

MR-to-MR propagated structures had significant less median surface distance and larger Dice similarity index compared to CT-MR propagation. The MR-MR propagation uncertainty was similar in magnitude to the intra-observer variation. Visual inspection of the deformed structures revealed that small anatomical differences between organs in source and destination image sets were generally well accounted for while large differences were not.

CONCLUSION

Both CT and MR based propagations require manual editing, but the current results show that MR-to-MR propagated structures require fewer corrections for high risk prostate cancer patients treated at a high-field MRL.

Male pelvic multi-organ segmentation aided by CBCT-based synthetic MRI

To develop an automated cone-beam computed tomography (CBCT) multi-organ segmentation method for potential CBCT-guided adaptive radiation therapy workflow. The proposed method combines the deep leaning-based image synthesis method, which generates magnetic resonance images (MRIs) with superior soft-tissue contrast from on-board setup CBCT images to aid CBCT segmentation, with a deep attention strategy, which focuses on learning discriminative features for differentiating organ margins. The whole segmentation method consists of 3 major steps. First, a cycle-consistent adversarial network (CycleGAN) was used to estimate a synthetic MRI (sMRI) from CBCT images. Second, a deep attention network was trained based on sMRI and its corresponding manual contours. Third, the segmented contours for a query patient was obtained by feeding the patient's CBCT images into the trained sMRI estimation and segmentation model. In our retrospective study, we included 100 prostate cancer patients, each of whom has CBCT acquired with prostate, bladder and rectum contoured by physicians with MRI guidance as ground truth. We trained and tested our model with separate datasets among these patients. The resulting segmentations were compared with physicians' manual contours. The Dice similarity coefficient and mean surface distance indices between our segmented and physicians' manual contours (bladder, prostate, and rectum) were 0.95  ±  0.02, 0.44  ±  0.22 mm, 0.86  ±  0.06, 0.73  ±  0.37 mm, and 0.91  ±  0.04, 0.72  ±  0.65 mm, respectively. We have proposed a novel CBCT-only pelvic multi-organ segmentation strategy using CBCT-based sMRI and validated its accuracy against manual contours. This technique could provide accurate organ volume for treatment planning without requiring MR images acquisition, greatly facilitating routine clinical workflow.

Advanced 4-dimensional cone-beam computed tomography reconstruction by combining motion estimation, motion-compensated reconstruction, biomechanical modeling and deep learning

4-Dimensional cone-beam computed tomography (4D-CBCT) offers several key advantages over conventional 3D-CBCT in moving target localization/delineation, structure de-blurring, target motion tracking, treatment dose accumulation and adaptive radiation therapy. However, the use of the 4D-CBCT in current radiation therapy practices has been limited, mostly due to its sub-optimal image quality from limited angular sampling of cone-beam projections. In this study, we summarized the recent developments of 4D-CBCT reconstruction techniques for image quality improvement, and introduced our developments of a new 4D-CBCT reconstruction technique which features simultaneous motion estimation and image reconstruction (SMEIR). Based on the original SMEIR scheme, biomechanical modeling-guided SMEIR (SMEIR-Bio) was introduced to further improve the reconstruction accuracy of fine details in lung 4D-CBCTs. To improve the efficiency of reconstruction, we recently developed a U-net-based deformation-vector-field (DVF) optimization technique to leverage a population-based deep learning scheme to improve the accuracy of intra-lung DVFs (SMEIR-Unet), without explicit biomechanical modeling. Details of each of the SMEIR, SMEIR-Bio and SMEIR-Unet techniques were included in this study, along with the corresponding results comparing the reconstruction accuracy in terms of CBCT images and the DVFs. We also discussed the application prospects of the SMEIR-type techniques in image-guided radiation therapy and adaptive radiation therapy, and presented potential schemes on future developments to achieve faster and more accurate 4D-CBCT imaging.

Robust contour propagation using deep learning and image registration for online adaptive proton therapy of prostate cancer

Purpose

To develop and validate a robust and accurate registration pipeline for automatic contour propagation for online adaptive Intensity‐Modulated Proton Therapy (IMPT) of prostate cancer using elastix software and deep learning.

Methods

A three‐dimensional (3D) Convolutional Neural Network was trained for automatic bladder segmentation of the computed tomography (CT) scans. The automatic bladder segmentation alongside the computed tomography (CT) scan is jointly optimized to add explicit knowledge about the underlying anatomy to the registration algorithm. We included three datasets from different institutes and CT manufacturers. The first was used for training and testing the ConvNet, where the second and the third were used for evaluation of the proposed pipeline. The system performance was quantified geometrically using the dice similarity coefficient (DSC), the mean surface distance (MSD), and the 95% Hausdorff distance (HD). The propagated contours were validated clinically through generating the associated IMPT plans and compare it with the IMPT plans based on the manual delineations. Propagated contours were considered clinically acceptable if their treatment plans met the dosimetric coverage constraints on the manual contours.

Results

The bladder segmentation network achieved a DSC of 88% and 82% on the test datasets. The proposed registration pipeline achieved a MSD of 1.29 ± 0.39, 1.48 ± 1.16, and 1.49 ± 0.44 mm for the prostate, seminal vesicles, and lymph nodes, respectively, on the second dataset and a MSD of 2.31 ± 1.92 and 1.76 ± 1.39 mm for the prostate and seminal vesicles on the third dataset. The automatically propagated contours met the dose coverage constraints in 86%, 91%, and 99% of the cases for the prostate, seminal vesicles, and lymph nodes, respectively. A Conservative Success Rate (CSR) of 80% was obtained, compared to 65% when only using intensity‐based registration.

Conclusion

The proposed registration pipeline obtained highly promising results for generating treatment plans adapted to the daily anatomy. With 80% of the automatically generated treatment plans directly usable without manual correction, a substantial improvement in system robustness was reached compared to a previous approach. The proposed method therefore facilitates more precise proton therapy of prostate cancer, potentially leading to fewer treatment‐related adverse side effects.

Enhancing liver tumor localization accuracy by prior-knowledge-guided motion modeling and a biomechanical model

BACKGROUND

Pre-treatment liver tumor localization remains a challenging task for radiation therapy, mostly due to the limited tumor contrast against normal liver tissues, and the respiration-induced liver tumor motion. Recently, we developed a biomechanical modeling-based, deformation-driven cone-beam CT estimation technique (Bio-CBCT), which achieved substantially improved accuracy on low-contrast liver tumor localization. However, the accuracy of Bio-CBCT is still affected by the limited tissue contrast around the caudal liver boundary, which reduces the accuracy of the boundary condition that is fed into the biomechanical modeling process. In this study, we developed a motion modeling and biomechanical modeling-guided CBCT estimation technique (MM-Bio-CBCT), to further improve the liver tumor localization accuracy by incorporating a motion model into the CBCT estimation process.

METHODS

MM-Bio-CBCT estimates new CBCT images through deforming a prior high-quality CT or CBCT volume. The deformation vector field (DVF) is solved by iteratively matching the digitally-reconstructed-radiographs (DRRs) of the deformed prior image to the acquired 2D cone-beam projections. Using the same solved DVF, the liver tumor volume contoured on the prior image can be transferred onto the new CBCT image for automatic tumor localization. To maximize the accuracy of the solved DVF, MM-Bio-CBCT employs two strategies for additional DVF optimization: (I) prior-knowledge-guided liver boundary motion modeling with motion patterns extracted from a prior 4D imaging set like 4D-CTs/4D-CBCTs, to improve the liver boundary DVF accuracy; and (II) finite-element-analysis-based biomechanical modeling of the liver volume to improve the intra-liver DVF accuracy. We evaluated the accuracy of MM-Bio-CBCT on both the digital extended-cardiac-torso (XCAT) phantom images and real liver patient images. The liver tumor localization accuracy of MM-Bio-CBCT was evaluated and compared with that of the purely intensity-driven 2D-3D deformation technique, the 2D-3D deformation technique with motion modeling, and the Bio-CBCT technique. Metrics including the DICE coefficient and the center-of-mass-error (COME) were assessed for quantitative evaluation.

RESULTS

Using limited-view 20 projections for CBCT estimation, the average (± SD) DICE coefficients between the estimated and the 'gold-standard' liver tumors of the XCAT study were 0.57±0.31, 0.78±0.26, 0.83±0.21, and 0.89±0.11 for 2D-3D deformation, 2D-3D deformation with motion modeling, Bio-CBCT and MM-Bio-CBCT techniques, respectively. Using 20 projections for estimation, the patient study yielded average DICE results of 0.63±0.21, 0.73±0.13 and 0.78±0.12, and 0.83±0.09, correspondingly. The MM-Bio-CBCT localized the liver tumor to an average COME of ~2 mm for both the XCAT and the liver patient studies.

CONCLUSIONS

Compared to Bio-CBCT, MM-Bio-CBCT further improves the accuracy of liver tumor localization. MM-Bio-CBCT can potentially be used towards pre-treatment liver tumor localization and intra-treatment liver tumor location verification to achieve substantial radiotherapy margin reduction.

Deformable image registration for radiation therapy: principle, methods, applications and evaluation

Background: Deformable image registration (DIR) is increasingly used in the field of radiation therapy (RT) to account for anatomical deformations. The aims of this paper are to describe the main applications of DIR in RT and discuss current DIR evaluation methods. Methods: Articles on DIR published from January 2000 to October 2018 were extracted from PubMed and Science Direct. Our search was restricted to articles that report data obtained from humans, were written in English, and address DIR methods for RT. A total of 207 articles were selected from among 2506 identified in the search process. Results: At planning, DIR is used for organ delineation using atlas-based segmentation, deformation-based planning target volume definition, functional planning and magnetic resonance imaging-based dose calculation. In image-guided RT, DIR is used for contour propagation and dose calculation on per-treatment imaging. DIR is also used to determine the accumulated dose from fraction to fraction in external beam RT and brachytherapy, both for dose reporting and adaptive RT. In the case of re-irradiation, DIR can be used to estimate the cumulated dose of the two irradiations. Finally, DIR can be used to predict toxicity in voxel-wise population analysis. However, the evaluation of DIR remains an open issue, especially when dealing with complex cases such as the disappearance of matter. To quantify DIR uncertainties, most evaluation methods are limited to geometry-based metrics. Software companies have now integrated DIR tools into treatment planning systems for clinical use, such as contour propagation and fraction dose accumulation. Conclusions: DIR is increasingly important in RT applications, from planning to toxicity prediction. DIR is routinely used to reduce the workload of contour propagation. However, its use for complex dosimetric applications must be carefully evaluated by combining quantitative and qualitative analyses.

Evolutionary Machine Learning for Multi-Objective Class Solutions in Medical Deformable Image Registration

Current state-of-the-art medical deformable image registration (DIR) methods optimize a weighted sum of key objectives of interest. Having a pre-determined weight combination that leads to high-quality results for any instance of a specific DIR problem (i.e., a class solution) would facilitate clinical application of DIR. However, such a combination can vary widely for each instance and is currently often manually determined. A multi-objective optimization approach for DIR removes the need for manual tuning, providing a set of high-quality trade-off solutions. Here, we investigate machine learning for a multi-objective class solution, i.e., not a single weight combination, but a set thereof, that, when used on any instance of a specific DIR problem, approximates such a set of trade-off solutions. To this end, we employed a multi-objective evolutionary algorithm to learn sets of weight combinations for three breast DIR problems of increasing difficulty: 10 prone-prone cases, 4 prone-supine cases with limited deformations and 6 prone-supine cases with larger deformations and image artefacts. Clinically-acceptable results were obtained for the first two problems. Therefore, for DIR problems with limited deformations, a multi-objective class solution can be machine learned and used to compute straightforwardly multiple high-quality DIR outcomes, potentially leading to more efficient use of DIR in clinical practice.

Autosegmentation of the rectum on megavoltage image guidance scans

Autosegmentation of image guidance (IG) scans is crucial for streamlining and optimising delivered dose calculation in radiotherapy. By accounting for interfraction motion, daily delivered dose can be accumulated and incorporated into automated systems for adaptive radiotherapy. Autosegmentation of IG scans is challenging due to poorer image quality than typical planning kilovoltage computed tomography (kVCT) systems, and the resulting reduction of soft tissue contrast in regions such as the pelvis makes organ boundaries less distinguishable. Current autosegmentation solutions generally involve propagation of planning contours to the IG scan by deformable image registration (DIR). Here, we present a novel approach for primary autosegmentation of the rectum on megavoltage IG scans acquired during prostate radiotherapy, based on the Chan-Vese algorithm. Pre-processing steps such as Hounsfield unit/intensity scaling, identifying search regions, dealing with air, and handling the prostate, are detailed. Post-processing features include identification of implausible contours (nominally those affected by muscle or air), 3D self-checking, smoothing, and interpolation. In cases where the algorithm struggles, the best estimate on a given slice may revert to the propagated kVCT rectal contour. Algorithm parameters were optimised systematically for a training cohort of 26 scans, and tested on a validation cohort of 30 scans, from 10 patients. Manual intervention was not required. Comparing Chan-Vese autocontours with contours manually segmented by an experienced clinical oncologist achieved a mean Dice Similarity Coefficient of 0.78 (SE < 0.011). This was comparable with DIR methods for kVCT and CBCT published in the literature. The autosegmentation system was developed within the VoxTox Research Programme for accumulation of delivered dose to the rectum in prostate radiotherapy, but may have applicability to further anatomical sites and imaging modalities.

Usefulness of hybrid deformable image registration algorithms in prostate radiation therapy

To evaluate the accuracy of commercially available hybrid deformable image registration (DIR) algorithms when using planning CT (pCT) and daily cone-beam computed tomography (CBCT) in radiation therapy for prostate cancer. The hybrid DIR algorithms in RayStation and MIM Maestro were evaluated. Contours of the prostate, bladder, rectum, and seminal vesicles (SVs) were used as region-of-interest (ROIs) to guide image deformation in the hybrid DIR and to compare the DIR accuracy. To evaluate robustness of the hybrid DIR for prostate cancer patients with organs with volume that vary on a daily basis, such as the bladder and rectum, the DIR algorithms were performed on ten pairs of CT volumes from ten patients who underwent prostate intensity-modulated radiation therapy or volumetric modulated arc therapy. In a visual evaluation, MIM caused unrealistic image deformation in soft tissues, organs, and pelvic bones. The mean dice similarity coefficient (DSC) ranged from 0.46 to 0.90 for the prostate, bladder, rectum, and SVs; the SVs had the lowest DSC. Target registration error (TRE) at the centroid of the ROIs was about 2 mm for the prostate and bladder, and about 6 mm for the rectum and SVs. RayStation did not cause unrealistic image deformation, and could maintain the shape of pelvic bones in most cases. The mean DSC and TRE at the centroid of the ROIs were about 0.9 and within 5 mm generally. In both software programs, the use of ROIs to guide image deformation had the possibility to reduce any unrealistic image deformation and might be effective to keep the DIR physically reasonable. The pCT/CBCT DIR for the prostate cancer did not reduce the DIR accuracy because of the use of ROIs to guide the image deformation.

Dose-guided patient positioning in proton radiotherapy using multicriteria-optimization

Proton radiotherapy (PT) requires accurate target alignment before each treatment fraction, ideally utilizing 3D in-room X-ray computed tomography (CT) imaging. Typically, the optimal patient position is determined based on anatomical landmarks or implanted markers. In the presence of non-rigid anatomical changes, however, the planning scenario cannot be exactly reproduced and positioning should rather aim at finding the optimal position in terms of the actually applied dose. In this work, dose-guided patient alignment, implemented as multicriterial optimization (MCO) problem, was investigated in the scope of intensity-modulated and double-scattered PT (IMPT and DSPT) for the first time. A method for automatically determining the optimal patient position with respect to pre-defined clinical goals was implemented. Linear dose interpolation was used to access a continuous space of potential patient shifts. Fourteen head and neck (H&N) and eight prostate cancer patients with up to five repeated CTs were included. Dose interpolation accuracy was evaluated and the potential dosimetric advantages of dose-guided over bony-anatomy-based patient alignment investigated by comparison of clinically relevant target and organ-at-risk (OAR) dose-volume histogram (DVH) parameters. Dose interpolation was found sufficiently accurate with average pass-rates of 90% and 99% for an exemplary H&N and prostate patient, respectively, using a 2% dose-difference criterion. Compared to bony-anatomy-based alignment, the main impact of automated MCO-based dose-guided positioning was a reduced dose to the serial OARs (spinal cord and brain stem) for the H&N cohort. For the prostate cohort, under-dosage of the target structures could be efficiently diminished. Limitations of dose-guided positioning were mainly found in reducing target over-dosage due to weight loss for H&N patients, which might require adaptation of the treatment plan. Since labor-intense online quality-assurance is not required for dose-guided patient positioning, it might, nevertheless, be considered an interesting alternative to full online re-planning for initially mitigating the effects of anatomical changes.

Dose evaluation of fast synthetic-CT generation using a generative adversarial network for general pelvis MR-only radiotherapy

To enable magnetic resonance (MR)-only radiotherapy and facilitate modelling of radiation attenuation in humans, synthetic CT (sCT) images need to be generated. Considering the application of MR-guided radiotherapy and online adaptive replanning, sCT generation should occur within minutes. This work aims at assessing whether an existing deep learning network can rapidly generate sCT images for accurate MR-based dose calculations in the entire pelvis. A study was conducted on data of 91 patients with prostate (59), rectal (18) and cervical (14) cancer who underwent external beam radiotherapy acquiring both CT and MRI for patients' simulation. Dixon reconstructed water, fat and in-phase images obtained from a conventional dual gradient-recalled echo sequence were used to generate sCT images. A conditional generative adversarial network (cGAN) was trained in a paired fashion on 2D transverse slices of 32 prostate cancer patients. The trained network was tested on the remaining patients to generate sCT images. For 30 patients in the test set, dose recalculations of the clinical plan were performed on sCT images. Dose distributions were evaluated comparing voxel-based dose differences, gamma and dose-volume histogram (DVH) analysis. The sCT generation required 5.6 s and 21 s for a single patient volume on a GPU and CPU, respectively. On average, sCT images resulted in a higher dose to the target of maximum 0.3%. The average gamma pass rates using the 3%, 3 mm and 2%, 2 mm criteria were above 97 and 91%, respectively, for all volumes of interests considered. All DVH points calculated on sCT differed less than  ±2.5% from the corresponding points on CT. Results suggest that accurate MR-based dose calculation using sCT images generated with a cGAN trained on prostate cancer patients is feasible for the entire pelvis. The sCT generation was sufficiently fast for integration in an MR-guided radiotherapy workflow.

Evaluation of automatic contour propagation in T2-weighted 4DMRI for normal-tissue motion assessment using internal organ-at-risk volume (IRV)

Purpose

The purpose of this study was to evaluate the quality of automatically propagated contours of organs at risk (OARs) based on respiratory‐correlated navigator‐triggered four‐dimensional magnetic resonance imaging (RC‐4DMRI) for calculation of internal organ‐at‐risk volume (IRV) to account for intra‐fractional OAR motion.

Methods and Materials

T2‐weighted RC‐4DMRI images were of 10 volunteers acquired and reconstructed using an internal navigator‐echo surrogate and concurrent external bellows under an IRB‐approved protocol. Four major OARs (lungs, heart, liver, and stomach) were delineated in the 10‐phase 4DMRI. Two manual‐contour sets were delineated by two clinical personnel and two automatic‐contour sets were propagated using free‐form deformable image registration. The OAR volume variation within the 10‐phase cycle was assessed and the IRV was calculated as the union of all OAR contours. The OAR contour similarity between the navigator‐triggered and bellows‐rebinned 4DMRI was compared. A total of 2400 contours were compared to the most probable ground truth with a 95% confidence level (S95) in similarity, sensitivity, and specificity using the simultaneous truth and performance level estimation (STAPLE) algorithm.

Results

Visual inspection of automatically propagated contours finds that approximately 5–10% require manual correction. The similarity, sensitivity, and specificity between manual and automatic contours are indistinguishable (P > 0.05). The Jaccard similarity indexes are 0.92 ± 0.02 (lungs), 0.89 ± 0.03 (heart), 0.92 ± 0.02 (liver), and 0.83 ± 0.04 (stomach). Volume variations within the breathing cycle are small for the heart (2.6 ± 1.5%), liver (1.2 ± 0.6%), and stomach (2.6 ± 0.8%), whereas the IRV is much larger than the OAR volume by: 20.3 ± 8.6% (heart), 24.0 ± 8.6% (liver), and 47.6 ± 20.2% (stomach). The Jaccard index is higher in navigator‐triggered than bellows‐rebinned 4DMRI by 4% (P < 0.05), due to the higher image quality of navigator‐based 4DMRI.

Conclusion

Automatic and manual OAR contours from Navigator‐triggered 4DMRI are not statistically distinguishable. The navigator‐triggered 4DMRI image provides higher contour quality than bellows‐rebinned 4DMRI. The IRVs are 20–50% larger than OAR volumes and should be considered in dose estimation.

Quantitative analysis of respiration-induced motion of each liver segment with helical computed tomography and 4-dimensional computed tomography

Background

To analyze the respiratory-induced motion of each liver segment using helical computed tomography (helical CT) and 4-dimensional computed tomography (4DCT), and to establish the individual segment expansion margin of internal target volume (ITV) to facilitate target delineation of tumors in different liver segments.

Methods

Twenty patients who received radiotherapy with CT-simulation scanning of the whole liver in both helical CT and 10-phase-gated 4DCT were investigated, including 2 patients with esophagus cancer, 4 with lung cancer, 10 with breast cancer, 2 with liver cancer, 1 with thymoma, and 1 with gastric diffuse large B-cell lymphoma (DLBCL). For each patient, 9 representative points were drawn on the helical CT images of liver segments 1, 2, 3, 4a, 4b, 5, 6, 7, and 8, respectively, and adaptively deformed to 2 phases of the 4DCT images at the end of inspiration (phase 0 CT) and expiration (phase 50 CT) in the treatment planning system. Using the amplitude of each point between phase 0 CT and phase 50 CT, we established quantitative data for the respiration-induced motion of each liver segment in 3-dimensional directions. Moreover, using the amplitude between the original helical CT and both 4DCT images, we rendered the individual segment expansion margin of ITV for hepatic target delineation to cover more than 95% of each tumor.

Results

The average amplitude (mean ± standard deviation) was 0.6 ± 3.0 mm in the left-right (LR) direction, 2.3 ± 2.4 mm in the anterior-posterior (AP) direction, and 5.7 ± 3.4 mm in the superior-inferior (SI) direction, respectively. All of the segments moved posteriorly and superiorly during expiration. Segment 7 had the largest amplitude in the SI direction, at 8.6 ± 3.4 mm. Otherwise, the segments over the lateral side, including segments 2, 3, 6, and 7, had greater excursion in the SI direction compared to the medial segments. To cover more than 95% of each tumor, the required expansion margin of ITV in the LR, AP, and SI directions were at least 2.5 mm, 2.5 mm, and 5.0 mm on average, respectively, with variations between different segments.

Conclusions

The greatest excursion occurred in liver segment 7, followed by the segments over the lateral side in the SI direction. The individual segment expansion margin of ITV is required to delineate targets for each segment and direction.

Feasibility of MR-only proton dose calculations for prostate cancer radiotherapy using a commercial pseudo-CT generation method

A magnetic resonance (MR)-only radiotherapy workflow can reduce cost, radiation exposure and uncertainties introduced by CT-MRI registration. A crucial prerequisite is generating the so called pseudo-CT (pCT) images for accurate dose calculation and planning. Many pCT generation methods have been proposed in the scope of photon radiotherapy. This work aims at verifying for the first time whether a commercially available photon-oriented pCT generation method can be employed for accurate intensity-modulated proton therapy (IMPT) dose calculation. A retrospective study was conducted on ten prostate cancer patients. For pCT generation from MR images, a commercial solution for creating bulk-assigned pCTs, called MR for Attenuation Correction (MRCAT), was employed. The assigned pseudo-Hounsfield Unit (HU) values were adapted to yield an increased agreement to the reference CT in terms of proton range. Internal air cavities were copied from the CT to minimise inter-scan differences. CT- and MRCAT-based dose calculations for opposing beam IMPT plans were compared by gamma analysis and evaluation of clinically relevant target and organ at risk dose volume histogram (DVH) parameters. The proton range in beam's eye view (BEV) was compared using single field uniform dose (SFUD) plans. On average, a [Formula: see text] mm) gamma pass rate of 98.4% was obtained using a [Formula: see text] dose threshold after adaptation of the pseudo-HU values. Mean differences between CT- and MRCAT-based dose in the DVH parameters were below 1 Gy ([Formula: see text]). The median proton range difference was [Formula: see text] mm, with on average 96% of all BEV dose profiles showing a range agreement better than 3 mm. Results suggest that accurate MR-based proton dose calculation using an automatic commercial bulk-assignment pCT generation method, originally designed for photon radiotherapy, is feasible following adaptation of the assigned pseudo-HU values.

A graphical approach to optimizing variable-kernel smoothing parameters for improved deformable registration of CT and cone beam CT images

A systematic method is presented for determining optimal parameters in variable-kernel deformable image registration of cone beam CT and CT images, in order to improve accuracy and convergence for potential use in online adaptive radiotherapy. Assessed conditions included the noise constant (symmetric force demons), the kernel reduction rate, the kernel reduction percentage, and the kernel adjustment criteria. Four such parameters were tested in conjunction with reductions of 5, 10, 15, 20, 30, and 40%. Noise constants ranged from 1.0 to 1.9 for pelvic images in ten prostate cancer patients. A total of 516 tests were performed and assessed using the structural similarity index. Registration accuracy was plotted as a function of iteration number and a least-squares regression line was calculated, which implied an average improvement of 0.0236% per iteration. This baseline was used to determine if a given set of parameters under- or over-performed. The most accurate parameters within this range were applied to contoured images. The mean Dice similarity coefficient was calculated for bladder, prostate, and rectum with mean values of 98.26%, 97.58%, and 96.73%, respectively; corresponding to improvements of 2.3%, 9.8%, and 1.2% over previously reported values for the same organ contours. This graphical approach to registration analysis could aid in determining optimal parameters for Demons-based algorithms. It also establishes expectation values for convergence rates and could serve as an indicator of non-physical warping, which often occurred in cases  >0.6% from the regression line.

Evaluation of the performance of deformable image registration between planning CT and CBCT images for the pelvic region: comparison between hybrid and intensity-based DIR

This study aimed to evaluate the performance of the hybrid deformable image registration (DIR) method in comparison with intensity-based DIR for pelvic cone-beam computed tomography (CBCT) images, using intensity and anatomical information. Ten prostate cancer patients treated with intensity-modulated radiation therapy (IMRT) were studied. Nine or ten CBCT scans were performed for each patient. First, rigid registration was performed between the planning CT and all CBCT images using gold fiducial markers, and then DIR was performed. The Dice similarity coefficient (DSC) and center of mass (COM) displacement were used to evaluate the quantitative DIR accuracy. The average DSCs for intensity-based DIR for the prostate, rectum, bladder, and seminal vesicles were 0.84 ± 0.05, 0.75 ± 0.05, 0.69 ± 0.07 and 0.65 ± 0.11, respectively, whereas those values for hybrid DIR were 0.98 ± 0.00, 0.97 ± 0.01, 0.98 ± 0.00 and 0.94 ± 0.03, respectively (P < 0.05). The average COM displacements for intensity-based DIR for the prostate, rectum, bladder, and seminal vesicles were 2.0 ± 1.5, 3.7 ± 1.4, 7.8 ± 2.2 and 3.6 ± 1.2 mm, whereas those values for hybrid DIR were 0.1 ± 0.0, 0.3 ± 0.2, 0.2 ± 0.1 and 0.6 ± 0.6 mm, respectively (P < 0.05). These results showed that the DSC for hybrid DIR had a higher DSC value and smaller COM displacement for all structures and all patients, compared with intensity-based DIR. Thus, the accumulative dose based on hybrid DIR might be trusted as a high-precision dose estimation method that takes into account organ movement during treatment radiotherapy.

Evaluation of deformable image registration for contour propagation between CT and cone-beam CT images in adaptive head and neck radiotherapy

Deformable image registration (DIR) is a critical technic in adaptive radiotherapy (ART) to propagate contours between planning computerized tomography (CT) images and treatment CT/Cone-beam CT (CBCT) image to account for organ deformation for treatment re-planning. To validate the ability and accuracy of DIR algorithms in organ at risk (OAR) contours mapping, seven intensity-based DIR strategies are tested on the planning CT and weekly CBCT images from six Head & Neck cancer patients who underwent a 6 ∼ 7 weeks intensity-modulated radiation therapy (IMRT). Three similarity metrics, i.e. the Dice similarity coefficient (DSC), the percentage error (PE) and the Hausdorff distance (HD), are employed to measure the agreement between the propagated contours and the physician delineated ground truths. It is found that the performance of all the evaluated DIR algorithms declines as the treatment proceeds. No statistically significant performance difference is observed between different DIR algorithms (p> 0.05), except for the double force demons (DFD) which yields the worst result in terms of DSC and PE. For the metric HD, all the DIR algorithms behaved unsatisfactorily with no statistically significant performance difference (p= 0.273). These findings suggested that special care should be taken when utilizing the intensity-based DIR algorithms involved in this study to deform OAR contours between CT and CBCT, especially for those organs with low contrast.