Clinical application and improvement of a CNN-based autosegmentation model for clinical target volumes in cervical cancer radiotherapy.

OBJECTIVE: Clinical target volume (CTV) autosegmentation for cervical cancer is desirable for radiation therapy. Data heterogeneity and interobserver variability (IOV) limit the clinical adaptability of such methods. The adaptive method is proposed to improve the adaptability of CNN-based autosegmentation of CTV contours in cervical cancer. METHODS: This study included 400 cervical cancer treatment planning cases with CTV delineated by radiation oncologists from three hospitals. The datasets were divided into five subdatasets (80 cases each). The cases in datasets 1, 2, and 3 were delineated by physicians A, B, and C, respectively. The cases in datasets 4 and 5 were delineated by multiple physicians. Dataset 1 was divided into training (50 cases), validation (10 cases), and testing (20 cases) cohorts, and they were used to construct the pretrained model. Datasets 2-5 were regarded as host datasets to evaluate the accuracy of the pretrained model. In the adaptive process, the pretrained model was fine-tuned to measure improvements by gradually adding more training cases selected from the host datasets. The accuracy of the autosegmentation model on each host dataset was evaluated using the corresponding test cases. The Dice similarity coefficient (DSC) and 95% Hausdorff distance (HD_95) were used to evaluate the accuracy. RESULTS: Before and after adaptive improvements, the average DSC values on the host datasets were 0.818 versus 0.882, 0.763 versus 0.810, 0.727 versus 0.772, and 0.679 versus 0.789, which are improvements of 7.82%, 6.16%, 6.19%, and 16.05%, respectively. The average HD_95 values were 11.143 mm versus 6.853 mm, 22.402 mm versus 14.076 mm, 28.145 mm versus 16.437 mm, and 33.034 mm versus 16.441 mm, which are improvements of 37.94%, 37.17%, 41.60%, and 50.23%, respectively. CONCLUSION: The proposed method improved the adaptability of the CNN-based autosegmentation model when applied to host datasets.

Patient-Specific Organ and Effective Dose Estimates in Adult Oncologic CT.

OBJECTIVE. Patient-specific organ and effective dose provides essential information for CT protocol optimization. However, such information is not readily available in the scan records. The purpose of this study was to develop a method to obtain accurate examination- and patient-specific organ and effective dose estimates by use of available scan data and patient body size information for a large cohort of patients. MATERIALS AND METHODS. The data were randomly collected for 1200 patients who underwent CT in a 2-year period. Physical characteristics of the patients and CT technique were processed as inputs for the dose estimator. Organ and effective doses were estimated by use of the inputs and computational human phantoms matched to patients on the basis of sex and effective diameter. Size-based ratios were applied to correct for patient-phantom body size differences. RESULTS. Patients received a mean of 59.9 mGy to the lens of the eye per brain scan, 10.1 mGy to the thyroid per chest scan, 17.5 mGy to the liver per abdomen and pelvis scan, and 19.0 mGy to the liver per body scan. A factor of 2 difference in dose estimates was observed between patients of various habitus. CONCLUSION. Examination- and patient-specific organ and effective doses were estimated for 1200 adult oncology patients undergoing CT. The dose conversion factors calculated facilitate rapid organ and effective dose estimation in clinics. Compared with nonspecific dose estimation methods, patient dose estimations with data specific to the patient and examination can differ by a factor of 2.

End-to-end unsupervised cycle-consistent fully convolutional network for 3D pelvic CT-MR deformable registration.

OBJECTIVE: To improve the efficiency of computed tomography (CT)-magnetic resonance (MR) deformable image registration while ensuring the registration accuracy. METHODS: Two fully convolutional networks (FCNs) for generating spatial deformable grids were proposed using the Cycle-Consistent method to ensure the deformed image consistency with the reference image data. In all, 74 pelvic cases consisting of both MR and CT images were studied, among which 64 cases were used as training data and 10 cases as the testing data. All training data were standardized and normalized, following simple image preparation to remove the redundant air. Dice coefficients and average surface distance (ASD) were calculated for regions of interest (ROI) of CT-MR image pairs, before and after the registration. The performance of the proposed method (FCN with Cycle-Consistent) was compared with that of Elastix software, MIM software, and FCN without cycle-consistent. RESULTS: The results show that the proposed method achieved the best performance among the four registration methods tested in terms of registration accuracy and the method was more stable than others in general. In terms of average registration time, Elastix took 64 s, MIM software took 28 s, and the proposed method was found to be significantly faster, taking <0.1 s. CONCLUSION: The proposed method not only ensures the accuracy of deformable image registration but also greatly reduces the time required for image registration and improves the efficiency of the registration process. In addition, compared with other deep learning methods, the proposed method is completely unsupervised and end-to-end.

A method of using deep learning to predict three-dimensional dose distributions for intensity-modulated radiotherapy of rectal cancer.

PURPOSE: To develop and test a three-dimensional (3D) deep learning model for predicting 3D voxel-wise dose distributions for intensity-modulated radiotherapy (IMRT). METHODS: A total of 122 postoperative rectal cancer cases treated by IMRT were considered in the study, of which 100 cases were randomly selected as the training-validating set and the remaining as the testing set. A 3D deep learning model named 3D U-Res-Net_B was constructed to predict 3D dose distributions. Eight types of 3D matrices from CT images, contoured structures, and beam configurations were fed into the independent input channel, respectively, and the 3D matrix of dose distributions was taken as the output to train the 3D model. The obtained 3D model was used to predict new 3D dose distributions. The predicted accuracy was evaluated in two aspects: (a) The dice similarity coefficients (DSCs) of different isodose volumes, the average dose difference of all voxels within the body, and 3%/5 mm global gamma passing rates of organs at risks (OARs) and planned target volume (PTV) were used to address the spatial correspondence between predicted and clinical delivered 3D dose distributions; (b) The dosimetric index (DI) including homogeneity index, conformity index, V50 , V45 for PTV and OARs between predicted and clinical truth were statistically analyzed with the paired-samples t test. The model was also compared with 3D U-Net and the same architecture model without beam configurations input (named as 3D U-Res-Net_O). RESULTS: The 3D U-Res-Net_B model predicted 3D dose distributions accurately. For the 22 testing cases, the average prediction bias ranged from -1.94% to 1.58%, and the overall mean absolute errors (MAEs) was 3.92 ± 4.16%; there was no statistically significant difference for nearly all DIs. The model had a DSCs value above 0.9 for most isodose volumes, and global 3D gamma passing rates varying from 0.81 to 0.90 for PTV and OARs, clearly outperforming 3D U-Res-Net_O and being slightly superior to 3D U-Net. CONCLUSIONS: This study developed a more general deep learning model by considering beam configurations input and achieved an accurate 3D voxel-wise dose prediction for rectal cancer treated by IMRT, a potentially easier clinical implementation for more comprehensive automatic planning.

Evaluation of deep learning-based auto-segmentation algorithms for delineating clinical target volume and organs at risk involving data for 125 cervical cancer patients.

OBJECTIVE: To evaluate the accuracy of a deep learning-based auto-segmentation mode to that of manual contouring by one medical resident, where both entities tried to mimic the delineation "habits" of the same clinical senior physician. METHODS: This study included 125 cervical cancer patients whose clinical target volumes (CTVs) and organs at risk (OARs) were delineated by the same senior physician. Of these 125 cases, 100 were used for model training and the remaining 25 for model testing. In addition, the medical resident instructed by the senior physician for approximately 8 months delineated the CTVs and OARs for the testing cases. The dice similarity coefficient (DSC) and the Hausdorff Distance (HD) were used to evaluate the delineation accuracy for CTV, bladder, rectum, small intestine, femoral-head-left, and femoral-head-right. RESULTS: The DSC values of the auto-segmentation model and manual contouring by the resident were, respectively, 0.86 and 0.83 for the CTV (P < 0.05), 0.91 and 0.91 for the bladder (P > 0.05), 0.88 and 0.84 for the femoral-head-right (P < 0.05), 0.88 and 0.84 for the femoral-head-left (P < 0.05), 0.86 and 0.81 for the small intestine (P < 0.05), and 0.81 and 0.84 for the rectum (P > 0.05). The HD (mm) values were, respectively, 14.84 and 18.37 for the CTV (P < 0.05), 7.82 and 7.63 for the bladder (P > 0.05), 6.18 and 6.75 for the femoral-head-right (P > 0.05), 6.17 and 6.31 for the femoral-head-left (P > 0.05), 22.21 and 26.70 for the small intestine (P > 0.05), and 7.04 and 6.13 for the rectum (P > 0.05). The auto-segmentation model took approximately 2 min to delineate the CTV and OARs while the resident took approximately 90 min to complete the same task. CONCLUSION: The auto-segmentation model was as accurate as the medical resident but with much better efficiency in this study. Furthermore, the auto-segmentation approach offers additional perceivable advantages of being consistent and ever improving when compared with manual approaches.

The impact of robustness of deformable image registration on contour propagation and dose accumulation for head and neck adaptive radiotherapy.

Deformable image registration (DIR) is the key process for contour propagation and dose accumulation in adaptive radiation therapy (ART). However, currently, ART suffers from a lack of understanding of "robustness" of the process involving the image contour based on DIR and subsequent dose variations caused by algorithm itself and the presetting parameters. The purpose of this research is to evaluate the DIR caused variations for contour propagation and dose accumulation during ART using the RayStation treatment planning system. Ten head and neck cancer patients were selected for retrospective studies. Contours were performed by a single radiation oncologist and new treatment plans were generated on the weekly CT scans for all patients. For each DIR process, four deformation vector fields (DVFs) were generated to propagate contours and accumulate weekly dose by the following algorithms: (a) ANACONDA with simple presetting parameters, (b) ANACONDA with detailed presetting parameters, (c) MORFEUS with simple presetting parameters, and (d) MORFEUS with detailed presetting parameters. The geometric evaluation considered DICE coefficient and Hausdorff distance. The dosimetric evaluation included D95 , Dmax , Dmean , Dmin , and Homogeneity Index. For geometric evaluation, the DICE coefficient variations of the GTV were found to be 0.78 ± 0.11, 0.96 ± 0.02, 0.64 ± 0.15, and 0.91 ± 0.03 for simple ANACONDA, detailed ANACONDA, simple MORFEUS, and detailed MORFEUS, respectively. For dosimetric evaluation, the corresponding Homogeneity Index variations were found to be 0.137 ± 0.115, 0.006 ± 0.032, 0.197 ± 0.096, and 0.006 ± 0.033, respectively. The coherent geometric and dosimetric variations also consisted in large organs and small organs. Overall, the results demonstrated that the contour propagation and dose accumulation in clinical ART were influenced by the DIR algorithm, and to a greater extent by the presetting parameters. A quality assurance procedure should be established for the proper use of a commercial DIR for adaptive radiation therapy.

A comparison of pediatric and adult CT organ dose estimation methods.

BACKGROUND: Computed Tomography (CT) contributes up to 50% of the medical exposure to the United States population. Children are considered to be at higher risk of developing radiation-induced tumors due to the young age of exposure and increased tissue radiosensitivity. Organ dose estimation is essential for pediatric and adult patient cancer risk assessment. The objective of this study is to validate the VirtualDose software in comparison to currently available software and methods for pediatric and adult CT organ dose estimation. METHODS: Five age groups of pediatric patients and adult patients were simulated by three organ dose estimators. Head, chest, abdomen-pelvis, and chest-abdomen-pelvis CT scans were simulated, and doses to organs both inside and outside the scan range were compared. For adults, VirtualDose was compared against ImPACT and CT-Expo. For pediatric patients, VirtualDose was compared to CT-Expo and compared to size-based methods from literature. Pediatric to adult effective dose ratios were also calculated with VirtualDose, and were compared with the ranges of effective dose ratios provided in ImPACT. RESULTS: In-field organs see less than 60% difference in dose between dose estimators. For organs outside scan range or distributed organs, a five times' difference can occur. VirtualDose agrees with the size-based methods within 20% difference for the organs investigated. Between VirtualDose and ImPACT, the pediatric to adult ratios for effective dose are compared, and less than 21% difference is observed for chest scan while more than 40% difference is observed for head-neck scan and abdomen-pelvis scan. For pediatric patients, 2 cm scan range change can lead to a five times dose difference in partially scanned organs. CONCLUSIONS: VirtualDose is validated against CT-Expo and ImPACT with relatively small discrepancies in dose for organs inside scan range, while large discrepancies in dose are observed for organs outside scan range. Patient-specific organ dose estimation is possible using the size-based methods, and VirtualDose agrees with size-based method for the organs investigated. Careful range selection for CT protocols is necessary for organ dose optimization for pediatric and adult patients.

ESTIMATION OF FETAL AND PEDIATRIC DOSES FROM CHEST CT EXAMINATIONS USING VIRTUALDOSE SOFTWARE.

Pregnant women and children sometimes had to undergo chest computed tomography (CT) scans during the Corona Virus Disease 2019 (COVID-19) pandemic. This study estimated the fetal and pediatric doses from chest CT scans. Organ doses and effective doses were calculated using the VirtualDose-CT software. Two groups of computational human phantoms, pregnant females and pediatric patients were used in this study. The results of doses normalized to volumetric CT Dose Index (CTDIvol) can be used universally for other dosimetry studies. Based on our calculations and international survey data of CTDIvol, fetal absorbed doses from COVID-19-related chest CT were found to be 0.04-0.36, 0.05-0.44 and 0.07-0.61 mGy for 3, 6 and 9 months of pregnancy, respectively. When the scan range is extended to the abdominal region, fetal doses increase by almost 4-fold. Effective doses for COVID-19-related chest CT were 1.62-13.77, 1.58-13.46, 1.57-13.33 and 1.29-10.98 mSv for the newborn, 1-, 5- and 10-y-old children, respectively. In addition, the effects of specific axial scan ranges exceeding the thorax region were evaluated. Although doses from chest CT scans are small, such data allow radiologists and patients to be informed of the dose levels and ways to avoid unnecessary radiation.

Development of a GPU-accelerated Monte Carlo dose calculation module for nuclear medicine, ARCHER-NM: demonstration for a PET/CT imaging procedure.

Objective.This paper describes the development and validation of a GPU-accelerated Monte Carlo (MC) dose computing module dedicated to organ dose calculations of individual patients undergoing nuclear medicine (NM) internal radiation exposures involving PET/CT examination.Approach. This new module extends the more-than-10-years-long ARCHER project that developed a GPU-accelerated MC dose engine by adding dedicated NM source-definition features. To validate the code, we compared dose distributions from the point ion source, including18F,11C,15O, and68Ga, calculated for a water phantom against a well-tested MC code, GATE. To demonstrate the clinical utility and advantage of ARCHER-NM, one set of18F-FDG PET/CT data for an adult male NM patient is calculated using the new code. Radiosensitive organs in the CT dataset are segmented using a CNN-based tool called DeepViewer. The PET image intensity maps are converted to radioactivity distributions to allow for MC radiation transport dose calculations at the voxel level. The dose rate maps and corresponding statistical uncertainties were calculated at the acquisition time of PET image.Main results.The water-phantom results show excellent agreement, suggesting that the radiation physics module in the new NM code is adequate. The dose rate results of the18F-FDG PET imaging patient show that ARCHER-NM's results agree very well with those of the GATE within -2.45% to 2.58% (for a total of 28 organs considered in this study). Most impressively, ARCHER-NM obtains such results in 22 s while it takes GATE about 180 min for the same number of 5 × 108simulated decay events.Significance.This is the first study presenting GPU-accelerated patient-specific MC internal radiation dose rate calculations for clinically realistic18F-FDG PET/CT imaging case involving autosegmentation of whole-body PET/CT images. This study suggests that the proposed computing tools-ARCHER-NM- are accurate and fast enough for routine internal dosimetry in NM clinics.

Real time 4D IMRT treatment planning based on a dynamic virtual patient model: proof of concept.

PURPOSE: To develop a novel four-dimensional (4D) intensity modulated radiation therapy (IMRT) treatment planning methodology based on dynamic virtual patient models. METHODS: The 4D model-based planning (4DMP) is a predictive tracking method which consists of two main steps: (1) predicting the 3D deformable motion of the target and critical structures as a function of time during treatment delivery; (2) adjusting the delivery beam apertures formed by the dynamic multi-leaf collimators (DMLC) to account for the motion. The key feature of 4DMP is the application of a dynamic virtual patient model in motion prediction, treatment beam adjustment, and dose calculation. A lung case was chosen to demonstrate the feasibility of the 4DMP. For the lung case, a dynamic virtual patient model (4D model) was first developed based on the patient's 4DCT images. The 4D model was capable of simulating respiratory motion of different patterns. A model-based registration method was then applied to convert the 4D model into a set of deformation maps and 4DCT images for dosimetric purposes. Based on the 4D model, 4DMP treatment plans with different respiratory motion scenarios were developed. The quality of 4DMP plans was then compared with two other commonly used 4D planning methods: maximum intensity projection (MIP) and planning on individual phases (IP). RESULTS: Under regular periodic motion, 4DMP offered similar target coverage as MIP with much better normal tissue sparing. At breathing amplitude of 2 cm, the lung V20 was 23.9% for a MIP plan and 16.7% for a 4DMP plan. The plan quality was comparable between 4DMP and IP: PTV V97 was 93.8% for the IP plan and 93.6% for the 4DMP plan. Lung V20 of the 4DMP plan was 2.1% lower than that of the IP plan and Dmax to cord was 2.2 Gy higher. Under a real time irregular breathing pattern, 4DMP had the best plan quality. PTV V97 was 90.4% for a MIP plan, 88.6% for an IP plan and 94.1% for a 4DMP plan. Lung V20 was 20.1% for the MIP plan, 17.8% for the IP plan and 17.5% for the 4DMP plan. The deliverability of the real time 4DMP plan was proved by calculating the maximum leaf speed of the DMLC. CONCLUSIONS: The 4D model-based planning, which applies dynamic virtual patient models in IMRT treatment planning, can account for the real time deformable motion of the tumor under different breathing conditions. Under regular motion, the quality of 4DMP plans was comparable with IP and superior to MIP. Under realistic motion in which breathing amplitude and period change, 4DMP gave the best plan quality of the three 4D treatment planning techniques.

Proton radiography and fluoroscopy of lung tumors: a Monte Carlo study using patient-specific 4DCT phantoms.

PURPOSE: Monte Carlo methods are used to simulate and optimize a time-resolved proton range telescope (TRRT) in localization of intrafractional and interfractional motions of lung tumor and in quantification of proton range variations. METHODS: The Monte Carlo N-Particle eXtended (MCNPX) code with a particle tracking feature was employed to evaluate the TRRT performance, especially in visualizing and quantifying proton range variations during respiration. Protons of 230 MeV were tracked one by one as they pass through position detectors, patient 4DCT phantom, and finally scintillator detectors that measured residual ranges. The energy response of the scintillator telescope was investigated. Mass density and elemental composition of tissues were defined for 4DCT data. RESULTS: Proton water equivalent length (WEL) was deduced by a reconstruction algorithm that incorporates linear proton track and lateral spatial discrimination to improve the image quality. 4DCT data for three patients were used to visualize and measure tumor motion and WEL variations. The tumor trajectories extracted from the WEL map were found to be within 1 mm agreement with direct 4DCT measurement. Quantitative WEL variation studies showed that the proton radiograph is a good representation of WEL changes from entrance to distal of the target. CONCLUSIONS: MCNPX simulation results showed that TRRT can accurately track the motion of the tumor and detect the WEL variations. Image quality was optimized by choosing proton energy, testing parameters of image reconstruction algorithm, and comparing to ground truth 4DCT. The future study will demonstrate the feasibility of using the time resolved proton radiography as an imaging tool for proton treatments of lung tumors.

COMPARISON OF PARTICLE-TRACKING FEATURES IN GEANT4 AND MCNPX CODES FOR APPLICATIONS IN MAPPING OF PROTON RANGE UNCERTAINTY.

The accuracy of proton therapy is partially limited by uncertainties that result from changing pathological conditions in the patient such as tumor motion and shrinkage. These uncertainties can be minimized with the help of a time-resolved range telescope. Monte Carlo methods can help improve the performance of range telescopes by tracking proton interactions on a particle-by-particle basis thus broadening our understanding on the behavior of protons within the patient and the detector. This paper compared the proton multiple coulomb scattering algorithms in the Monte Carlo codes MCNPX and Geant4 to well-established scattering theories. We focus only on beam energies associated with proton imaging. Despite slight discrepancies between scattering algorithms, both codes appear to be capable of providing useful particle-tracking information for applications such as the proton range telescope.

Feasibility evaluation of PET scan-time reduction for diagnosing amyloid-ß levels in Alzheimer's disease patients using a deep-learning-based denoising algorithm.

PURPOSE: To shorten positron emission tomography (PET) scanning time in diagnosing amyloid-ß levels thus increasing the workflow in centers involving Alzheimer's Disease (AD) patients. METHODS: PET datasets were collected for 25 patients injected with 18F-AV45 radiopharmaceutical. To generate necessary training data, PET images from both normal-scanning-time (20-min) as well as so-called "shortened-scanning-time" (1-min, 2-min, 5-min, and 10-min) were reconstructed for each patient. Building on our earlier work on MCDNet (Monte Carlo Denoising Net) and a new Wasserstein-GAN algorithm, we developed a new denoising model called MCDNet-2 to predict normal-scanning-time PET images from a series of shortened-scanning-time PET images. The quality of the predicted PET images was quantitatively evaluated using objective metrics including normalized-root-mean-square-error (NRMSE), structural similarity (SSIM), and peak signal-to-noise ratio (PSNR). Furthermore, two radiologists performed subjective evaluations including the qualitative evaluation and a five-point grading evaluation. The denoising performance of the proposed MCDNet-2 was finally compared with those of U-Net, MCDNet, and a traditional denoising method called Gaussian Filtering. RESULTS: The proposed MCDNet-2 can yield good denoising performance in 5-min PET images. In the comparison of denoising methods, MCDNet-2 yielded the best performance in the subjective evaluation although it is comparable with MCDNet in objective comparison (NRMSE, PSNR, and SSIM). In the qualitative evaluation of amyloid-ß positive or negative results, MCDNet-2 was found to achieve a classification accuracy of 100%. CONCLUSIONS: The proposed denoising method has been found to reduce the PET scan time from the normal level of 20 min to 5 min but still maintaining acceptable image quality in correctly diagnosing amyloid-ß levels. These results suggest strongly that deep learning-based methods such as ours can be an attractive solution to the clinical needs to improve PET imaging workflow.

A comparative study on the risk of second primary cancers in out-of-field organs associated with radiotherapy of localized prostate carcinoma using Monte Carlo-based accelerator and patient models.

PURPOSE: A physician's decision regarding an ideal treatment approach (i.e., radiation, surgery, and/or hormonal) for prostate carcinoma is traditionally based on a variety of metrics. One of these metrics is the risk of radiation-induced second primary cancer following radiation treatments. The aim of this study was to investigate the significance of second cancer risks in out-of-field organs from 3D-CRT and IMRT treatments of prostate carcinoma compared to baseline cancer risks in these organs. METHODS: Monte Carlo simulations were performed using a detailed medical linear accelerator model and an anatomically realistic adult male whole-body phantom. A four-field box treatment, a four-field box treatment plus a six-field boost, and a seven-field IMRT treatment were simulated. Using BEIR VII risk models, the age-dependent lifetime attributable risks to various organs outside the primary beam with a known predilection for cancer were calculated using organ-averaged equivalent doses. RESULTS: The four-field box treatment had the lowest treatment-related second primary cancer risks to organs outside the primary beam ranging from 7.3 x 10(-9) to 2.54 x 10(-5)%/MU depending on the patients age at exposure and second primary cancer site. The risks to organs outside the primary beam from the four-field box and six-field boost and the seven-field IMRT were nearly equivalent. The risks from the four-field box and six-field boost ranged from 1.39 x 10(-8) to 1.80 x 10(-5)%/MU, and from the seven-field IMRT ranged from 1.60 x 10(-9) to 1.35 x 10(-5)%/MU. The second cancer risks in all organs considered from each plan were below the baseline risks. CONCLUSIONS: The treatment-related second cancer risks in organs outside the primary beam due to 3D-CRT and IMRT is small. New risk assessment techniques need to be investigated to address the concern of radiation-induced second cancers from prostate treatments, particularly focusing on risks to organs inside the primary beam.

Monte Carlo calculation of imaging doses from diagnostic multidetector CT and kilovoltage cone-beam CT as part of prostate cancer treatment plans.

PURPOSE: To calculate imaging doses to the rectum, bladder, and femoral heads as part of a prostate cancer treatment plans, assuming an image guided radiation therapy (IGRT) procedure involving either the multidetector CT (MDCT) or kilovoltage cone-beam CT (kV CBCT). METHODS: This study considered an IGRT treatment plan for a prostate carcinoma patient involving 50.4 Gy from 28 initial fractions and a boost of 28.8 Gy from 16 fractions. A total of 45 CT imaging procedures, each involving a MDCT or a kV CBCT scan procedure, were carefully modeled using the MCNPX code version 2.5.0. The MDCT scanner model is based on the GE LightSpeed 16-MDCT scanner and the kV CBCT scanner model is based on the Varian On-Board Imager using parameters reported by the CT manufacturers and literatures. A patient-specific treatment planning CT data set was used to construct the phantom for the dose calculation. The target, organs-at-risk (OARs), and background voxels in the CT data set were categorized into six tissue types according to CT numbers for Monte Carlo calculations. RESULTS: For a total of 45 imaging procedures, it was found that the rectum received 78.4 and 76.7 cGy from MDCT and kV CBCT, respectively. The bladder received slightly greater doses of 82.4 and 77.9 cGy, while the femoral heads received much higher doses of 182.3 and 141.3 cGy from MDCT and kV CBCT, respectively. To investigate the impact of these imaging doses on treatment planning, OAR doses from MDCT or kV CBCT imaging procedures were added to the corresponding dose matrix reported by the original treatment plans to construct dose volume histograms. It was found that after the imaging dose is added, the rectum volumes irradiated to 75 and 70 Gy increased from 13.9% and 21.2%, respectively, in the original plan to 14.8% and 21.8%. The bladder volumes receiving 80 Gy increased to 4.6% from 4.1% in the original plan and the volume receiving 75 Gy increased to 7.9% from 7.5%. All values remained within the tolerance levels: V70<25%, V75 <15% for rectum and V75 < 25%, V80 < 15% for bladder. The irradiation of femoral heads was also acceptable with no volume receiving >45 Gy. CONCLUSIONS: IGRT procedures can irradiate the OARs to an imaging dose level that is great enough to require careful evaluation and perhaps even adjustment of original treatment planning in order to still satisfy the dose constraints. This study only considered one patient CT because the CT x rays cover a relatively larger volume of the body and the dose distribution is considerably more uniform than those associated with the therapeutic beams. As a result, the dose to an organ from CT imaging doses does not vary much from one patient to the other for the same CT settings. One factor that would potentially affect such CT dose level is the size of the patient body. More studies are needed to develop accurate and convenient methods of accounting for the imaging doses as part of treatment planning.

Comparison of organ doses for patients undergoing balloon brachytherapy of the breast with HDR 192Ir or electronic sources using monte carlo simulations in a heterogeneous human phantom.

PURPOSE: Accelerated partial breast irradiation via interstitial balloon brachytherapy is a fast and effective treatment method for certain early stage breast cancers. The radiation can be delivered using a conventional high-dose rate (HDR) 192Ir gamma-emitting source or a novel electronic brachytherapy (eBx) source which uses lower energy x rays that do not penetrate as far within the patient. A previous study [A. Dickler, M. C. Kirk, N. Seif, K. Griem, K. Dowlatshahi, D. Francescatti, and R. A. Abrams, "A dosimetric comparison of MammoSite high-dose-rate brachytherapy and Xoft Axxent electronic brachytherapy," Brachytherapy 6, 164-168 (2007)] showed that the target dose is similar for HDR 192Ir and eBx. This study compares these sources based on the dose received by healthy organs and tissues away from the treatment site. METHODS: A virtual patient with left breast cancer was represented by a whole-body, tissue-heterogeneous female voxel phantom. Monte Carlo methods were used to calculate the dose to healthy organs in a virtual patient undergoing balloon brachytherapy of the left breast with HDR 192Ir or eBx sources. The dose-volume histograms for a few organs which received large doses were also calculated. Additional simulations were performed with all tissues in the phantom defined as water to study the effect of tissue inhomogeneities. RESULTS: For both HDR 192Ir and eBx, the largest mean organ doses were received by the ribs, thymus gland, left lung, heart, and sternum which were close to the brachytherapy source in the left breast, eBx yielded mean healthy organ doses that were more than a factor of approximately 1.4 smaller than for HDR 192Ir for all organs considered, except for the three closest ribs. Excluding these ribs, the average and median dose-reduction factors were approximately 28 and approximately 11, respectively. The volume distribution of doses in nearby soft tissue organs that were outside the PTV were also improved with eBx. However, the maximum dose to the closest rib with the eBx source was 5.4 times greater than that of the HDR 192Ir source. The ratio of tissue-to-water maximum rib dose for the eBx source was approximately 5. CONCLUSIONS: The results of this study indicate that eBx may offer lower toxicity to most healthy tissues, except nearby bone. TG-43 methods have a tendency to underestimate dose to bone, especially the ribs. Clinical studies evaluating the negative health effects caused by irradiating healthy organs are needed so that physicians can better understand when HDR 192Ir or eBx might best benefit a patient.

CONCEPTUAL DESIGN AND PRELIMINARY RESULTS OF A VR-BASED RADIATION SAFETY TRAINING SYSTEM FOR INTERVENTIONAL RADIOLOGISTS.

To help minimise occupational radiation exposure in interventional radiology, we conceptualised a virtual reality-based radiation safety training system to help operators understand complex radiation fields and to avoid high radiation areas through game-like interactive simulations. The preliminary development of the system has yielded results suggesting that the training system can calculate and report the radiation exposure after each training session based on a database precalculated from computational phantoms and Monte Carlo simulations and the position information provided by the Microsoft HoloLens headset. In addition, real-time dose rate and cumulative dose will be displayed to the trainee to help them adjust their practice. This paper presents the conceptual design of the overall hardware and software design, as well as preliminary results to combine HoloLens headset and complex 3D X-ray field spatial distribution data to create a mixed reality environment for safety training purpose in interventional radiology.

A method of rapid quantification of patient-specific organ doses for CT using deep-learning-based multi-organ segmentation and GPU-accelerated Monte Carlo dose computing.

PURPOSE: One technical barrier to patient-specific computed tomography (CT) dosimetry has been the lack of computational tools for the automatic patient-specific multi-organ segmentation of CT images and rapid organ dose quantification. When previous CT images are available for the same body region of the patient, the ability to obtain patient-specific organ doses for CT - in a similar manner as radiation therapy treatment planning - will open the door to personalized and prospective CT scan protocols. This study aims to demonstrate the feasibility of combining deep-learning algorithms for automatic segmentation of multiple radiosensitive organs from CT images with the GPU-based Monte Carlo rapid organ dose calculation. METHODS: A deep convolutional neural network (CNN) based on the U-Net for organ segmentation is developed and trained to automatically delineate multiple radiosensitive organs from CT images. Two databases are used: The lung CT segmentation challenge 2017 (LCTSC) dataset that contains 60 thoracic CT scan patients, each consisting of five segmented organs, and the Pancreas-CT (PCT) dataset, which contains 43 abdominal CT scan patients each consisting of eight segmented organs. A fivefold cross-validation method is performed on both sets of data. Dice similarity coefficients (DSCs) are used to evaluate the segmentation performance against the ground truth. A GPU-based Monte Carlo dose code, ARCHER, is used to calculate patient-specific CT organ doses. The proposed method is evaluated in terms of relative dose errors (RDEs). To demonstrate the potential improvement of the new method, organ dose results are compared against those obtained for population-average patient phantoms used in an off-line dose reporting software, VirtualDose, at Massachusetts General Hospital. RESULTS: The median DSCs are found to be 0.97 (right lung), 0.96 (left lung), 0.92 (heart), 0.86 (spinal cord), 0.76 (esophagus) for the LCTSC dataset, along with 0.96 (spleen), 0.96 (liver), 0.95 (left kidney), 0.90 (stomach), 0.87 (gall bladder), 0.80 (pancreas), 0.75 (esophagus), and 0.61 (duodenum) for the PCT dataset. Comparing with organ dose results from population-averaged phantoms, the new patient-specific method achieved smaller absolute RDEs (mean ± standard deviation) for all organs: 1.8% ± 1.4% (vs 16.0% ± 11.8%) for the lung, 0.8% ± 0.7% (vs 34.0% ± 31.1%) for the heart, 1.6% ± 1.7% (vs 45.7% ± 29.3%) for the esophagus, 0.6% ± 1.2% (vs 15.8% ± 12.7%) for the spleen, 1.2% ± 1.0% (vs 18.1% ± 15.7%) for the pancreas, 0.9% ± 0.6% (vs 20.0% ± 15.2%) for the left kidney, 1.7% ± 3.1% (vs 19.1% ± 9.8%) for the gallbladder, 0.3% ± 0.3% (vs 24.2% ± 18.7%) for the liver, and 1.6% ± 1.7% (vs 19.3% ± 13.6%) for the stomach. The trained automatic segmentation tool takes <5 s per patient for all 103 patients in the dataset. The Monte Carlo radiation dose calculations performed in parallel to the segmentation process using the GPU-accelerated ARCHER code take <4 s per patient to achieve <0.5% statistical uncertainty in all organ doses for all 103 patients in the database. CONCLUSION: This work shows the feasibility to perform combined automatic patient-specific multi-organ segmentation of CT images and rapid GPU-based Monte Carlo dose quantification with clinically acceptable accuracy and efficiency.

Monte Carlo modeling of a 6 and 18 MV Varian Clinac medical accelerator for in-field and out-of-field dose calculations: development and validation.

There is a serious and growing concern about the increased risk of radiation-induced second cancers and late tissue injuries associated with radiation treatment. To better understand and to more accurately quantify non-target organ doses due to scatter and leakage radiation from medical accelerators, a detailed Monte Carlo model of the medical linear accelerator is needed. This paper describes the development and validation of a detailed accelerator model of the Varian Clinac operating at 6 and 18 MV beam energies. Over 100 accelerator components have been defined and integrated using the Monte Carlo code MCNPX. A series of in-field and out-of-field dose validation studies were performed. In-field dose distributions calculated using the accelerator models were tuned to match measurement data that are considered the de facto 'gold standard' for the Varian Clinac accelerator provided by the manufacturer. Field sizes of 4 cm x 4 cm, 10 cm x 10 cm, 20 cm x 20 cm and 40 cm x 40 cm were considered. The local difference between calculated and measured dose on the percent depth dose curve was less than 2% for all locations. The local difference between calculated and measured dose on the dose profile curve was less than 2% in the plateau region and less than 2 mm in the penumbra region for all locations. Out-of-field dose profiles were calculated and compared to measurement data for both beam energies for field sizes of 4 cm x 4 cm, 10 cm x 10 cm and 20 cm x 20 cm. For all field sizes considered in this study, the average local difference between calculated and measured dose for the 6 and 18 MV beams was 14 and 16%, respectively. In addition, a method for determining neutron contamination in the 18 MV operating model was validated by comparing calculated in-air neutron fluence with reported calculations and measurements. The average difference between calculated and measured neutron fluence was 20%. As one of the most detailed accelerator models for both in-field and out-of-field dose calculations, the model will be combined with anatomically realistic computational patient phantoms into a computational framework to calculate non-target organ doses to patients from various radiation treatment plans.

Calculated organ doses from selected prostate treatment plans using Monte Carlo simulations and an anatomically realistic computational phantom.

There is growing concern about radiation-induced second cancers associated with radiation treatments. Particular attention has been focused on the risk to patients treated with intensity-modulated radiation therapy (IMRT) due primarily to increased monitor units. To address this concern we have combined a detailed medical linear accelerator model of the Varian Clinac 2100 C with anatomically realistic computational phantoms to calculate organ doses from selected treatment plans. This paper describes the application to calculate organ-averaged equivalent doses using a computational phantom for three different treatments of prostate cancer: a 4-field box treatment, the same box treatment plus a 6-field 3D-CRT boost treatment and a 7-field IMRT treatment. The equivalent doses per MU to those organs that have shown a predilection for second cancers were compared between the different treatment techniques. In addition, the dependence of photon and neutron equivalent doses on gantry angle and energy was investigated. The results indicate that the box treatment plus 6-field boost delivered the highest intermediate- and low-level photon doses per treatment MU to the patient primarily due to the elevated patient scatter contribution as a result of an increase in integral dose delivered by this treatment. In most organs the contribution of neutron dose to the total equivalent dose for the 3D-CRT treatments was less than the contribution of photon dose, except for the lung, esophagus, thyroid and brain. The total equivalent dose per MU to each organ was calculated by summing the photon and neutron dose contributions. For all organs non-adjacent to the primary beam, the equivalent doses per MU from the IMRT treatment were less than the doses from the 3D-CRT treatments. This is due to the increase in the integral dose and the added neutron dose to these organs from the 18 MV treatments. However, depending on the application technique and optimization used, the required MU values for IMRT treatments can be two to three times greater than 3D CRT. Therefore, the total equivalent dose in most organs would be higher from the IMRT treatment compared to the box treatment and comparable to the organ doses from the box treatment plus the 6-field boost. This is the first time when organ dose data for an adult male patient of the ICRP reference anatomy have been calculated and documented. The tools presented in this paper can be used to estimate the second cancer risk to patients undergoing radiation treatment.