Deep-Learning for Rapid Estimation of the Out-of-Field Dose in External Beam Photon Radiation Therapy - A Proof of Concept.

PURPOSE: The dose deposited outside of the treatment field during external photon beam radiation therapy treatment, also known as out-of-field dose, is the subject of extensive study as it may be associated with a higher risk of developing a second cancer and could have deleterious effects on the immune system that compromise the efficiency of combined radio-immunotherapy treatments. Out-of-field dose estimation tools developed today in research, including Monte Carlo simulations and analytical methods, are not suited to the requirements of clinical implementation because of their lack of versatility and their cumbersome application. We propose a proof of concept based on deep learning for out-of-field dose map estimation that addresses these limitations. METHODS AND MATERIALS: For this purpose, a 3D U-Net, considering as inputs the in-field dose, as computed by the treatment planning system, and the patient's anatomy, was trained to predict out-of-field dose maps. The cohort used for learning and performance evaluation included 3151 pediatric patients from the FCCSS database, treated in 5 clinical centers, whose whole-body dose maps were previously estimated with an empirical analytical method. The test set, composed of 433 patients, was split into 5 subdata sets, each containing patients treated with devices unseen during the training phase. Root mean square deviation evaluated only on nonzero voxels located in the out-of-field areas was computed as performance metric. RESULTS: Root mean square deviations of 0.28 and 0.41 cGy/Gy were obtained for the training and validation data sets, respectively. Values of 0.27, 0.26, 0.28, 0.30, and 0.45 cGy/Gy were achieved for the 6 MV linear accelerator, 16 MV linear accelerator, Alcyon cobalt irradiator, Mobiletron cobalt irradiator, and betatron device test sets, respectively. CONCLUSIONS: This proof-of-concept approach using a convolutional neural network has demonstrated unprecedented generalizability for this task, although it remains limited, and brings us closer to an implementation compatible with clinical routine.

Anatomy-based correction of kidney PVE on [Formula: see text] SPECT images.

BACKGROUND: In peptide receptor radionuclide therapy (PRRT), accurate quantification of kidney activity on post-treatment SPECT images paves the way for patient-specific treatment. Due to the limited spatial resolution of SPECT images, the partial volume effect (PVE) is a significant source of quantitative bias. In this study, we aimed to evaluate the performance and robustness of anatomy-based partial volume correction (PVC) algorithms to recover the accurate activity concentration of realistic kidney geometries on [Formula: see text]Lu SPECT images recorded under clinical conditions. METHODS: Based on the CT scan data from patients, three sets of fillable kidneys with surface-to-volume (S:V) ratios ranging from 1.5 to 2.8 cm-1, were 3D printed and attached in a IEC phantom. Quantitative [Formula: see text]Lu SPECT/CT acquisitions were performed on a GE Discovery NM CT 870 DR camera for the three modified IEC phantoms and for 6 different Target-To-Background ratios (TBRs: 2, 4, 6, 8, 10, 12). Two region-based (GTM and Labbé) and five voxel-based (GTM + MTC, Labbé + MTC, GTM + RBV, Labbé + RBV and IY) methods were evaluated with this data set. Additionally, the robustness of PVC methods to Point Spread Function (PSF) discrepancies, registration mismatches and background heterogeneity was evaluated. RESULTS: Without PVC, the average kidney RCs across all TBRs ranged from 0.66 ± 0.05 (smallest kidney) to 0.80 ± 0.03 (largest kidney). For a TBR of 12, all anatomy-based method were able to recover the kidneys activity concentration with an error < 6%. All methods result in a comparable decline in RC restoration with decreasing TBR. The Labbé method was the most robust against PSF and registration mismatches but was also the most sensitive to background heterogeneity. Among the voxel-based methods, MTC images were less uniform than RBV and IY images at the outer edge of high uptake areas (kidneys and spheres). CONCLUSION: Anatomy-based PVE correction allows for accurate SPECT quantification of the [Formula: see text]Lu activity concentration with realistic kidney geometries. Combined with recent progress in deep-learning algorithms for automatic anatomic segmentation of whole-body CT, these methods could be of particular interest for a fully automated OAR dosimetry pipeline with PVE correction.

Ultra-hypofractionated prostate cancer radiotherapy: Dosimetric impact of real-time intrafraction prostate motion and daily anatomical changes.

PURPOSE: To retrospectively assess the differences between planned and delivered dose during ultra-hypofractionated (UHF) prostate cancer treatments, by evaluating the dosimetric impact of daily anatomical variations alone, and in combination with prostate intrafraction motion. METHODS: Prostate intrafraction motion was recorded with a transperineal ultrasound probe in 15 patients treated by UHF radiotherapy (36.25 Gy/5 fractions). The dosimetric objective was to cover 99 % of the clinical target volume with the 100 % prescription isodose line. After treatment, planning CT (pCT) images were deformably registered onto daily Cone Beam CT to generate pseudo-CT for dose accumulation (accumulated CT, aCT). The interplay effect was accounted by synchronizing prostatic shifts and beam geometry. Finally, the shifted dose maps were accumulated (moved-accumulated CT, maCT). RESULTS: No significant change in daily CTV volumes was observed. Conversely, CTV V100% was 98.2 ± 0.8 % and 94.7 ± 2.6 % on aCT and maCT, respectively, compared with 99.5 ± 0.2 % on pCT (p < 0.0001). Bladder volume was smaller than planned in 76 % of fractions and D5cc was 33.8 ± 3.2 Gy and 34.4 ± 3.4 Gy on aCT (p = 0.02) and maCT (p = 0.01) compared with the pCT (36.0 ± 1.1 Gy). The rectum was smaller than planned in 50.3 % of fractions, but the dosimetric differences were not statistically significant, except for D1cc, found smaller on the maCT (33.2 ± 3.2 Gy, p = 0.02) compared with the pCT (35.3 ± 0.7 Gy). CONCLUSIONS: Anatomical variations and prostate movements had more important dosimetric impact than anatomical variations alone, although, in some cases, the two phenomena compensated. Therefore, an efficient IGRT protocol is required for treatment implementation to reduce setup errors and control intrafraction motion.

Performance study of a 360° CZT camera for monitoring 177Lu-PSMA treatment.

BACKGROUND: The aim of this study was to investigate the quantification performance of a 360° CZT camera for 177Lu-based treatment monitoring. METHODS: Three phantoms with known 177Lu activity concentrations were acquired: (1) a uniform cylindrical phantom for calibration, (2) a NEMA IEC body phantom for analysis of different-sized spheres to optimise quantification parameters and (3) a phantom containing two large vials simulating organs at risk for tests. Four sets of reconstruction parameters were tested: (1) Scatter, (2) Scatter and Point Spread Function Recovery (PSFR), (3) PSFR only and (4) Penalised likelihood option and Scatter, varying the number of updates (iterations × subsets) with CT-based attenuation correction only. For each, activity concentration (ARC) and contrast recovery coefficients (CRC) were estimated as well as root mean square. Visualisation and quantification parameters were applied to reconstructed patient image data. RESULTS: Optimised quantification parameters were determined to be: CT-based attenuation correction, scatter correction, 12 iterations, 8 subsets and no filter. ARC, CRC and RMS results were dependant on the methodology used for calculations. Two different reconstruction parameters were recommended for visualisation and for quantification. 3D whole-body SPECT images were acquired and reconstructed for 177Lu-PSMA patients in 2-3 times faster than the time taken for a conventional gamma camera. CONCLUSION: Quantification of whole-body 3D images of patients treated with 177Lu-PSMA is feasible and an optimised set of parameters has been determined. This camera greatly reduces procedure time for whole-body SPECT.

A radiomic- and dosiomic-based machine learning regression model for pretreatment planning in 177 Lu-DOTATATE therapy.

BACKGROUND: Standardized patient-specific pretreatment dosimetry planning is mandatory in the modern era of nuclear molecular radiotherapy, which may eventually lead to improvements in the final therapeutic outcome. Only a comprehensive definition of a dosage therapeutic window encompassing the range of absorbed doses, that is, helpful without being detrimental can lead to therapy individualization and improved outcomes. As a result, setting absorbed dose safety limits for organs at risk (OARs) requires knowledge of the absorbed dose-effect relationship. Data sets of consistent and reliable inter-center dosimetry findings are required to characterize this relationship. PURPOSE: We developed and standardized a new pretreatment planning model consisting of a predictive dosimetry procedure for OARs in patients with neuroendocrine tumors (NETs) treated with 177 Lu-DOTATATE (Lutathera). In the retrospective study described herein, we used machine learning (ML) regression algorithms to predict absorbed doses in OARs by exploiting a combination of radiomic and dosiomic features extracted from patients' imaging data. METHODS: Pretreatment and posttreatment data for 20 patients with NETs treated with 177 Lu-DOTATATE were collected from two clinical centers. A total of 3412 radiomic and dosiomic features were extracted from the patients' computed tomography (CT) scans and dose maps, respectively. All dose maps were generated using Monte Carlo simulations. An ML regression model was designed based on ML algorithms for predicting the absorbed dose in every OAR (liver, left kidney, right kidney, and spleen) before and after the therapy and between each therapy session, thus predicting any possible radiotoxic effects. RESULTS: We evaluated nine ML regression algorithms. Our predictive model achieved a mean absolute dose error (MAE, in Gy) of 0.61 for the liver, 1.58 for the spleen, 1.30 for the left kidney, and 1.35 for the right kidney between pretherapy 68 Ga-DOTATOC positron emission tomography (PET)/CT and posttherapy 177 Lu-DOTATATE single photon emission (SPECT)/CT scans. Τhe best predictive performance observed was based on the gradient boost for the liver, the left kidney and the right kidney, and on the extra tree regressor for the spleen. Evaluation of the model's performance according to its ability to predict the absorbed dose in each OAR in every possible combination of pretherapy 68 Ga-DOTATOC PET/CT and any posttherapy 177 Lu-DOTATATE treatment cycle SPECT/CT scans as well as any 177 Lu-DOTATATE SPECT/CT treatment cycle and the consequent 177 Lu-DOTATATE SPECT/CT treatment cycle revealed mean absorbed dose differences ranges from -0.55 to 0.68 Gy. Incorporating radiodosiomics features from the 68 Ga-DOTATOC PET/CT and first 177 Lu-DOTATATE SPECT/CT treatment cycle scans further improved the precision and minimized the standard deviation of the predictions in nine out of 12 instances. An average improvement of 57.34% was observed (range: 17.53%-96.12%). However, it's important to note that in three instances (i.e., Ga,C.1 â†’ C3 in spleen and left kidney, and Ga,C.1 â†’ C2 in right kidney) we did not observe an improvement (absolute differences of 0.17, 0.08, and 0.05 Gy, respectively). Wavelet-based features proved to have high correlated predictive value, whereas non-linear-based ML regression algorithms proved to be more capable than the linear-based of producing precise prediction in our case. CONCLUSIONS: The combination of radiomics and dosiomics has potential utility for personalized molecular radiotherapy (PMR) response evaluation and OAR dose prediction. These radiodosiomic features can potentially provide information on any possible disease recurrence and may be highly useful in clinical decision-making, especially regarding dose escalation issues.

Evaluation of different algorithms for automatic segmentation of head-and-neck lymph nodes on CT images.

PURPOSE: To investigate the performance of 4 atlas-based (multi-ABAS) and 2 deep learning (DL) solutions for head-and-neck (HN) elective nodes (CTVn) automatic segmentation (AS) on CT images. MATERIAL AND METHODS: Bilateral CTVn levels of 69 HN cancer patients were delineated on contrast-enhanced planning CT. Ten and 49 patients were used for atlas library and for training a mono-centric DL model, respectively. The remaining 20 patients were used for testing. Additionally, three commercial multi-ABAS methods and one commercial multi-centric DL solution were investigated. Quantitative evaluation was assessed using volumetric Dice Similarity Coefficient (DSC) and 95-percentile Hausdorff distance (HD95%). Blind evaluation was performed for 3 solutions by 4 physicians. One recorded the time needed for manual corrections. A dosimetric study was finally conducted using automated planning. RESULTS: Overall DL solutions had better DSC and HD95% results than multi-ABAS methods. No statistically significant difference was found between the 2 DL solutions. However, the contours provided by multi-centric DL solution were preferred by all physicians and were also faster to correct (1.1 min vs 4.17 min, on average). Manual corrections for multi-ABAS contours took on average 6.52 min Overall, decreased contour accuracy was observed from CTVn2 to CTVn3 and to CTVn4. Using the AS contours in treatment planning resulted in underdosage of the elective target volume. CONCLUSION: Among all methods, the multi-centric DL method showed the highest delineation accuracy and was better rated by experts. Manual corrections remain necessary to avoid elective target underdosage. Finally, AS contours help reducing the workload of manual delineation task.

Contour-guided deep learning based deformable image registration for dose monitoring during CBCT-guided radiotherapy of prostate cancer.

PURPOSE: To evaluate deep learning (DL)-based deformable image registration (DIR) for dose accumulation during radiotherapy of prostate cancer patients. METHODS AND MATERIALS: Data including 341 CBCTs (209 daily, 132 weekly) and 23 planning CTs from 23 patients was retrospectively analyzed. Anatomical deformation during treatment was estimated using free-form deformation (FFD) method from Elastix and DL-based VoxelMorph approaches. The VoxelMorph method was investigated using anatomical scans (VMorph_Sc) or label images (VMorph_Msk), or the combination of both (VMorph_Sc_Msk). Accumulated doses were compared with the planning dose. RESULTS: The DSC ranges, averaged for prostate, rectum and bladder, were 0.60-0.71, 0.67-0.79, 0.93-0.98, and 0.89-0.96 for the FFD, VMorph_Sc, VMorph_Msk, and VMorph_Sc_Msk methods, respectively. When including both anatomical and label images, VoxelMorph estimated more complex deformations resulting in heterogeneous determinant of Jacobian and higher percentage of deformation vector field (DVF) folding (up to a mean value of 1.90% in the prostate). Large differences were observed between DL-based methods regarding estimation of the accumulated dose, showing systematic overdosage and underdosage of the bladder and rectum, respectively. The difference between planned mean dose and accumulated mean dose with VMorph_Sc_Msk reached a median value of +6.3 Gy for the bladder and -5.1 Gy for the rectum. CONCLUSION: The estimation of the deformations using DL-based approach is feasible for male pelvic anatomy but requires the inclusion of anatomical contours to improve organ correspondence. High variability in the estimation of the accumulated dose depending on the deformable strategy suggests further investigation of DL-based techniques before clinical deployment.

Management of reirradiations: A clinical and technical overview based on a French survey.

INTRODUCTION: The reirradiation number increased due to systemic therapies and patient survival. Few guidelines regarding acceptable cumulative doses to organs at risk (OARs) and appropriate dose accumulation tools need, made reirradiation challenging. The survey objective was to present the French current technical and clinical practices in reirradiations. METHODS: A group of physician and physicists developed a survey gathering major issues of the topic. The questionnaire consisted in 4 parts: data collection, demographic, clinical and technical aspects. It was delivered through the SFRO and the SFPM. Data collection lasted 2 months and were gathered to compute statistical analysis. RESULTS: 48 institutions answered the survey. Difficulties about patient data collection were related to patient safety, administrative and technical limitations. Half of the institutions discussed reirradiation cases during a multidisciplinary meeting. It mainly aimed at discussing the indication and the new treatment total dose (92%). 79% of the respondents used various references but only 6% of them were specific to reirradiations. Patients with pain and clinical deficit were ranked as best inclusion criteria. 54.2% of the institutions considered OARs recovery, especially for spinal cord and brainstem. A commercial software was used for dose accumulation for 52% of respondents. Almost all institutions performed equivalent dose conversion (94%). A quarter of the institutions estimated not to have the appropriate equipment for reirradiation. CONCLUSION: This survey showed the various approaches and tools used in reirradiation management. It highlighted issues in collecting data, and the guidelines necessity for safe practices, to increase clinicians confidence in retreating patients.

From netrin-1-targeted SPECT/CT to internal radiotherapy for management of advanced solid tumors.

Targeted radionuclide therapy is a revolutionary tool for the treatment of highly spread metastatic cancers. Most current approaches rely on the use of vectors to deliver radionuclides to tumor cells, targeting membrane-bound cancer-specific moieties. Here, we report the embryonic navigation cue netrin-1 as an unanticipated target for vectorized radiotherapy. While netrin-1, known to be re-expressed in tumoral cells to promote cancer progression, is usually characterized as a diffusible ligand, we demonstrate here that netrin-1 is actually poorly diffusible and bound to the extracellular matrix. A therapeutic anti-netrin-1 monoclonal antibody (NP137) has been preclinically developed and was tested in various clinical trials showing an excellent safety profile. In order to provide a companion test detecting netrin-1 in solid tumors and allowing the selection of therapy-eligible patients, we used the clinical-grade NP137 agent and developed an indium-111-NODAGA-NP137 single photon emission computed tomography (SPECT) contrast agent. NP137-111 In provided specific detection of netrin-1-positive tumors with an excellent signal-to-noise ratio using SPECT/CT imaging in different mouse models. The high specificity and strong affinity of NP137 paved the way for the generation of lutetium-177-DOTA-NP137, a novel vectorized radiotherapy, which specifically accumulated in netrin-1-positive tumors. We demonstrate here, using tumor cell-engrafted mouse models and a genetically engineered mouse model, that a single systemic injection of NP137-177 Lu provides important antitumor effects and prolonged mouse survival. Together, these data support the view that NP137-111 In and NP137-177 Lu may represent original and unexplored imaging and therapeutic tools against advanced solid cancers.

Dosimetric impact of 3D motion-compensated SPECT reconstruction for SIRT planning.

BACKGROUND: In selective internal radiation therapy, 99mTc SPECT images are used to optimize patient treatment planning, but they are affected by respiratory motion. In this study, we evaluated on patient data the dosimetric impact of motion-compensated SPECT reconstruction on several volumes of interest (VOI), on the tumor-to-normal liver (TN) ratio and on the activity to be injected. METHODS: Twenty-nine patients with liver cancer or hepatic metastases treated by radioembolization were included in this study. The biodistribution of 90Y is assumed to be the same as that of 99mTc when predictive dosimetry is implemented. A total of 31 99mTc SPECT images were acquired and reconstructed with two methods: conventional OSEM (3D) and motion-compensated OSEM (3Dcomp). Seven VOI (liver, lungs, tumors, perfused liver, hepatic reserve, healthy perfused liver and healthy liver) were delineated on the CT or obtained by thresholding SPECT images followed by Boolean operations. Absorbed doses were calculated for each reconstruction using Monte Carlo simulations. Percentages of dose difference (PDD) between 3Dcomp and 3D reconstructions were estimated as well as the relative differences for TN ratio and activities to be injected. The amplitude of movement was determined with local rigid registration of the liver between the 3Dcomp reconstructions of the extreme phases of breathing. RESULTS: The mean amplitude of the liver was 9.5 ± 2.7 mm. Medians of PDD were closed to zero for all VOI except for lungs (6.4%) which means that the motion compensation overestimates the absorbed dose to the lungs compared to the 3D reconstruction. The smallest lesions had higher PDD than the largest ones. Between 3D and 3Dcomp reconstructions, means of differences in lung dose and TN ratio were not statistically significant, but in some cases these differences exceed 1 Gy (4/31) and 8% (2/31). The absolute differences in activity were on average 3.1% ± 5.1% and can reach 22.8%. CONCLUSION: The correction of respiratory motion mainly impacts the lung and tumor doses but only for some patients. The largest dose differences are observed for the smallest lesions.

Comparison of atlas-based and deep learning methods for organs at risk delineation on head-and-neck CT images using an automated treatment planning system.

BACKGROUND AND PURPOSE: To investigate the performance of head-and-neck (HN) organs-at-risk (OAR) automatic segmentation (AS) using four atlas-based (ABAS) and two deep learning (DL) solutions. MATERIAL AND METHODS: All patients underwent iodine contrast-enhanced planning CT. Fourteen OAR were manually delineated. DL.1 and DL.2 solutions were trained with 63 mono-centric patients and > 1000 multi-centric patients, respectively. Ten and 15 patients with varied anatomies were selected for the atlas library and for testing, respectively. The evaluation was based on geometric indices (DICE coefficient and 95th percentile-Hausdorff Distance (HD95%)), time needed for manual corrections and clinical dosimetric endpoints obtained using automated treatment planning. RESULTS: Both DICE and HD95% results indicated that DL algorithms generally performed better compared with ABAS algorithms for automatic segmentation of HN OAR. However, the hybrid-ABAS (ABAS.3) algorithm sometimes provided the highest agreement to the reference contours compared with the 2 DL. Compared with DL.2 and ABAS.3, DL.1 contours were the fastest to correct. For the 3 solutions, the differences in dose distributions obtained using AS contours and AS + manually corrected contours were not statistically significant. High dose differences could be observed when OAR contours were at short distances to the targets. However, this was not always interrelated. CONCLUSION: DL methods generally showed higher delineation accuracy compared with ABAS methods for AS segmentation of HN OAR. Most ABAS contours had high conformity to the reference but were more time consuming than DL algorithms, especially when considering the computing time and the time spent on manual corrections.

Modeling families of particle distributions with conditional GAN for Monte Carlo SPECT simulations.

Objective.We propose a method to model families of distributions of particles exiting a phantom with a conditional generative adversarial network (condGAN) during Monte Carlo simulation of single photon emission computed tomography imaging devices.Approach.The proposed condGAN is trained on a low statistics dataset containing the energy, the time, the position and the direction of exiting particles. In addition, it also contains a vector of conditions composed of four dimensions: the initial energy and the position of emitted particles within the phantom (a total of 12 dimensions). The information related to the gammas absorbed within the phantom is also added in the dataset. At the end of the training process, one component of the condGAN, the generator (G), is obtained.Main results.Particles with specific energies and positions of emission within the phantom can then be generated withGto replace the tracking of particle within the phantom, allowing reduced computation time compared to conventional Monte Carlo simulation.Significance.The condGAN generator is trained only once for a given phantom but can generate particles from various activity source distributions.

Imaging of polychromatic sources through Compton spectral reconstruction.

Objective.Study the performance of a spectral reconstruction method for Compton imaging of polychromatic sources and compare it to standard Compton reconstruction based on the selection of photopeak events.Approach.The proposed spectral and the standard photopeak reconstruction methods are used to reconstruct images from simulated sources emitting simultaneously photons of 140, 245, 364 and 511 keV. Data are simulated with perfect and realistic energy resolutions and including Doppler broadening. We compare photopeak and spectral reconstructed images both qualitatively and quantitatively by means of activity recovery coefficient and spatial resolution.Main results.The presented method allows improving the images of polychromatic sources with respect to standard reconstruction methods. The main reasons for this improvement are the increase of available statistics and the reduction of contamination from higher initial photon energies. The reconstructed images present lower noise, higher activity recovery coefficient and better spatial resolution. The improvements become more sensible as the energy resolution of the detectors decreases.Significance.Compton cameras have been studied for their capability of imaging polychromatic sources, thus allowing simultaneous imaging of multiple radiotracers. In such scenarios, Compton images are conventionally reconstructed for each emission energy independently, selecting only those measured events depositing a total energy within a fixed window around the known emission lines. We propose to employ a spectral image reconstruction method for polychromatic sources, which allows increasing the available statistics by using the information from events with partial energy deposition. The detector energy resolution influences the energy window used to select photopeak events and therefore the level of contamination by higher energies. The spectral method is expected to have a more important impact as the detector resolution worsens. In this paper we focus on energy ranges from nuclear medical imaging and we consider realistic energy resolutions.

The OpenGATE ecosystem for Monte Carlo simulation in medical physics.

This paper reviews the ecosystem of GATE, an open-source Monte Carlo toolkit for medical physics. Based on the shoulders of Geant4, the principal modules (geometry, physics, scorers) are described with brief descriptions of some key concepts (Volume, Actors, Digitizer). The main source code repositories are detailed together with the automated compilation and tests processes (Continuous Integration). We then described how the OpenGATE collaboration managed the collaborative development of about one hundred developers during almost 20 years. The impact of GATE on medical physics and cancer research is then summarized, and examples of a few key applications are given. Finally, future development perspectives are indicated.

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.

Patient-specific dosimetry adapted to variable number of SPECT/CT time-points per cycle for [Formula: see text]Lu-DOTATATE therapy.

BACKGROUND: The number of SPECT/CT time-points is important for accurate patient dose estimation in peptide receptor radionuclide therapy. However, it may be limited by the patient's health and logistical reasons. Here,  an image-based dosimetric workflow adapted to the number of SPECT/CT acquisitions available throughout the treatment cycles was proposed, taking into account patient-specific pharmacokinetics and usable in clinic for all organs at risk. METHODS: Thirteen patients with neuroendocrine tumors were treated with four injections of 7.4 GBq of [Formula: see text]Lu-DOTATATE. Three SPECT/CT images were acquired during the first cycle (1H, 24H and 96H or 144H post-injection) and a single acquisition (24H) for following cycles. Absorbed doses were estimated for kidneys (LK and RK), liver (L), spleen (S), and three surrogates of bone marrow (L2 to L4, L1 to L5 and T9 to L5) that were compared. 3D dose rate distributions were computed with Monte Carlo simulations. Voxel dose rates were averaged at the organ level. The obtained Time Dose-Rate Curves (TDRC) were fitted with a tri-exponential model and time-integrated. This method modeled patient-specific uptake and clearance phases observed at cycle 1. Obtained fitting parameters were reused for the following cycles, scaled to the measure organ dose rate at 24H. An alternative methodology was proposed when some acquisitions were missing based on population average TDRC (named STP-Inter). Seven other patients with three SPECT/CT acquisitions at cycles 1 and 4 were included to estimate the uncertainty of the proposed methods. RESULTS: Absorbed doses (in Gy) per cycle available were: 3.1 ± 1.1 (LK), 3.4 ± 1.5 (RK), 4.5 ± 2.8 (L), 4.6 ± 1.8 (S), 0.3 ± 0.2 (bone marrow). There was a significant difference between bone marrow surrogates (L2 to L4 and L1 to L5, Wilcoxon's test: p value < 0.05), and while depicting very doses, all three surrogates were significantly different than dose in background (p value < 0.01). At cycle 1, if the acquisition at 24H is missing and approximated, medians of percentages of dose difference (PDD) compared to the initial tri-exponential function were inferior to 3.3% for all organs. For cycles with one acquisition, the median errors were smaller with a late time-point. For STP-Inter, medians of PDD were inferior to 7.7% for all volumes, but it was shown to depend on the homogeneity of TDRC. CONCLUSION: The proposed workflow allows the estimation of organ doses, including bone marrow, from a variable number of time-points acquisitions for patients treated with [Formula: see text]Lu-DOTATATE.

Minimum non-isotropic and asymmetric margins for taking into account intrafraction prostate motion during moderately hypofractionated radiotherapy.

PURPOSE: To investigate the impact on dose distribution of intrafraction motion during moderate hypofractionated prostate cancer treatments and to estimate minimum non-isotropic and asymmetric (NI-AS) treatment margins taking motion into account. METHODS: Prostate intrafraction 3D displacements were recorded with a transperineal ultrasound probe and were evaluated in 46 prostate cancer patients (876 fractions) treated by moderate hypofractionated radiation therapy (60 Gy in 20 fractions). For 18 patients (346 fractions), treatment plans were recomputed increasing CTV-to-PTV margins from 0 to 6 mm with an auto-planning optimization algorithm. Dose distribution was estimated using the voxel shifting method by displacing CTV structure according to the retrieved movements. Time-dependent margins were finally calculated using both van Herk's formula and the voxel shifting method. RESULTS: Mean intrafraction prostate displacements observed were -0.02 ± 0.52 mm, 0.27 ± 0.78 mm and -0.43 ± 1.06 mm in left-right, supero-inferior and antero-posterior directions, respectively. The CTV dosimetric coverage increased with increased CTV-to-PTV margins but it decreased with time. Hence using van Herk's formula, after 7 min of treatment, a margin of 0.4 and 0.5 mm was needed in left and right, 1.5 and 0.7 mm in inferior and superior and 1.1 and 3.2 mm in anterior and posterior directions, respectively. Conversely, using the voxel shifting method, a margin of 0 mm was needed in left-right, 2 mm in superior, 3 mm in inferior and anterior and 5 mm in posterior directions, respectively. With this latter NI-AS margin strategy, the dosimetric target coverage was equivalent to the one obtained with a 5 mm homogeneous margin. CONCLUSIONS: NI-AS margins would be required to optimally take into account intrafraction motion.

Improving motion-mask segmentation in thoracic CT with multiplanar U-nets.

PURPOSE: Motion-mask segmentation from thoracic computed tomography (CT) images is the process of extracting the region that encompasses lungs and viscera, where large displacements occur during breathing. It has been shown to help image registration between different respiratory phases. This registration step is, for example, useful for radiotherapy planning or calculating local lung ventilation. Knowing the location of motion discontinuity, that is, sliding motion near the pleura, allows a better control of the registration preventing unrealistic estimates. Nevertheless, existing methods for motion-mask segmentation are not robust enough to be used in clinical routine. This article shows that it is feasible to overcome this lack of robustness by using a lightweight deep-learning approach usable on a standard computer, and this even without data augmentation or advanced model design. METHODS: A convolutional neural-network architecture with three 2D U-nets for the three main orientations (sagittal, coronal, axial) was proposed. Predictions generated by the three U-nets were combined by majority voting to provide a single 3D segmentation of the motion mask. The networks were trained on a database of nonsmall cell lung cancer 4D CT images of 43 patients. Training and evaluation were done with a K-fold cross-validation strategy. Evaluation was based on a visual grading by two experts according to the appropriateness of the segmented motion mask for the registration task, and on a comparison with motion masks obtained by a baseline method using level sets. A second database (76 CT images of patients with early-stage COVID-19), unseen during training, was used to assess the generalizability of the trained neural network. RESULTS: The proposed approach outperformed the baseline method in terms of quality and robustness: the success rate increased from 53 % to 79 % without producing any failure. It also achieved a speed-up factor of 60 with GPU, or 17 with CPU. The memory footprint was low: less than 5 GB GPU RAM for training and less than 1 GB GPU RAM for inference. When evaluated on a dataset with images differing by several characteristics (CT device, pathology, and field of view), the proposed method improved the success rate from 53 % to 83 % . CONCLUSION: With 5-s processing time on a mid-range GPU and success rates around 80 % , the proposed approach seems fast and robust enough to be routinely used in clinical practice. The success rate can be further improved by incorporating more diversity in training data via data augmentation and additional annotated images from different scanners and diseases. The code and trained model are publicly available.

Projection-based dynamic tomography.

Objective. This paper proposes a 4D dynamic tomography framework that allows a moving sample to be imaged in a tomograph as well as the associated space-time kinematics to be measured with nothing more than a single conventional x-ray scan.Approach. The method exploits the consistency of the projection/reconstruction operations through a multi-scale procedure. The procedure is composed of two main parts solved alternatively: a motion-compensated reconstruction algorithm and a projection-based measurement procedure that estimates the displacement field directly on each projection.Main results. The method is applied to two studies: a numerical simulation of breathing from chest computed tomography (4D-CT) and a clinical cone-beam CT of a breathing patient acquired for image guidance of radiotherapy. The reconstructed volume, initially blurred by the motion, is cleaned from motion artifacts.Significance. Applying the proposed approach results in an improved reconstruction quality showing sharper edges and finer details.

Yttrium-90 quantitative phantom study using digital photon counting PET.

BACKGROUND: PET imaging of 90Y-microsphere distribution following radioembolisation is challenging due to the count-starved statistics from the low branching ratio of e+/e- pair production during 90Y decay. PET systems using silicon photo-multipliers have shown better 90Y image quality compared to conventional photo-multiplier tubes. The main goal of the present study was to evaluate reconstruction parameters for different phantom configurations and varying listmode acquisition lengths to improve quantitative accuracy in 90Y dosimetry, using digital photon counting PET/CT. METHODS: Quantitative PET and dosimetry accuracy were evaluated using two uniform cylindrical phantoms specific for PET calibration validation. A third body phantom with a 9:1 hot sphere-to-background ratio was scanned at different activity concentrations of 90Y. Reconstructions were performed using OSEM algorithm with varying parameters. Time-of-flight and point-spread function modellings were included in all reconstructions. Absorbed dose calculations were carried out using voxel S-values convolution and were compared to reference Monte Carlo simulations. Dose-volume histograms and root-mean-square deviations were used to evaluate reconstruction parameter sets. Using listmode data, phantom and patient datasets were rebinned into various lengths of time to assess the influence of count statistics on the calculation of absorbed dose. Comparisons between the local energy deposition method and the absorbed dose calculations were performed. RESULTS: Using a 2-mm full width at half maximum post-reconstruction Gaussian filter, the dosimetric accuracy was found to be similar to that found with no filter applied but also reduced noise. Larger filter sizes should not be used. An acquisition length of more than 10 min/bed reduces image noise but has no significant impact in the quantification of phantom or patient data for the digital photon counting PET. 3 iterations with 10 subsets were found suitable for large spheres whereas 1 iteration with 30 subsets could improve dosimetry for smaller spheres. CONCLUSION: The best choice of the combination of iterations and subsets depends on the size of the spheres. However, one should be careful on this choice, depending on the imaging conditions and setup. This study can be useful in this choice for future studies for more accurate 90Y post-dosimetry using a digital photon counting PET/CT.