Search

Background: Real‐time tumor tracking is one motion management method to address motion‐induced uncertainty. To date, fiducial markers are often required to reliably track lung tumors with X‐ray imaging, which carries risks of complications and leads to prolonged treatment time. A markerless tracking approach is thus desirable. Deep learning‐based approaches have shown promise for markerless tracking, but systematic evaluation and procedures to investigate applicability in individual cases are missing. Moreover, few efforts have been made to provide bounding box prediction and mask segmentation simultaneously, which could allow either rigid or deformable multi‐leaf collimator tracking. Purpose: The purpose of this study was to implement a deep learning‐based markerless lung tumor tracking model exploiting patient‐specific training which outputs both a bounding box and a mask segmentation simultaneously. We also aimed to compare the two kinds of predictions and to implement a specific procedure to understand the feasibility of markerless tracking on individual cases. Methods: We first trained a Retina U‐Net baseline model on digitally reconstructed radiographs (DRRs) generated from a public dataset containing 875 CT scans and corresponding lung nodule annotations. Afterwards, we used an independent cohort of 97 lung patients to develop a patient‐specific refinement procedure. In order to determine the optimal hyperparameters for automatic patient‐specific training, we selected 13 patients for validation where the baseline model predicted a bounding box on planning CT (PCT)‐DRR with intersection over union (IoU) with the ground‐truth higher than 0.7. The final test set contained the remaining 84 patients with varying PCT‐DRR IoU. For each testing patient, the baseline model was refined on the PCT‐DRR to generate a patient‐specific model, which was then tested on a separate 10‐phase 4DCT‐DRR to mimic the intrafraction motion during treatment. A template matching algorithm served as benchmark model. The testing results were evaluated by four metrics: the center of mass (COM) error and the Dice similarity coefficient (DSC) for segmentation masks, and the center of box (COB) error and the DSC for bounding box detections. Performance was compared to the benchmark model including statistical testing for significance. Results: A PCT‐DRR IoU value of 0.2 was shown to be the threshold dividing inconsistent (68%) and consistent (100%) success (defined as mean bounding box DSC > 0.6) of PS models on 4DCT‐DRRs. Thirty‐seven out of the eighty‐four testing cases had a PCT‐DRR IoU above 0.2. For these 37 cases, the mean COM error was 2.6 mm, the mean segmentation DSC was 0.78, the mean COB error was 2.7 mm, and the mean box DSC was 0.83. Including the validation cases, the model was applicable to 50 out of 97 patients when using the PCT‐DRR IoU threshold of 0.2. The inference time per frame was 170 ms. The model outperformed the benchmark model on all metrics, and the comparison was significant (p < 0.001) over the 37 PCT‐DRR IoU > 0.2 cases, but not over the undifferentiated 84 testing cases. Conclusions: The implemented patient‐specific refinement approach based on a pre‐trained baseline model was shown to be applicable to markerless tumor tracking in simulated radiographs for lung cases. [ABSTRACT FROM AUTHOR]

Background: Cone beam computed tomography (CBCT) is widely used in many medical fields. However, conventional CBCT circular scans suffer from cone beam (CB) artifacts that limit the quality and reliability of the reconstructed images due to incomplete data. Purpose: Saddle trajectories in theory might be able to improve the CBCT image quality by providing a larger region with complete data. Therefore, we investigated the feasibility and performance of saddle trajectory CBCT scans and compared them to circular trajectory scans. Methods: We performed circular and saddle trajectory scans using a novel robotic CBCT scanner (Mobile ImagingRing (IRm); medPhoton, Salzburg, Austria). For the saddle trajectory, the gantry executed yaw motion up to ±10∘$\pm 10^{\circ }$ using motorized wheels driving on the floor. An infrared (IR) tracking device with reflective markers was used for online geometric calibration correction (mainly floor unevenness). All images were reconstructed using penalized least‐squares minimization with the conjugate gradient algorithm from RTK with 0.5×0.5×0.5mm3$0.5 \times 0.5\times 0.5 \text{ mm}^3$ voxel size. A disk phantom and an Alderson phantom were scanned to assess the image quality. Results were correlated with the local incompleteness value represented by tan(ψ)$\tan (\psi)$, which was calculated at each voxel as a function of the source trajectory and the voxel's 3D coordinates. We assessed the magnitude of CB artifacts using the full width half maximum (FWHM) of each disk profile in the axial center of the reconstructed images. Spatial resolution was also quantified by the modulation transfer function at 10% (MTF10). Results: When using the saddle trajectory, the region without CB artifacts was increased from 43 to 190 mm in the SI direction compared to the circular trajectory. This region coincided with low values for tan(ψ)$\tan (\psi)$. When tan(ψ)$\tan (\psi)$ was larger than 0.02, we found there was a linear relationship between the FWHM and tan(ψ)$\tan (\psi)$. For the saddle, IR tracking allowed the increase of MTF10 from 0.37 to 0.98 lp/mm. Conclusions: We achieved saddle trajectory CBCT scans with a novel CBCT system combined with IR tracking. The results show that the saddle trajectory provides a larger region with reliable reconstruction compared to the circular trajectory. The proposed method can be used to evaluate other non‐circular trajectories. [ABSTRACT FROM AUTHOR]

Background: Current commercially available hybrid magnetic resonance linear accelerators (MR‐Linac) use 2D+t cine MR imaging to provide intra‐fractional motion monitoring. However, given the limited temporal resolution of cine MR imaging, target intra‐frame motion deterioration effects, resulting in effective time latency and motion artifacts in the image domain, can be appreciable, especially in the case of fast breathing. Purpose: The aim of this work is to investigate intra‐frame motion deterioration effects in MR‐guided radiotherapy (MRgRT) by simulating the motion‐corrupted image acquisition, and to explore the feasibility of deep‐learning‐based compensation approaches, relying on the intra‐frame motion information which is spatially and temporally encoded in the raw data (k‐space). Methods: An intra‐frame motion model was defined to simulate motion‐corrupted MR images, with 4D anthropomorphic digital phantoms being exploited to provide ground truth 2D+t cine MR sequences. A total number of 10 digital phantoms were generated for lung cancer patients, with randomly selected eight patients for training or validation and the remaining two for testing. The simulation code served as the data generator, and a dedicated motion pattern perturbation scheme was proposed to build the intra‐frame motion database, where three degrees of freedom were designed to guarantee the diversity of intra‐frame motion trajectories, enabling a thorough exploration in the domain of the potential anatomical structure positions. U‐Nets with three types of loss functions: L1 or L2 loss defined in image or Fourier domain, referred to as NNImgLoss‐L1, NNFloss‐L1 and NNL2‐Loss were trained to extract information from the motion‐corrupted image and used to estimate the ground truth final‐position image, corresponding to the end of the acquisition. Images before and after compensation were evaluated in terms of (i) image mean‐squared error (MSE) and mean absolute error (MAE), and (ii) accuracy of gross tumor volume (GTV) contouring, based on optical‐flow image registration. Results: Image degradation caused by intra‐frame motion was observed: for a linearly and fully acquired Cartesian readout k‐space trajectory, intra‐frame motion resulted in an imaging latency of approximately 50% of the acquisition time; in comparison, the motion artifacts exhibited only a negligible contribution to the overall geometric errors. All three compensation models led to a decrease in image MSE/MAE and GTV position offset compared to the motion‐corrupted image. In the investigated testing dataset for GTV contouring, the average dice similarity coefficients (DSC) improved from 88% to 96%, and the 95th percentile Hausdorff distance (HD95) dropped from 4.8 mm to 2.1 mm. Different models showed slight performance variations across different intra‐frame motion amplitude categories: NNImgLoss‐L1 excelled for small/medium amplitudes, whereas NNFloss‐L1 demonstrated higher DSC median values at larger amplitudes. The saliency maps of the motion‐corrupted image highlighted the major contribution of the later acquired k‐space data, as well as the edges of the moving anatomical structures at their final positions, during the model inference stage. Conclusions: Our results demonstrate the deep‐learning‐based approaches have the potential to compensate for intra‐frame motion by utilizing the later acquired data to drive the convergence of the earlier acquired k‐space components. [ABSTRACT FROM AUTHOR]

Background: Magnetic resonance imaging (MRI)‐guided radiotherapy with multileaf collimator (MLC)‐tracking is a promising technique for intra‐fractional motion management, achieving high dose conformality without prolonging treatment times. To improve beam‐target alignment, the geometric error due to system latency should be reduced by using temporal prediction. Purpose: To experimentally compare linear regression (LR) and long‐short‐term memory (LSTM) motion prediction models for MLC‐tracking on an MRI‐linac using multiple patient‐derived traces with different complexities. Methods: Experiments were performed on a prototype 1.0 T MRI‐linac capable of MLC‐tracking. A motion phantom was programmed to move a target in superior‐inferior (SI) direction according to eight lung cancer patient respiratory motion traces. Target centroid positions were localized from sagittal 2D cine MRIs acquired at 4 Hz using a template matching algorithm. The centroid positions were input to one of four motion prediction models. We used (1) a LSTM network which had been optimized in a previous study on patient data from another cohort (offline LSTM). We also used (2) the same LSTM model as a starting point for continuous re‐optimization of its weights during the experiment based on recent motion (offline+online LSTM). Furthermore, we implemented (3) a continuously updated LR model, which was solely based on recent motion (online LR). Finally, we used (4) the last available target centroid without any changes as a baseline (no‐predictor). The predictions of the models were used to shift the MLC aperture in real‐time. An electronic portal imaging device (EPID) was used to visualize the target and MLC aperture during the experiments. Based on the EPID frames, the root‐mean‐square error (RMSE) between the target and the MLC aperture positions was used to assess the performance of the different motion predictors. Each combination of motion trace and prediction model was repeated twice to test stability, for a total of 64 experiments. Results: The end‐to‐end latency of the system was measured to be (389 ± 15) ms and was successfully mitigated by both LR and LSTM models. The offline+online LSTM was found to outperform the other models for all investigated motion traces. It obtained a median RMSE over all traces of (2.8 ± 1.3) mm, compared to the (3.2 ± 1.9) mm of the offline LSTM, the (3.3 ± 1.4) mm of the online LR and the (4.4 ± 2.4) mm when using the no‐predictor. According to statistical tests, differences were significant (p‐value <0.05) among all models in a pair‐wise comparison, but for the offline LSTM and online LR pair. The offline+online LSTM was found to be more reproducible than the offline LSTM and the online LR with a maximum deviation in RMSE between two measurements of 10%. Conclusions: This study represents the first experimental comparison of different prediction models for MRI‐guided MLC‐tracking using several patient‐derived respiratory motion traces. We have shown that among the investigated models, continuously re‐optimized LSTM networks are the most promising to account for the end‐to‐end system latency in MRI‐guided radiotherapy with MLC‐tracking. [ABSTRACT FROM AUTHOR]

Background: The treatment of moving tumor entities is expected to have superior clinical outcomes, using image‐guided adaptive intensity‐modulated proton therapy (IMPT). Purpose: For 21 lung cancer patients, IMPT dose calculations were performed on scatter‐corrected 4D cone beam CTs (4DCBCTcor) to evaluate their potential for triggering treatment adaptation. Additional dose calculations were performed on corresponding planning 4DCTs and day‐of‐treatment 4D virtual CTs (4DvCTs). Methods: A 4DCBCT correction workflow, previously validated on a phantom, generates 4DvCT (CT‐to‐CBCT deformable registration) and 4DCBCTcor images (projection‐based correction using 4DvCT as a prior) with 10 phase bins, using day‐of‐treatment free‐breathing CBCT projections and planning 4DCT images as input. Using a research planning system, robust IMPT plans administering eight fractions of 7.5 Gy were created on a free‐breathing planning CT (pCT) contoured by a physician. The internal target volume (ITV) was overridden with muscle tissue. Robustness settings for range and setup uncertainties were 3% and 6 mm, and a Monte Carlo dose engine was used. On every phase of planning 4DCT, day‐of‐treatment 4DvCT, and 4DCBCTcor, the dose was recalculated. For evaluation, image analysis as well as dose analysis were performed using mean error (ME) and mean absolute error (MAE) analysis, dose–volume histogram (DVH) parameters, and 2%/2‐mm gamma pass rate analysis. Action levels (1.6% ITV D98 and 90% gamma pass rate) based on our previous phantom validation study were set to determine which patients had a loss of dosimetric coverage. Results: Quality enhancements of 4DvCT and 4DCBCTcor over 4DCBCT were observed. ITV D98% and bronchi D2% had its largest agreement for 4DCBCTcor–4DvCT, and the largest gamma pass rates (>94%, median 98%) were found for 4DCBCTcor–4DvCT. Deviations were larger and gamma pass rates were smaller for 4DvCT–4DCT and 4DCBCTcor–4DCT. For five patients, deviations were larger than the action levels, suggesting substantial anatomical changes between pCT and CBCT projections acquisition. Conclusions: This retrospective study shows the feasibility of daily proton dose calculation on 4DCBCTcor for lung tumor patients. The applied method is of clinical interest as it generates up‐to‐date in‐room images, accounting for breathing motion and anatomical changes. This information could be used to trigger replanning. [ABSTRACT FROM AUTHOR]

Background: Stereotactic body radiation therapy (SBRT) of central lung tumors with photon or proton therapy has a risk of increased toxicity. Treatment planning studies comparing accumulated doses for state‐of‐the‐art treatment techniques, such as MR‐guided radiotherapy (MRgRT) and intensity modulated proton therapy (IMPT), are currently lacking. Purpose: We conducted a comparison of accumulated doses for MRgRT, robustly optimized non‐adaptive IMPT, and online adaptive IMPT for central lung tumors. A special focus was set on analyzing the accumulated doses to the bronchial tree, a parameter linked to high‐grade toxicities. Methods: Data of 18 early‐stage central lung tumor patients, treated at a 0.35 T MR‐linac in eight or five fractions, were analyzed. Three gated treatment scenarios were compared: (S1) online adaptive MRgRT, (S2) non‐adaptive IMPT, and (S3) online adaptive IMPT. The treatment plans were recalculated or reoptimized on the daily imaging data acquired during MRgRT, and accumulated over all treatment fractions. Accumulated dose‐volume histogram (DVH) parameters of the gross tumor volume (GTV), lung, heart, and organs‐at‐risk (OARs) within 2 cm of the planning target volume (PTV) were extracted for each scenario and compared in Wilcoxon signed‐rank tests between S1 & S2, and S1 & S3. Results: The accumulated GTV D98% was above the prescribed dose for all patients and scenarios. Significant reductions (p < 0.05) of the mean ipsilateral lung dose (S2: –8%; S3: –23%) and mean heart dose (S2: –79%; S3: –83%) were observed for both proton scenarios compared to S1. The bronchial tree D0.1cc was significantly lower for S3 (S1: 48.1 Gy; S3: 39.2 Gy; p = 0.005), but not significantly different for S2 (S2: 45.0 Gy; p = 0.094), compared to S1. The D0.1cc for S2 and S3 compared to S1 was significantly (p < 0.05) smaller for OARs within 1–2 cm of the PTV (S1: 30.2 Gy; S2: 24.6 Gy; S3: 23.1 Gy), but not significantly different for OARs within 1 cm of the PTV. Conclusions: A significant dose sparing potential of non‐adaptive and online adaptive proton therapy compared to MRgRT for OARs in close, but not direct proximity of central lung tumors was identified. The near‐maximum dose to the bronchial tree was not significantly different for MRgRT and non‐adaptive IMPT. Online adaptive IMPT achieved significantly lower doses to the bronchial tree compared to MRgRT. [ABSTRACT FROM AUTHOR]

Background: Online adaptive radiation therapy (RT) using hybrid magnetic resonance linear accelerators (MR-Linacs) can administer a tailored radiation dose at each treatment fraction. Daily MR imaging followed by organ and target segmentation adjustments allow to capture anatomical changes, improve target volume coverage, and reduce the risk of side effects. The introduction of automatic segmentation techniques could help to further improve the online adaptive workflow by shortening the re-contouring time and reducing intra- and inter-observer variability. In fractionated RT, prior knowledge, such as planning images and manual expert contours, is usually available before irradiation, but not used by current artificial intelligence-based autocontouring approaches. Purpose: The goal of this study was to train convolutional neural networks (CNNs) for automatic segmentation of bladder, rectum (organs at risk, OARs), and clinical target volume (CTV) for prostate cancer patients treated at 0.35 T MR-Linacs. Furthermore, we tested the CNNs generalization on data from independent facilities and compared them with the MR-Linac treatment planning system (TPS) propagated structures currently used in clinics. Finally, expert planning delineations were utilized for patient- (PS) and facility-specific (FS) transfer learning to improve auto-segmentation of CTV and OARs on fraction images. Methods: In this study, data from fractionated treatments at 0.35 T MR-Linacs were leveraged to develop a 3D U-Net-based automatic segmentation. Cohort C1 had 73 planning images and cohort C2 had 19 planning and 240 fraction images. The baseline models (BMs) were trained solely on C1 planning data using 53 MRIs for training and 10 for validation. To assess their accuracy, the models were tested on three data subsets: (i) 10 C1 planning images not used for training, (ii) 19 C2 planning,and (iii) 240 C2 fraction images.BMs also served as a starting point for FS and PS transfer learning, where the planning images from C2 were used for network parameter fine tuning.The segmentation output of the different trained models was compared against expert ground truth by means of geometric metrics.Moreover, a trained physician graded the network segmentations as well as the segmentations propagated by the clinical TPS. Results: The BMs showed dice similarity coefficients (DSC) of 0.88(4) and 0.93(3) for the rectum and the bladder, respectively, independent of the facility. CTV segmentation with the BM was the best for intermediate- and high-risk cancer patients from C1 with DSC=0.84(5) and worst for C2 with DSC=0.74(7). The PS transfer learning brought a significant improvement in the CTV segmentation, yielding DSC=0.72(4) for post-prostatectomy and low-risk patients and DSC=0.88(5) for intermediate- and high-risk patients. The FS training did not improve the segmentation accuracy considerably. The physician’s assessment of the TPS-propagated versus network-generated structures showed a clear advantage of the latter. Conclusions: The obtained results showed that the presented segmentation technique has potential to improve automatic segmentation for MR-guided RT. [ABSTRACT FROM AUTHOR]

In modern radiotherapy (RT), especially for stereotactic radiotherapy or stereotactic radiosurgery treatments, image guidance is essential. Recently, the ExacTrac Dynamic (EXTD) system, a new combined surface‐guided RT and image‐guided RT (IGRT) system for patient positioning, monitoring, and tumor targeting, was introduced in clinical practice. The purpose of this study was to provide more information about the geometric accuracy of EXTD and its workflow in a clinical environment. The surface optical/thermal‐ and the stereoscopic X‐ray imaging positioning systems of EXTD was evaluated and compared to cone‐beam computed tomography (CBCT). Additionally, the congruence with the radiation isocenter was tested. A Winston Lutz test was executed several times over 1 year, and repeated end‐to‐end positioning tests were performed. The magnitude of the displacements between all systems, CBCT, stereoscopic X‐ray, optical‐surface imaging, and MV portal imaging was within the submillimeter range, suggesting that the image guidance provided by EXTD is accurate at any couch angle. Additionally, results from the evaluation of 14 patients with intracranial tumors treated with open‐face masks are reported, and limited differences with a maximum of 0.02 mm between optical/thermal‐ and stereoscopic X‐ray imaging were found. As the optical/thermal positioning system showed a comparable accuracy to other IGRT systems, and due to its constant monitoring capability, it can be an efficient tool for detecting intra‐fractional motion and for real‐time tracking of the surface position during RT. [ABSTRACT FROM AUTHOR]

Purpose: The implementation of volumetric in‐room imaging for online adaptive radiotherapy makes extensive testing of this image data for treatment planning necessary. Especially for proton beams the higher sensitivity to stopping power properties of the tissue results in more stringent requirements. Current approaches mainly focus on recalculation of the plans on the new image data, lacking experimental verification, and ignoring the impact on the plan re‐optimization process. The aim of this study was to use gel and film dosimetry coupled with a three‐dimensional (3D) printed head phantom (based on the planning CT of the patient) for 3D range verification of intensity‐corrected cone beam computed tomography (CBCT) image data for adaptive proton therapy. Methods: Single field uniform dose pencil beam scanning proton plans were optimized for three different patients on the patients' planning CT (planCT) and the patients' intensity‐corrected CBCT (scCBCT) for the same target volume using the same optimization constraints. The CBCTs were corrected on projection level using the planCT as a prior. The dose optimized on planCT and recalculated on scCBCT was compared in terms of proton range differences (80% distal fall‐off, recalculation). Moreover, the dose distribution resulting from recalculation of the scCBCT‐optimized plan on the planCT and the original planCT dose distribution were compared (simulation). Finally, the two plans of each patient were irradiated on the corresponding patient‐specific 3D printed head phantom using gel dosimetry inserts for one patient and film dosimetry for all three patients. Range differences were extracted from the measured dose distributions. The measured and the simulated range differences were corrected for range differences originating from the initial plans and evaluated. Results: The simulation approach showed high agreement with the standard recalculation approach. The median values of the range differences of these two methods agreed within 0.1 mm and the interquartile ranges (IQRs) within 0.3 mm for all three patients. The range differences of the film measurement were accurately matching with the simulation approach in the film plane. The median values of these range differences deviated less than 0.1 mm and the IQRs less than 0.4 mm. For the full 3D evaluation of the gel range differences, the median value and IQR matched those of the simulation approach within 0.7 and 0.5 mm, respectively. scCBCT‐ and planCT‐based dose distributions were found to have a range agreement better than 3 mm (median and IQR) for all considered scenarios (recalculation, simulation, and measurement). Conclusions: The results of this initial study indicate that an online adaptive proton workflow based on scatter‐corrected CBCT image data for head irradiations is feasible. The novel presented measurement‐ and simulation‐based method was shown to be equivalent to the standard literature recalculation approach. Additionally, it has the capability to catch effects of image differences on the treatment plan optimization. This makes the measurement‐based approach particularly interesting for quality assurance of CBCT‐based online adaptive proton therapy. The observed uncertainties could be kept within those of the registration and positioning. The proposed validation could also be applied for other alternative in‐room images, e.g. for MR‐based pseudoCTs. [ABSTRACT FROM AUTHOR]