ABSTRACT
In this work we present a framework for robust deep learning-based VMAT forward dose calculations for the 1.5T MR-linac. A convolutional neural network was trained on the dose of individual multi-leaf-collimator VMAT segments and was used to predict the dose per segment for a set of MR-linac-deliverable VMAT test plans. The training set consisted of prostate, rectal, lung and esophageal tumour data. All patients were previously treated in our clinic with VMAT on a conventional linac. The clinical data were converted to an MR-linac environment prior to training. During training time, gantry and collimator angles were randomized for each training sample, while the multi-leaf-collimator shapes were rigidly shifted to ensure robust learning. A Monte Carlo dose engine was used for the generation of the ground truth data at 1% statistical uncertainty per control point. For a set of 17 MR-linac-deliverable VMAT test plans, generated on a research treatment planning system, our method predicted highly accurate dose distributions, reporting 99.7% ± 0.5% for the full plan prediction at the 3%/3 mm gamma criterion. Additional evaluation on previously unseen IMRT patients passed all clinical requirements resulting in 99.0% ± 0.6% for the 3%/3 mm analysis. The overall performance of our method makes it a promising plan validation solution for IMRT and VMAT workflows, robust to tumour anatomies and tissue density variations.
Subject(s)
Deep Learning , Radiotherapy, Intensity-Modulated , Humans , Male , Particle Accelerators , Radiotherapy Dosage , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Intensity-Modulated/methodsABSTRACT
Objective.GPU-oriented Monte Carlo dose (GPUMCD) is a fast dose calculation algorithm used for treatment planning on the Unity MR-linac. Treatments for the MR-linac must be calculated quickly and accurately, and must account for two important MR-linac aspects: off-axis positions and angular transmission through the cryostat, couch and MR-coils. Therefore, the aim of this research is to quantify the system-related errors for GPUMCD calculations over the range of clinically-relevant field configurations and gantry angles.Approach.Dose profiles (crossline, inline and PDD) were measured and calculated for varying field sizes, off-axis positions and depths. Eleven different (off-axis) positions were included. The angular transmission was investigated by measuring and calculating the transmission for multiple angles, taking the cryostat, couch and coils into account.Main results.Differences between absolute point doses were found to be within 1.7% for field sizes 2 × 2 cm2and larger. The relative dose profiles in the crossline, inline and PDD direction illustrated maximum mean dose differences of 0.9pp, 0.8pp and 0.7pp ofDmaxin the central region for field sizes 2 × 2 cm2and larger. The 1 × 1 cm2field size showed larger dosimetric errors for absolute point doses and relative dose profiles. The maximum mean DTA in the penumbra was 0.7 mm. The mean difference in angular transmission ranged from -0.33% ± 0.60% to 0.27% ± 0.91% using three treatment machines. Additionally, 77.1%-93.7% of the datapoints remained within 1% transmission difference. The largest transmission differences were present at the edges of the table.Significance.This research showed that the GPUMCD algorithm provides reliable dose calculations with a low uncertainty for field sizes 2 × 2 cm2and larger, focusing on off-axis fields and angular transmission.
Subject(s)
Particle Accelerators , Radiometry , Algorithms , Monte Carlo Method , Radiotherapy Planning, Computer-Assisted/methodsABSTRACT
Objective.Stereotactic arrhythmia radioablation (STAR) is a novel, non-invasive treatment for refractory ventricular tachycardia (VT). The VT isthmus is subject to both respiratory and cardiac motion. Rapid cardiac motion presents a unique challenge. In this study, we provide first experimental evidence for real-time cardiorespiratory motion-mitigated MRI-guided STAR on the 1.5 T Unity MR-linac (Elekta AB, Stockholm, Sweden) aimed at simultaneously compensating cardiac and respiratory motions.Approach.A real-time cardiorespiratory motion-mitigated radiotherapy workflow was developed on the Unity MR-linac in research mode. A 15-beam intensity-modulated radiation therapy treatment plan (1 × 25 Gy) was created in Monaco v.5.40.01 (Elekta AB) for the Quasar MRI4Dphantom (ModusQA, London, ON). A film dosimetry insert was moved by combining either artificial (cos4, 70 bpm, 10 mm peak-to-peak) or subject-derived (59 average bpm, 15.3 mm peak-to-peak) cardiac motion with respiratory (sin, 12 bpm, 20 mm peak-to-peak) motion. A balanced 2D cine MRI sequence (13 Hz, field-of-view = 400 × 207 mm2, resolution = 3 × 3 × 15 mm3) was developed to estimate cardiorespiratory motion. Cardiorespiratory motion was estimated by rigid registration and then deconvoluted into cardiac and respiratory components. For beam gating, the cardiac component was used, whereas the respiratory component was used for MLC-tracking. In-silico dose accumulation experiments were performed on three patient data sets to simulate the dosimetric effect of cardiac motion on VT targets.Main results.Experimentally, a duty cycle of 57% was achieved when simultaneously applying respiratory MLC-tracking and cardiac gating. Using film, excellent agreement was observed compared to a static reference delivery, resulting in a 1%/1 mm gamma pass rate of 99%. The end-to-end gating latency was 126 ms on the Unity MR-linac. Simulations showed that cardiac motion decreased the target's D98% dose between 0.1 and 1.3 Gy, with gating providing effective mitigation.Significance.Real-time MRI-guided cardiorespiratory motion management greatly reduces motion-induced dosimetric uncertainty and warrants further research and development for potential future use in STAR.
Subject(s)
Imaging, Three-Dimensional , Tachycardia, Ventricular , Arrhythmias, Cardiac , Heart/diagnostic imaging , Humans , Magnetic Resonance Imaging , MotionABSTRACT
PURPOSE: To evaluate seminal vesicle (SV) intrafraction motion using cinematic magnetic resonance imaging (cine-MR) during the delivery of online adaptive MR-Linac radiotherapy fractions, in preparation of MR-guided extremely hypofractionated radiotherapy for intermediate to high-risk prostate cancer patients. MATERIAL AND METHODS: Fifty prostate cancer patients were treated with 5 × 7.25 Gy on a 1.5 Tesla MR-Linac. 3D Cine-MR imaging was started simultaneously and acquired over the full beam-on period. Intrafraction motion in this cine-MR was determined for each SV separately with a previously validated soft-tissue contrast-based tracking algorithm. Motion statistics and coverage probability for the SVs and prostate were determined based on the obtained results. RESULTS: SV motion was automatically determined during the beam-on period (approx. 10 min) for 247 fractions. SV intrafraction motion shows larger spread than prostate intrafraction motion and increases over time. This difference is especially evident in the anterior and cranial translation directions. Significant difference in rotation about the left-right axis was found, with larger rotation for the SVs than the prostate. Intra-fraction coverage probability of 99% can be achieved when using 5 mm isometric expansion for the left and right SV and 3 mm for the prostate. CONCLUSION: This is the first study to investigate SV intrafraction motion during MR-guided RT sessions on an MR-Linac. We have shown that high quality 3D cine-MR imaging and SV tracking during RT is feasible with beam-on. The tracking method as described may be used as input for a fast replanning algorithm, which allows for intrafraction plan adaptation.
Subject(s)
Prostatic Neoplasms , Radiosurgery , Radiotherapy, Image-Guided , Radiotherapy, Intensity-Modulated , Humans , Magnetic Resonance Imaging , Male , Movement , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiotherapy Planning, Computer-Assisted , Rotation , Seminal Vesicles/diagnostic imagingABSTRACT
The integration of real-time magnetic resonance imaging (MRI) guidance and proton therapy would potentially improve the proton dose steering capability by reducing daily uncertainties due to anatomical variations. The use of a fixed beamline coupled with an axial patient couch rotation would greatly simplify the proton delivery with MRI guidance. Nonetheless, it is mandatory to assure that the plan quality is not deteriorated by the anatomical deformations due to patient rotation. In this work, an in-house tool allowing for intra-fractional per-beam adaptation of intensity-modulated proton plans (BeamAdapt) was implemented through features available in RayStation. A set of three MRIs was acquired for two healthy volunteers (V1,V2): (1) no rotation/static, (2) rotation to the right and (3) left.V1was rotated by 15°, to simulate a clinical pediatric abdominal case andV2by 45°, to simulate an extreme patient rotation case. For each volunteer, a total of four intensity-modulated pencil beam scanning plans were optimized on the static MRI using virtual abdominal targets and two-three posterior-oblique beams. Beam angles were defined according to the angulations on the rotated MRIs. With BeamAdapt, each original plan was initially converted into separate plans with one beam per plan. In an iterative order, individual beam doses were non-rigidly deformed to the rotated anatomies and re-optimized accounting for the consequent deformations and the beam doses delivered so far. For evaluation, the final accumulated dose distribution was propagated back to the static MRI. Planned and adapted dose distributions were compared by computing relative differences between dose-volume histogram metrics. Absolute target dose differences were on average below 1% and organs-at-risk mean dose differences were below 3%. With BeamAdapt, not only intra-fractional per-beam proton plan adaptation coupled with axial patient rotation is possible but also the need for a rotating gantry during MRI guidance might be mitigated.
Subject(s)
Proton Therapy , Radiotherapy, Intensity-Modulated , Child , Humans , Magnetic Resonance Imaging , Protons , Radiotherapy Dosage , Radiotherapy Planning, Computer-Assisted , WorkflowABSTRACT
Purpose.To assess the feasibility of prostate cancer radiotherapy for patients with a hip implant on an 1.5 T MRI-Linac (MRL) in terms of geometrical image accuracy, image quality, and plan quality.Methods.Pretreatment MRI images on a 1.5 T MRL and 3 T MRI consisting of a T2-weighted 3D delineation scan and main magnetic field homogeneity (B0) scan were performed in six patients with a unilateral hip implant. System specific geometrical errors due to gradient nonlinearity were determined for the MRL. Within the prostate and skin contour,B0inhomogeneity, gradient nonlinearity error and the total geometrical error (vector summation of the prior two) was determined. Image quality was determined by visually scoring the extent of implant-born image artifacts. A treatment planning study was performed on five patients to quantify the impact of the implant on plan quality, in which conventional MRL IMRT plans were created, as well as plans which avoid radiation through the left or right femur.Results.The total maximum geometrical error in the prostate was <1 mm and the skin contour <1.7 mm; in all cases the machine-specific gradient error was most dominant. TheB0error for the MRlinac MRI could partly be predicted based on the pre-treatment 3 T scan. Image quality for all patients was sufficient at 1.5 T MRL. Plan comparison showed that, even with avoidance of the hips, in all cases sufficient target coverage could be obtained with similar D1cc and D5cc to rectum and bladder, while V28Gy was slightly poorer in only the rectum for femur avoidance.Conclusion.We showed that geometrical accuracy, image quality and plan quality for six prostate patients with a hip implant or hip fixation treated on a 1.5 T MRL did not show relevant deterioration for the used image settings, which allowed safe treatment.
Subject(s)
Prostatic Neoplasms , Radiotherapy, Intensity-Modulated , Humans , Magnetic Resonance Imaging , Male , Particle Accelerators , Prostate , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiotherapy Dosage , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Intensity-Modulated/methodsABSTRACT
4D-MRI is becoming increasingly important for daily guidance of thoracic and abdominal radiotherapy. This study exploits the simultaneous multi-slice (SMS) technique to accelerate the acquisition of a balanced turbo field echo (bTFE) and a turbo spin echo (TSE) coronal 4D-MRI sequence performed on 1.5 T MRI scanners. SMS single-shot bTFE and TSE sequences were developed to acquire a stack of 52 coronal 2D images over 30 dynamics. Simultaneously excited slices were separated by half the field of view. Slices intersecting with the liver-lung interface were used as navigator slices. For each navigator slice location, an end-exhale dynamic was automatically identified, and used to derive the self-sorting signal by rigidly registering the remaining dynamics. Navigator slices were sorted into 10 amplitude bins, and the temporal relationship of simultaneously excited slices was used to generate sorted 4D-MRIs for 12 healthy volunteers. The self-sorting signal was validated using anin vivopeak-to-peak motion analysis. The smoothness of the liver-lung interface was quantified by comparing to sagittal cine images acquired directly after the SMS-4D-MRI sequence. To ensure compatibility with the MR-linac radiotherapy workflow, the 4D-MRIs were transformed into 3D mid-position (MidP) images using deformable image registration. Consistency of the deformable vector fields was quantified in terms of the distance discordance metric (DDM) in the body. The SMS-4D-TSE sequence was additionally acquired for 3 lung cancer patients to investigate tumor visibility. SMS-4D-MRI acquisition and processing took approximately 7 min. 4D-MRI reconstruction was possible for 26 out of 27 acquired datasets. Missing data in the sorted 4D-MRIs varied from 4%-26% for the volunteers and varied from 8%-24% for the patients. Peak-to-peak (SD) amplitudes analysis agreed within 1.8 (1.1) mm and 0.9 (0.4) mm between the sorted 4D-MRIs and the self-sorting signals of the volunteers and patients, respectively. Liver-lung interface smoothness was found to be in the range of 0.6-3.1 mm for volunteers. The percentage of DDM values smaller than 2 mm was in the range of 85%-89% and 86%-92% for the volunteers and patients, respectively. Lung tumors were clearly visibility in the SMS-4D-TSE images and MidP images. Two fast SMS-accelerated 4D-MRI sequences were developed resulting in T2/T1or T2weighted contrast. The SMS-4D-MRIs and derived 3D MidP-MRIs yielded anatomically plausible images and good tumor visibility. SMS-4D-MRI is therefore a strong candidate to be used for treatment simulation and daily guidance of thoracic and abdominal MR-guided radiotherapy.
Subject(s)
Magnetic Resonance Imaging , Humans , Imaging, Three-Dimensional , Liver Neoplasms , Motion , Particle AcceleratorsABSTRACT
We present a robust deep learning-based framework for dose calculations of abdominal tumours in a 1.5 T MRI radiotherapy system. For a set of patient plans, a convolutional neural network is trained on the dose of individual multi-leaf-collimator segments following the DeepDose framework. It can then be used to predict the dose distribution per segment for a set of patient anatomies. The network was trained using data from three anatomical sites of the abdomen: prostate, rectal and oligometastatic tumours. A total of 216 patient fractions were used, previously treated in our clinic with fixed-beam IMRT using the Elekta MR-linac. For the purpose of training, 176 fractions were used with random gantry angles assigned to each segment, while 20 fractions were used for the validation of the network. The ground truth data were calculated with a Monte Carlo dose engine at 1% statistical uncertainty per segment. For a total of 20 independent abdominal test fractions with the clinical angles, the network was able to accurately predict the dose distributions, achieving 99.4% ± 0.6% for the whole plan prediction at the 3%/3 mm gamma test. The average dose difference and standard deviation per segment was 0.3% ± 0.7%. Additional dose prediction on one cervical and one pancreatic case yielded high dose agreement of 99.9% and 99.8% respectively for the 3%/3 mm criterion. Overall, we show that our deep learning-based dose engine calculates highly accurate dose distributions for a variety of abdominal tumour sites treated on the MR-linac, in terms of performance and generality.
Subject(s)
Abdominal Neoplasms/diagnostic imaging , Abdominal Neoplasms/radiotherapy , Deep Learning , Magnetic Resonance Imaging/methods , Neural Networks, Computer , Particle Accelerators , Radiotherapy Dosage , Radiotherapy Planning, Computer-Assisted/instrumentation , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Intensity-Modulated/methods , Humans , Male , Monte Carlo Method , Neoplasm Metastasis , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Rectal Neoplasms/diagnostic imaging , Rectal Neoplasms/drug therapy , Reproducibility of ResultsABSTRACT
Accurate spatial dose delivery in radiotherapy is frequently complicated due to changes in the patient's internal anatomy during and in-between therapy segments. The recent introduction of hybrid MRI radiotherapy systems allows unequaled soft-tissue visualization during radiation delivery and can be used for dose reconstruction to quantify the impact of motion. To this end, knowledge of anatomical deformations obtained from continuous monitoring during treatment has to be combined with information on the spatio-temporal dose delivery to perform motion-compensated dose accumulation (MCDA). Here, the influence of the choice of deformable image registration algorithm, dose warping strategy, and magnetic resonance image resolution and signal-to-noise-ratio on the resulting MCDA is investigated. For a quantitative investigation, four 4D MRI-datasets representing typical patient observed motion patterns are generated using finite element modeling and serve as a gold standard. Energy delivery is simulated intra-fractionally in the deformed image space and, subsequently, MCDA-processed. Finally, the results are substantiated by comparing MCDA strategies on clinically acquired patient data. It is shown that MCDA is needed for correct quantitative dose reconstruction. For prostate treatments, using the energy per mass transfer dose warping strategy has the largest influence on decreasing dose estimation errors.
Subject(s)
Magnetic Resonance Imaging/methods , Motion , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Image-Guided/methods , Signal-To-Noise Ratio , Algorithms , Humans , Image Processing, Computer-Assisted , Imaging, Three-Dimensional , Male , Prostate/diagnostic imaging , Rectum/physiopathology , Reproducibility of ResultsABSTRACT
In this work we present the first delivery of intensity modulated arc therapy on the Elekta Unity 1.5 T MR-linac. The machine's current intensity modulated radiation therapy based control system was modified suitably to enable dynamic delivery of radiation, for the purpose of exploring MRI-guided radiation therapy adaptation modes in a research setting. The proof-of-concept feasibility was demonstrated by planning and delivering two types of plans, each investigating the performance of different parts of a dynamic treatment. A series of fixed-speed arc plans was used to show the high-speed capabilities of the gantry during radiation, while several fully modulated prostate plans-optimised following the volumetric modulated arc therapy approach-were delivered in order to establish the performance of its multi-leaf collimator and diaphragms. These plans were delivered to Delta4 Phantom+ MR and film phantoms passing the clinical quality assurance criteria used in our clinic. In addition, we also performed some initial MR imaging experiments during dynamic therapy, demonstrating that the impact of radiation and moving gantry/collimator components on the image quality is negligible. These results show that arc therapy is feasible on the Elekta Unity system. The machine's high performance components enable dynamic delivery during fast gantry rotation and can be controlled in a stable fashion to deliver fully modulated plans.
Subject(s)
Magnetic Resonance Imaging/instrumentation , Particle Accelerators , Radiotherapy, Intensity-Modulated/instrumentation , Humans , Male , Phantoms, Imaging , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiotherapy Dosage , Radiotherapy Planning, Computer-Assisted , RotationABSTRACT
OBJECTIVE: Dose prediction using deep learning networks prior to radiotherapy might lead tomore efficient modality selections. The study goal was to predict proton and photon dose distributions based on the patient-specific anatomy and to assess their clinical usage for paediatric abdominal tumours. MATERIAL AND METHODS: Data from 80 patients with neuroblastoma or Wilms' tumour was included. Pencil beam scanning (PBS) (5 mm/ 3%) and volumetric-modulated arc therapy (VMAT) plans (5 mm) were robustly optimized on the internal target volume (ITV). Separate 3-dimensional patch-based U-net networks were trained to predict PBS and VMAT dose distributions. Doses, planning-computed tomography images and relevant optimization masks (ITV, vertebra and organs-at-risk) of 60 patients were used for training with a 5-fold cross validation. The networks' performance was evaluated by computing the relative error between planned and predicted dose-volume histogram (DVH) parameters for 20 inference patients. In addition, the organs-at-risk mean dose difference between modalities was calculated using planned and predicted dose distributions (ΔDmean = DVMAT-DPBS). Two radiation oncologists performed a blind PBS/VMAT modality selection based on either planned or predicted ΔDmean. RESULTS: Average DVH differences between planned and predicted dose distributions were ≤ |6%| for both modalities. The networks classified the organs-at-risk Dmean difference as a gain (ΔDmean > 0) with 98% precision. An identical modality selection based on planned compared to predicted ΔDmean was made for 18/20 patients. CONCLUSION: Deep learning networks for accurate prediction of proton and photon dose distributions for abdominal paediatric tumours were established. These networks allowing fast dose visualisation might aid in identifying the optimal radiotherapy technique when experience and/or resources are unavailable.
Subject(s)
Abdominal Neoplasms , Deep Learning , Proton Therapy , Radiotherapy, Intensity-Modulated , Abdominal Neoplasms/radiotherapy , Child , Humans , Organs at Risk , Protons , Radiotherapy Dosage , Radiotherapy Planning, Computer-AssistedABSTRACT
Image-guided radiotherapy (IGRT) allows observation of the location and shape of the tumor and organs-at-risk (OAR) over the course of a radiation cancer treatment. Such information may in turn be used for reducing geometric uncertainties during therapeutic planning, dose delivery and response assessment. However, given the multiple imaging modalities and/or contrasts potentially included within the imaging protocol over the course of the treatment, the current manual approach to determining tissue displacement may become time-consuming and error prone. In this context, variational multi-modal deformable image registration (DIR) algorithms allow automatic estimation of tumor and OAR deformations across the acquired images. In addition, they require short computational times and a low number of input parameters, which is particularly beneficial for online adaptive applications, which require on-the-fly adaptions with the patient on the treatment table. However, the majority of such DIR algorithms assume that all structures across the entire field-of-view (FOV) undergo a similar deformation pattern. Given that various anatomical structures may behave considerably different, this may lead to the estimation of anatomically implausible deformations at some locations, thus limiting their validity. Therefore, in this paper we propose an anatomically-adaptive variational multi-modal DIR algorithm, which employs a regionalized registration model in accordance with the local underlying anatomy. The algorithm was compared against two existing methods which employ global assumptions on the estimated deformations patterns. Compared to the existing approaches, the proposed method has demonstrated an improved anatomical plausibility of the estimated deformations over the entire FOV as well as displaying overall higher accuracy. Moreover, despite the more complex registration model, the proposed approach is very fast and thus suitable for online scenarios. Therefore, future adaptive IGRT workflows may benefit from an anatomically-adaptive registration model for precise contour propagation and dose accumulation, in areas showcasing considerable variations in anatomical properties.
Subject(s)
Image Processing, Computer-Assisted/methods , Multimodal Imaging , Radiotherapy, Image-Guided , Algorithms , Humans , Radiotherapy Planning, Computer-AssistedABSTRACT
PURPOSE: To evaluate prostate intrafraction motion using MRI during the full course of online adaptive MR-Linac radiotherapy (RT) fractions, in preparation of MR-guided extremely hypofractionated RT. MATERIAL AND METHODS: Five low and intermediate risk prostate cancer patients were treated with 20 × 3.1 Gy fractions on a 1.5T MR-Linac. Each fraction, initial MRI (Pre) scans were obtained at the start of every treatment session. Pre-treatment planning MRI contours were propagated and adapted to this Pre scan after which plan re-optimization was started in the treatment planning system followed by dose delivery. 3D Cine-MR imaging was started simultaneously with beam-on and acquired over the full beam-on period. Prostate intrafraction motion in this cine-MR was determined with a previously validated soft-tissue contrast based tracking algorithm. In addition, absolute accuracy of the method was determined using a 4D phantom. RESULTS: Prostate motion was completely automatically determined over the full on-couch period (approx. 45 min) with no identified mis-registrations. The translation 95% confidence intervals are within clinically applied margins of 5 mm, and plan adaption for intrafraction motion was required in only 4 out of 100 fractions. CONCLUSION: This is the first study to investigate prostate intrafraction motions during entire MR-guided RT sessions on an MR-Linac. We have shown that high quality 3D cine-MR imaging and prostate tracking during RT is feasible with beam-on. The clinically applied margins of 5 mm have proven to be sufficient for these treatments and may potentially be further reduced using intrafraction plan adaptation guided by cine-MR imaging.
Subject(s)
Prostatic Neoplasms , Radiotherapy Planning, Computer-Assisted , Humans , Magnetic Resonance Imaging , Male , Movement , Particle Accelerators , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapyABSTRACT
We present DeepDose, a deep learning framework for fast dose calculations in radiation therapy. Given a patient anatomy and linear-accelerator IMRT multi-leaf-collimator shape or segment, a novel set of physics-based inputs is calculated that encode the linac machine parameters into the underlying anatomy. These inputs are then used to train a deep convolutional network to derive the dose distribution of individual MLC shapes on a given patient anatomy. In this work we demonstrate the proof-of-concept application of DeepDose on 101 prostate patients treated in our clinic with fixed-beam IMRT. The ground-truth data used for training, validation and testing of the prediction were calculated with a state-of-the-art Monte Carlo dose engine at 1% statistical uncertainty per segment. A deep convolution network was trained using the data of 80 patients at the clinically used 3 mm3 grid spacing while 10 patients were used for validation. For another 11 independent test patients, the network was able to accurately estimate the segment doses from the clinical plans of each patient passing the clinical QA when compared with the Monte Carlo calculations, yielding on average 99.9%±0.3% for the forward calculated patient plans at 3%/3 mm gamma tests. Dose prediction using the trained network was very fast at approximately 0.9 seconds for the input generation and 0.6 seconds for single GPU inference per segment and 1 minute per patient in total. The overall performance of this dose calculation framework in terms of both accuracy and inference speed, makes it compelling for online adaptive workflows where fast segment dose calculations are needed.
Subject(s)
Deep Learning , Radiation Dosage , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Intensity-Modulated , Algorithms , Humans , Male , Monte Carlo Method , Particle Accelerators , Prostatic Neoplasms/radiotherapy , Radiotherapy DosageABSTRACT
PURPOSE: To quickly and automatically propagate organ contours from pretreatment to fraction images in magnetic resonance (MR)-guided prostate external-beam radiotherapy. METHODS: Five prostate cancer patients underwent 20 fractions of image-guided external-beam radiotherapy on a 1.5 T MR-Linac system. For each patient, a pretreatment T2-weighted three-dimensional (3D) MR imaging (MRI) scan was used to delineate the clinical target volume (CTV) contours. The same scan was repeated during each fraction, with the CTV contour being manually adapted if necessary. A convolutional neural network (CNN) was trained for combined image registration and contour propagation. The network estimated the propagated contour and a deformation field between the two input images. The training set consisted of a synthetically generated ground truth of randomly deformed images and prostate segmentations. We performed a leave-one-out cross-validation on the five patients and propagated the prostate segmentations from the pretreatment to the fraction scans. Three variants of the CNN, aimed at investigating supervision based on optimizing segmentation overlap, optimizing the registration, and a combination of the two were compared to results of the open-source deformable registration software package Elastix. RESULTS: The neural networks trained on segmentation overlap or the combined objective achieved significantly better Hausdorff distances between predicted and ground truth contours than Elastix, at the much faster registration speed of 0.5 s. The CNN variant trained to optimize both the prostate overlap and deformation field, and the variant trained to only maximize the prostate overlap, produced the best propagation results. CONCLUSIONS: A CNN trained on maximizing prostate overlap and minimizing registration errors provides a fast and accurate method for deformable contour propagation for prostate MR-guided radiotherapy.
Subject(s)
Image Processing, Computer-Assisted/methods , Magnetic Resonance Imaging , Neural Networks, Computer , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiotherapy, Image-Guided , Dose Fractionation, Radiation , Humans , Male , Time FactorsABSTRACT
To investigate the dosimetric impact of intrafraction translation and rotation motion of the prostate, as extracted from daily acquired post-treatment 3D cine-MR based on soft-tissue contrast, in extremely hypofractionated (SBRT) prostate patients. Accurate dose reconstruction is performed by using a prostate intrafraction motion trace which is obtained with a soft-tissue based rigid registration method on 3D cine-MR dynamics with a temporal resolution of 11 s. The recorded motion of each time-point was applied to the planning CT, resulting in the respective dynamic volume used for dose calculation. For each treatment fraction, the treatment delivery record was generated by proportionally splitting the plan into 11 s intervals based on the delivered monitor units. For each fraction the doses of all partial plan/dynamic volume combinations were calculated and were summed to lead to the motion-affected fraction dose. Finally, for each patient the five fraction doses were summed, yielding the total treatment dose. Both daily and total doses were compared to the original reference dose of the respective patient to assess the impact of the intrafraction motion. Depending on the underlying motion of the prostate, different types of motion-affected dose distributions were observed. The planning target volumes (PTVs) ensured CTV_30 (seminal vesicles) D99% coverage for all patients, CTV_35 (prostate corpus) coverage for 97% of the patients and GTV_50 (local boost) for 83% of the patients when compared against the strict planning target D99% value. The dosimetric impact due to prostate intrafraction motion in extremely hypofractionated treatments was determined. The presented study is an essential step towards establishing the actual delivered dose to the patient during radiotherapy fractions.
Subject(s)
Dose Fractionation, Radiation , Imaging, Three-Dimensional , Movement , Prostatic Neoplasms/diagnostic imaging , Prostatic Neoplasms/radiotherapy , Radiosurgery/methods , Algorithms , Humans , Male , Radiometry , Radiotherapy Planning, Computer-Assisted , RotationABSTRACT
To develop a method to automatically determine intrafraction motion of the prostate based on soft tissue contrast on 3D cine-magnetic resonance (MR) images with high spatial and temporal resolution. Twenty-nine patients who underwent prostate stereotactic body radiotherapy (SBRT), with four implanted cylindrical gold fiducial markers (FMs), had cine-MR imaging sessions after each of five weekly fractions. Each cine-MR session consisted of 55 sequentially obtained 3D data sets ('dynamics') and was acquired over an 11 s period, covering a total of 10 min. The prostate was delineated on the first dynamic of every dataset and this delineation was used as the starting position for the soft tissue tracking (SST). Each subsequent dynamic was rigidly aligned to the first dynamic, based on the contrast of the prostate. The obtained translation and rotation describes the intrafraction motion of the prostate. The algorithm was applied to 6270 dynamics over 114 scans of 29 patients and the results were validated by comparing to previously obtained fiducial marker tracking data of the same dataset. Our proposed tracking method was also retro-perspectively applied to cine-MR images acquired during MR-guided radiotherapy of our first prostate patient treated on the MR-Linac. The difference in the 3D translation results between the soft tissue and marker tracking was below 1 mm for 98.2% of the time. The mean translation at 10 min were X: 0.0 [Formula: see text] 0.8 mm, Y: 1.0 [Formula: see text] 1.8 mm and Z: [Formula: see text] mm. The mean rotation results at 10 min were X: [Formula: see text], Y: 0.1 [Formula: see text] 0.6° and Z: 0.0 [Formula: see text] 0.7°. A fast, robust and accurate SST algorithm was developed which obviates the need for FMs during MR-guided prostate radiotherapy. To our knowledge, this is the first data using full 3D cine-MR images for real-time soft tissue prostate tracking, which is validated against previously obtained marker tracking data.
Subject(s)
Imaging, Three-Dimensional/methods , Magnetic Resonance Imaging, Cine/methods , Prostatic Neoplasms/radiotherapy , Radiosurgery/methods , Radiotherapy Planning, Computer-Assisted/methods , Radiotherapy, Image-Guided/methods , Algorithms , Fiducial Markers , Humans , Imaging, Three-Dimensional/standards , Magnetic Resonance Imaging, Cine/standards , Male , Movement , Prostate/diagnostic imaging , Prostatic Neoplasms/diagnostic imaging , Radiosurgery/standards , Radiotherapy Planning, Computer-Assisted/standards , Radiotherapy, Image-Guided/standards , RotationABSTRACT
Radiation therapy (RT) is an essential component of effective cancer care and is used across nearly all cancer types. The delivery of RT is becoming more precise through rapid advances in both computing and imaging. The direct integration of magnetic resonance imaging (MRI) with linear accelerators represents an exciting development with the potential to dramatically impact cancer research and treatment. These impacts extend beyond improved imaging and dose deposition. Real-time MRI-guided RT is actively transforming the work flows and capabilities of virtually every aspect of RT. It has the opportunity to change entirely the delivery methods and response assessments of numerous malignancies. This review intends to approach the topic of MRI-based RT guidance from a vendor neutral and international perspective. It also aims to provide an introduction to this topic targeted towards oncologists without a speciality focus in RT. Speciality implications, areas for physician education and research opportunities are identified as they are associated with MRI-guided RT. The uniquely disruptive implications of MRI-guided RT are discussed and placed in context. We further aim to describe and outline important future changes to the speciality of radiation oncology that will occur with MRI-guided RT. The impacts on RT caused by MRI guidance include target identification, RT planning, quality assurance, treatment delivery, training, clinical workflow, tumour response assessment and treatment scheduling. In addition, entirely novel research areas that may be enabled by MRI guidance are identified for future investigation.
Subject(s)
Magnetic Resonance Imaging/methods , Radiation Oncology , Radiotherapy, Image-Guided/methods , Humans , Quality Assurance, Health Care/methods , Radiation Oncology/education , Radiation Oncology/methods , Radiation Oncology/standardsABSTRACT
With the recent advent of hybrid MRI-guided radiotherapy systems, continuous intra-fraction MR imaging for motion monitoring has become feasible. The ability to perform real-time custom image reconstructions is however often lacking. In this work we present a low-latency streaming solution, ReconSocket, which provides a real-time stream of k-space data from the magnetic resonance imaging (MRI) to custom reconstruction servers. We determined the performance of the data streaming by measuring the streaming latency (i.e. non-zero time delay due to data transfer and processing) and jitter (i.e. deviations from periodicity) using an ultra-fast 1D MRI acquisition of a moving phantom. Simultaneously, its position was recorded with near-zero time delay. The feasibility of low-latency custom reconstructions was tested by measuring the imaging latency (i.e. time delay between physical change and appearance of that change on the image) for several non-Cartesian 2D and 3D acquisitions using an in-house implemented reconstruction server. The measured streaming latency of the ReconSocket interface was [Formula: see text] ms. 98% of the incoming data packets arrived within a jitter range of 367 [Formula: see text]s. This shows that the ReconSocket interface can provide reliable real-time access to MRI data, acquired during the course of a MRI-guided radiotherapy fraction. The total imaging latency was measured to be 221 ms (2D) and 3889 ms (3D) for exemplary acquisitions, using the custom image reconstruction server. These imaging latencies are approximately equal to half of the temporal footprint (T acq /2) of the respective 2D and 3D golden-angle radial sequences. For radial sequences, it was previously showed that T acq /2 is the expected contribution of only the data acquisition to the total imaging latency. Indeed, the contribution of the non-Cartesian reconstruction to the total imaging latency was minor (<10%): 21 ms for 2D, 300 ms for 3D, indicating that the acquisition, i.e. the physical encoding of the image itself is the major contributor to the total imaging latency.
Subject(s)
Image Interpretation, Computer-Assisted/methods , Magnetic Resonance Imaging/methods , Phantoms, Imaging , Radiotherapy, Image-Guided/methods , Software , Humans , Imaging, Three-Dimensional/methods , MovementABSTRACT
Recently, multileaf collimator (MLC)-tracking has been technically and clinically demonstrated showing promising improvements of radiotherapy of mobile sites. Furthermore, magnetic resonance imaging (MRI)-guided treatments have shown to provide superior targetting performance due to on-line soft-tissue imaging. Hitherto, the combination of MLC-tracking and MRI has not been investigated using clinically released hardware. In this note we aim to describe the technical feasibilty of such a combination on a clinically operating MRI-linac. The MLC-tracking system is characterized by quantifying the latencies and geometric errors produced by the system. In order to reach optimization recommendations, the tracking system was first characterized using a quasi-ideal position sensor, isolating the performance of the MLC only. Subsequently, the analysis was repeated using real-time MRI as the positioning source for the MLC. For the isolated MLC, we found latencies of 20.67 ms and minimal overshooting behaviour. The latencies for MRI-guidance were 347.45 ms at 4 Hz imaging and 204 ms at 8 Hz. We showed that MLC-tracking on the Elekta Unity using integrated MRI is technically supported and feasible. The isolated analysis of the MLC demonstrated the negligible contribution of the MLC in MRI-guided tracking. The latency and geometric errors caused by the sampling properties of MRI exceed the MLC-related errors by several factors. Most gain for real-time MRI-based adaptive radiotherapy can therefore be realized by optimizing and accelerating the MRI acquisition process.