Abstract Objective . Deep-learning (DL)-based dose engines have been developed to alleviate the intrinsic compromise between the calculation accuracy and efficiency of the traditional dose calculation algorithms. However, current DL-based engines typically possess high computational complexity and require powerful computing devices. Therefore, to mitigate their computational burdens and broaden their applicability to a clinical setting where resource-limited devices are available, we proposed a compact dose engine via knowledge distillation (KD) framework that offers an ultra-fast calculation speed with high accuracy for prostate Volumetric Modulated Arc Therapy (VMAT). Approach . The KD framework contains two sub-models: a large pre-trained teacher and a small to-be-trained student. The student receives knowledge transferred from the teacher for better generalization. The trained student serves as the final engine for dose calculation. The model input is patient computed tomography and VMAT dose in water, and the output is DL-calculated patient dose. The ground-truth \dose was computed by the Monte Carlo module of the Monaco treatment planning system. Twenty and ten prostate cases were included for model training and assessment, respectively. The model’s performance (teacher/student/student-only) was evaluated by Gamma analysis and inference efficiency. Main results . The dosimetric comparisons (input/DL-calculated/ground-truth doses) suggest that the proposed engine can effectively convert low-accuracy doses in water to high-accuracy patient doses. The Gamma passing rate (2%/2 mm, 10% threshold) between the DL-calculated and ground-truth doses was 98.64 ± 0.62% (teacher), 98.13 ± 0.76% (student), and 96.95 ± 1.02% (student-only). The inference time was 16 milliseconds (teacher) and 11 milliseconds (student/student-only) using a graphics processing unit device, while it was 936 milliseconds (teacher) and 374 milliseconds (student/student-only) using a central processing unit device. Significance . With the KD framework, a compact dose engine can achieve comparable accuracy to that of a larger one. Its compact size reduces the computational burdens and computing device requirements, and thus such an engine can be more clinically applicable.
more »
« less
Investigation of autosegmentation techniques on T2‐weighted MRI for off‐line dose reconstruction in MR‐linac workflow for head and neck cancers
In order to accurately accumulate delivered dose for head and neck cancer patients treated with the Adapt to Position workflow on the 1.5T magnetic resonance imaging (MRI)‐linear accelerator (MR‐linac), the low‐resolution T2‐weighted MRIs used for daily setup must be segmented to enable reconstruction of the delivered dose at each fraction.
In this pilot study, we evaluate various autosegmentation methods for head and neck organs at risk (OARs) on on‐board setup MRIs from the MR‐linac for off‐line reconstruction of delivered dose.
Seven OARs (parotid glands, submandibular glands, mandible, spinal cord, and brainstem) were contoured on 43 images by seven observers each. Ground truth contours were generated using a simultaneous truth and performance level estimation (STAPLE) algorithm. Twenty total autosegmentation methods were evaluated in ADMIRE: 1–9) atlas‐based autosegmentation using a population atlas library (PAL) of 5/10/15 patients with STAPLE, patch fusion (PF), random forest (RF) for label fusion; 10–19) autosegmentation using images from a patient's 1–4 prior fractions (individualized patient prior [IPP]) using STAPLE/PF/RF; 20) deep learning (DL) (3D ResUNet trained on 43 ground truth structure sets plus 45 contoured by one observer). Execution time was measured for each method. Autosegmented structures were compared to ground truth structures using the Dice similarity coefficient, mean surface distance (MSD), Hausdorff distance (HD), and Jaccard index (JI). For each metric and OAR, performance was compared to the inter‐observer variability using Dunn's test with control. Methods were compared pairwise using the Steel‐Dwass test for each metric pooled across all OARs. Further dosimetric analysis was performed on three high‐performing autosegmentation methods (DL, IPP with RF and 4 fractions [IPP_RF_4], IPP with 1 fraction [IPP_1]), and one low‐performing (PAL with STAPLE and 5 atlases [PAL_ST_5]). For five patients, delivered doses from clinical plans were recalculated on setup images with ground truth and autosegmented structure sets. Differences in maximum and mean dose to each structure between the ground truth and autosegmented structures were calculated and correlated with geometric metrics.
DL and IPP methods performed best overall, all significantly outperforming inter‐observer variability and with no significant difference between methods in pairwise comparison. PAL methods performed worst overall; most were not significantly different from the inter‐observer variability or from each other. DL was the fastest method (33 s per case) and PAL methods the slowest (3.7–13.8 min per case). Execution time increased with a number of prior fractions/atlases for IPP and PAL. For DL, IPP_1, and IPP_RF_4, the majority (95%) of dose differences were within ± 250 cGy from ground truth, but outlier differences up to 785 cGy occurred. Dose differences were much higher for PAL_ST_5, with outlier differences up to 1920 cGy. Dose differences showed weak but significant correlations with all geometric metrics (
The autosegmentation methods offering the best combination of performance and execution time are DL and IPP_1. Dose reconstruction on on‐board T2‐weighted MRIs is feasible with autosegmented structures with minimal dosimetric variation from ground truth, but contours should be visually inspected prior to dose reconstruction in an end‐to‐end dose accumulation workflow.
- NSF-PAR ID:
- 10441516
- Publisher / Repository:
- Wiley Blackwell (John Wiley & Sons)
- Date Published:
- Journal Name:
- Medical Physics
- ISSN:
- 0094-2405
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
Abstract Background Stereotactic radiosurgery (SRS) is an established treatment for patients with brain metastases (BMs). However, damage to the healthy brain may limit the tumor dose for patients with multiple lesions.
Purpose In this study, we investigate the potential of spatiotemporal fractionation schemes to reduce the biological dose received by the healthy brain in SRS of multiple BMs, and also demonstrate a novel concept of spatiotemporal fractionation for polymetastatic cancer patients that faces less hurdles for clinical implementation.
Methods Spatiotemporal fractionation (STF) schemes aim at partial hypofractionation in the metastases along with more uniform fractionation in the healthy brain. This is achieved by delivering distinct dose distributions in different fractions, which are designed based on their cumulative biologically effective dose () such that each fraction contributes with high doses to complementary parts of the target volume, while similar dose baths are delivered to the normal tissue. For patients with multiple brain metastases, a novel constrained approach to spatiotemporal fractionation (cSTF) is proposed, which is more robust against setup and biological uncertainties. The approach aims at irradiating entire metastases with possibly different doses, but spatially similar dose distributions in every fraction, where the optimal dose contribution of every fraction to each metastasis is determined using a new planning objective to be added to the BED‐based treatment plan optimization problem. The benefits of spatiotemporal fractionation schemes are evaluated for three patients, each with >25 BMs.
Results For the same tumor BED10and the same brain volume exposed to high doses in all plans, the mean brain BED2can be reduced compared to uniformly fractionated plans by 9%–12% with the cSTF plans and by 13%–19% with the STF plans. In contrast to the STF plans, the cSTF plans avoid partial irradiation of the individual metastases and are less sensitive to misalignments of the fractional dose distributions when setup errors occur.
Conclusion Spatiotemporal fractionation schemes represent an approach to lower the biological dose to the healthy brain in SRS‐based treatments of multiple BMs. Although cSTF cannot achieve the full BED reduction of STF, it improves on uniform fractionation and is more robust against both setup errors and biological uncertainties related to partial tumor irradiation.
-
more » « less
Abstract Purpose The main purpose of this work was to generate and validate the dosimetric accuracy of proton beams of dimensions that are appropriate for in vivo small animal and in vitro ultrahigh dose rate (FLASH) radiotherapy experiments using a synchrotron‐based treatment delivery system. This study was performed to enable future investigations of the relevance of a spread‐out Bragg peak (SOBP) under FLASH conditions.
Methods The spill characteristics of the small field fixed horizontal beam line were modified to deliver accelerated protons in times as short as 2 ms and to control the dose delivered. A Gaussian‐like transverse beam profile was transformed into a square uniform one at FLASH dose rates, while avoiding low‐dose regions, a crucial requirement to protect normal tissue during FLASH irradiation. Novel beam‐shaping devices were designed using Monte Carlo techniques to produce up to about 6 cm3of uniform dose in SOBPs while maximizing the dose rate. These included a scattering foil, a conical flattening filter to maximize the flux of protons into the region of interest, energy filters, range compensators, and collimators. The shapes, sizes, and positions of the components were varied to provide the required field sizes and SOBPs.
Results The designed and fabricated devices were used to produce 10‐, 15‐, and 20‐mm diameter, circular field sizes and 10‐, 15‐, and 20‐mm SOBP modulation widths at uniform physical dose rates of up to 375 Gy/s at the center of the SOBP and a minimum dose rate of about 255 Gy/s at the entrance, respectively, in cylindrical volumes. The flatness of lateral dose profiles at the center could be adjusted to within ±1.5% at the center of the SOBP. Assessment of systematic uncertainties, such as impact of misalignments and positioning uncertainties, was performed using simulations, and the results were used to provide appropriate adjustments to ensure high‐accuracy FLASH beam delivery for both in vitro and in vivo preclinical experiments.
Conclusions It is feasible to use synchrotron‐generated proton beams of sufficient dimensions for FLASH radiobiology experiments. We expect to use the system we developed to acquire in vitro and in vivo small animal FLASH radiobiology data as a function of dose, dose rate, oxygen content, and linear energy transfer to help us understand the underlying mechanisms of the FLASH phenomenon.
-
more » « less
Abstract Purpose Most commercially available treatment planning systems (TPSs) approximate the continuous delivery of volumetric modulated arc therapy (VMAT) plans with a series of discretized static beams for treatment planning, which can make VMAT dose computation extremely inefficient. In this study, we developed a polar‐coordinate‐based pencil beam (PB) algorithm for efficient VMAT dose computation with high‐resolution gantry angle sampling that can improve the computational efficiency and reduce the dose discrepancy due to the angular under‐sampling effect.
Methods and Materials 6 MV pencil beams were simulated on a uniform cylindrical phantom under an EGSnrc Monte Carlo (MC) environment. The MC‐generated PB kernels were collected in the polar coordinate system for each bixel on a fluence map and subsequently fitted via a series of Gaussians. The fluence was calculated using a detectors’ eye view with off‐axis and MLC transmission factors corrected. Doses of VMAT arc on the phantom were computed by summing the convolution results between the corresponding PB kernels and fluence for each bixel in the polar coordinate system. The convolution was performed using fast Fourier transform to expedite the computing speed. The calculated doses were converted to the Cartesian coordinate system and compared with the reference dose computed by a collapsed cone convolution (CCC) algorithm of the TPS. A heterogeneous phantom was created to study the heterogeneity corrections using the proposed algorithm. Ten VMAT arcs were included to evaluate the algorithm performance. Gamma analysis and computation complexity theory were used to measure the dosimetric accuracy and computational efficiency, respectively.
Results The dosimetric comparisons on the homogeneous phantom between the proposed PB algorithm and the CCC algorithm for 10 VMAT arcs demonstrate that the proposed algorithm can achieve a dosimetric accuracy comparable to that of the CCC algorithm with average gamma passing rates of 96% (2%/2mm) and 98% (3%/3mm). In addition, the proposed algorithm can provide better computational efficiency for VMAT dose computation using a PC equipped with a 4‐core processor, compared to the CCC algorithm utilizing a dual 10‐core server. Moreover, the computation complexity theory reveals that the proposed algorithm has a great advantage with regard to computational efficiency for VMAT dose computation on homogeneous medium, especially when a fine angular sampling rate is applied. This can support a reduction in dose errors from the angular under‐sampling effect by using a finer angular sampling rate, while still preserving a practical computing speed. For dose calculation on the heterogeneous phantom, the proposed algorithm with heterogeneity corrections can still offer a reasonable dosimetric accuracy with comparable computational efficiency to that of the CCC algorithm.
Conclusions We proposed a novel polar‐coordinate‐based pencil beam algorithm for VMAT dose computation that enables a better computational efficiency while maintaining clinically acceptable dosimetric accuracy and reducing dose error caused by the angular under‐sampling effect. It also provides a flexible VMAT dose computation structure that allows adjustable sampling rates and direct dose computation in regions of interest, which makes the algorithm potentially useful for clinical applications such as independent dose verification for VMAT patient‐specific QA.
-
more » « less
Abstract Purpose The radiobiological benefits afforded by spatially fractionated (GRID) radiation therapy pairs well with the dosimetric advantages of proton therapy. Inspired by the emergence of energy‐layer specific collimators in pencil beam scanning (PBS), this work investigates how the spot spacing and collimation can be optimized to maximize the therapeutic gains of a GRID treatment while demonstrating the integration of a dynamic collimation system (DCS) within a commercial beamline to deliver GRID treatments and experimentally benchmark Monte Carlo calculation methods.
Methods GRID profiles were experimentally benchmarked using a clinical DCS prototype that was mounted to the nozzle of the IBA‐dedicated nozzle system. Integral depth dose (IDD) curves and lateral profiles were measured for uncollimated and GRID‐collimated beamlets. A library of collimated GRID dose distributions were simulated by placing beamlets within a specified uniform grid and weighting the beamlets to achieve a volume‐averaged tumor cell survival equivalent to an open field delivery. The healthy tissue sparing afforded by the GRID distribution was then estimated across a range of spot spacings and collimation widths, which were later optimized based on the radiosensitivity of the tumor cell line and the nominal spot size of the PBS system. This was accomplished by using validated models of the IBA universal and dedicated nozzles.
Results Excellent agreement was observed between the measured and simulated profiles. The IDDs matched above 98.7% when analyzed using a 1%/1‐mm gamma criterion with some minor deviation observed near the Bragg peak for higher beamlet energies. Lateral profile distributions predicted using Monte Carlo methods agreed well with the measured profiles; a gamma passing rate of 95% or higher was observed for all in‐depth profiles examined using a 3%/2‐mm criteria. Additional collimation was shown to improve PBS GRID treatments by sharpening the lateral penumbra of the beamlets but creates a trade‐off between enhancing the valley‐to‐peak ratio of the GRID delivery and the dose‐volume effect. The optimal collimation width and spot spacing changed as a function of the tumor cell radiosensitivity, dose, and spot size. In general, a spot spacing below 2.0 cm with a collimation less than 1.0 cm provided a superior dose distribution among the specific cases studied.
Conclusions The ability to customize a GRID dose distribution using different collimation sizes and spot spacings is a useful advantage, especially to maximize the overall therapeutic benefit. In this regard, the capabilities of the DCS, and perhaps alternative dynamic collimators, can be used to enhance GRID treatments. Physical dose models calculated using Monte Carlo methods were experimentally benchmarked in water and were found to accurately predict the respective dose distributions of uncollimated and DCS‐collimated GRID profiles.
