Scalable radiotherapy data curation infrastructure for deep-learning based autosegmentation of organs-at-risk: A case study in head and neck cancer.

In this era of patient-centered, outcomes-driven and adaptive radiotherapy, deep learning is now being successfully applied to tackle imaging-related workflow bottlenecks such as autosegmentation and dose planning. These applications typically require supervised learning approaches enabled by relatively large, curated radiotherapy datasets which are highly reflective of the contemporary standard of care. However, little has been previously published describing technical infrastructure, recommendations, methods or standards for radiotherapy dataset curation in a holistic fashion. Our radiation oncology department has recently embarked on a large-scale project in partnership with an external partner to develop deep-learning-based tools to assist with our radiotherapy workflow, beginning with autosegmentation of organs-at-risk. This project will require thousands of carefully curated radiotherapy datasets comprising all body sites we routinely treat with radiotherapy. Given such a large project scope, we have approached the need for dataset curation rigorously, with an aim towards building infrastructure that is compatible with efficiency, automation and scalability. Focusing on our first use-case pertaining to head and neck cancer, we describe our developed infrastructure and novel methods applied to radiotherapy dataset curation, inclusive of personnel and workflow organization, dataset selection, expert organ-at-risk segmentation, quality assurance, patient de-identification, data archival and transfer. Over the course of approximately 13 months, our expert multidisciplinary team generated 490 curated head and neck radiotherapy datasets. This task required approximately 6000 human-expert hours in total (not including planning and infrastructure development time). This infrastructure continues to evolve and will support ongoing and future project efforts.

A comprehensive system for dosimetric commissioning and Monte Carlo validation for the small animal radiation research platform.

Our group has constructed the small animal radiation research platform (SARRP) for delivering focal, kilo-voltage radiation to targets in small animals under robotic control using cone-beam CT guidance. The present work was undertaken to support the SARRP's treatment planning capabilities. We have devised a comprehensive system for characterizing the radiation dosimetry in water for the SARRP and have developed a Monte Carlo dose engine with the intent of reproducing these measured results. We find that the SARRP provides sufficient therapeutic dose rates ranging from 102 to 228 cGy min(-1) at 1 cm depth for the available set of high-precision beams ranging from 0.5 to 5 mm in size. In terms of depth-dose, the mean of the absolute percentage differences between the Monte Carlo calculations and measurement is 3.4% over the full range of sampled depths spanning 0.5-7.2 cm for the 3 and 5 mm beams. The measured and computed profiles for these beams agree well overall; of note, good agreement is observed in the profile tails. Especially for the smallest 0.5 and 1 mm beams, including a more realistic description of the effective x-ray source into the Monte Carlo model may be important.

High throughput film dosimetry in homogeneous and heterogeneous media for a small animal irradiator.

PURPOSE: We have established a high-throughput Gafchromic film dosimetry protocol for narrow kilovoltage beams in homogeneous and heterogeneous media for small-animal radiotherapy applications. The kV beam characterization is based on extensive Gafchromic film dosimetry data acquired in homogeneous and heterogeneous media. An empirical model is used for parameterization of depth and off-axis dependence of measured data. METHODS: We have modified previously published methods of film dosimetry to suit the specific tasks of the study. Unlike film protocols used in previous studies, our protocol employs simultaneous multi-channel scanning and analysis of up to nine Gafchromic films per scan. A scanner and background correction were implemented to improve accuracy of the measurements. Measurements were taken in homogeneous and inhomogeneous phantoms at 220 kVp and a field size of 5 × 5 mm(2). The results were compared against Monte Carlo simulations. RESULTS: Dose differences caused by variations in background signal were effectively removed by the corrections applied. Measurements in homogeneous phantoms were used to empirically characterize beam data in homogeneous and heterogeneous media. Film measurements in inhomogeneous phantoms and their empirical parameterization differed by about 2%-3%. The model differed from MC by about 1% (water, lung) to 7% (bone). Good agreement was found for measured and modelled off-axis ratios. CONCLUSIONS: EBT2 films are a valuable tool for characterization of narrow kV beams, though care must be taken to eliminate disturbances caused by varying background signals. The usefulness of the empirical beam model in interpretation and parameterization of film data was demonstrated.

A fast analytic dose calculation method for arc treatments for kilovoltage small animal irradiators.

Arc treatments require calculation of dose for collections of discrete gantry angles. The sampling of angles must balance between short computation time of small angle sets and the better calculation reliability of large sets. In this paper, an analytical formula is presented that allows calculation of dose delivered during continuous rotation of the gantry. The formula holds valid for continuous short arcs of up to about 30° and is derived by integrating a dose formula over gantry angles within a small angle approximation. Doses for longer arcs may be obtained in terms of doses for shorter arcs. The formula is derived with an empirical beam model in water and extended to inhomogeneous media. It is validated with experimental data obtained by applying arc treatment using kV small animal irradiator to a phantom of solid water and lung-equivalent material. The results are a promising step towards efficient 3D dose calculation and inverse planning purposes. In principle, this method also applies to VMAT dose calculation and optimization but requires extensions.

Deformable image registration for cone-beam CT guided transoral robotic base-of-tongue surgery.

Transoral robotic surgery (TORS) offers a minimally invasive approach to resection of base-of-tongue tumors. However, precise localization of the surgical target and adjacent critical structures can be challenged by the highly deformed intraoperative setup. We propose a deformable registration method using intraoperative cone-beam computed tomography (CBCT) to accurately align preoperative CT or MR images with the intraoperative scene. The registration method combines a Gaussian mixture (GM) model followed by a variation of the Demons algorithm. First, following segmentation of the volume of interest (i.e. volume of the tongue extending to the hyoid), a GM model is applied to surface point clouds for rigid initialization (GM rigid) followed by nonrigid deformation (GM nonrigid). Second, the registration is refined using the Demons algorithm applied to distance map transforms of the (GM-registered) preoperative image and intraoperative CBCT. Performance was evaluated in repeat cadaver studies (25 image pairs) in terms of target registration error (TRE), entropy correlation coefficient (ECC) and normalized pointwise mutual information (NPMI). Retraction of the tongue in the TORS operative setup induced gross deformation >30 mm. The mean TRE following the GM rigid, GM nonrigid and Demons steps was 4.6, 2.1 and 1.7 mm, respectively. The respective ECC was 0.57, 0.70 and 0.73, and NPMI was 0.46, 0.57 and 0.60. Registration accuracy was best across the superior aspect of the tongue and in proximity to the hyoid (by virtue of GM registration of surface points on these structures). The Demons step refined registration primarily in deeper portions of the tongue further from the surface and hyoid bone. Since the method does not use image intensities directly, it is suitable to multi-modality registration of preoperative CT or MR with intraoperative CBCT. Extending the 3D image registration to the fusion of image and planning data in stereo-endoscopic video is anticipated to support safer, high-precision base-of-tongue robotic surgery.

SU-E-T-385: Accelerated Beam Delivery with MLC Gaps in IMRT Fields.

PURPOSE: By default, the 4-mm wide leaf-pairs of the Elekta Beam Modulator MLC not used for field shaping are moved to under the primary collimator to minimize the leakage dose. The trade-off is the increased beam delivery time in the case of IMRT. This study examines the dosimetric impact and improved delivery efficiency by allowing the unused leaf-pair as 2mm gap to stay within the PTV aperture during beam-on. METHODS: The capability of the treatment planning system (Pinnacle) in modeling the central and off-axis dose through 2mm MLC gaps, respectively, was investigated. EBT film measurement performed in a solid water phantom was compared with the computed dose. For two representative IMRT cases, the dosimetric features of treatment plans generated with and without MLC gaps in the field were compared. Similarly, the delivery times for 13 IMRT cases with and without MLC gaps were compared. RESULTS: The peak dose of the leaf gap computed with Pinnacle was dependent on dose grid resolution. The difference between a 0.5mm and 2mm dose grid was about 16.5% for 6MV and 15MV beam respectively. Using a 0.5mm dose grid, the maximum difference between measured and calculated dose was < 1% for a central axis gap, and < 3% for an off-axis gap. The maximum dose through a 2mm×4mm leaf gap was estimated < 10cGy for a typical IMRT plan. DVH comparison showed an increase of target dose up to 2% for the IMRT plans with gap, with minimal dose change to all critical structures. For the 13 IMRT cases, the beam delivery times were reduced by 27+/-9.7%, from 8.7+/-4.5 minutes to 6.5+/-4.0 minutes. CONCLUSIONS: The beam delivery is accelerated by utilizing minimal gaps in the IMRT fields. The increased dose to the target is acceptable and is minimal to all critical structures.

Reduction of the spin-orbit splittings at the n = 28 shell closure.

The N = 28 shell closure has been investigated via the 46Ar(d,p)47Ar transfer reaction in inverse kinematics. Energies and spectroscopic factors of the neutron p(3/2), p(1/2), and f(5/2) states in 47Ar were determined and compared to those of the 49Ca isotone. We deduced a reduction of the N = 28 gap by 330(90) keV and spin-orbit weakenings of approximately 10(2) and 45(10)% for the f and p states, respectively. Such large variations for the f and p spin-orbit splittings could be accounted for by the proton-neutron tensor force and by the density dependence of the spin-orbit interaction, respectively. This contrasts with the picture of the spin-orbit interaction as a surface term only.