Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “algorithmic”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 343 records · Page 19

Quantum-classical hybrid algorithm for the simulation of all-electron correlation

While chemical systems containing hundreds to thousands of electrons remain beyond the reach of quantum devices, hybrid quantum-classical algorithms present a promising pathway toward a quantum advantage. Hybrid algorithms treat the exponentially scaling part of the calculation-the static correlation-on the quantum computer and the non-exponentially scaling part-the dynamic correlation-on the classical computer. While a variety of algorithms have been proposed, the dependence of many methods on the total wave function limits the development of easy-to-use classical post-processing implementations. Here, we present a novel combination of quantum and classical algorithms, which computes the all-electron energy of a strongly correlated molecular system on the classical computer from the 2-electron reduced density matrix (2-RDM) evaluated on the quantum device. Significantly, we circumvent the wave function in the all-electron calculations by using density matrix methods that only require input of the statically correlated 2-RDM. Although the algorithm is completely general, we test it with two classical density matrix methods, the anti-Hermitian contracted Schrödinger equation (ACSE) and multiconfiguration pair-density functional theories, using the recently developed quantum ACSE method for simulating the statically correlated 2-RDM. Furthermore, we obtain experimental accuracy for the relative energies of all three benzyne isomers and thereby demonstrate the ability of the developed algorithm to achieve chemically relevant and accurate results on noisy intermediate-scale quantum devices.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Evaluating algorithmic bias on biomarker classification of breast cancer pathology reports

Objectives: This work evaluated algorithmic bias in biomarkers classification using electronic pathology reports from female breast cancer cases. Bias was assessed across 5 subgroups: cancer registry, race, Hispanic ethnicity, age at diagnosis, and socioeconomic status. Materials and Methods: We utilized 594 875 electronic pathology reports from 178 121 tumors diagnosed in Kentucky, Louisiana, New Jersey, New Mexico, Seattle, and Utah to train 2 deep-learning algorithms to classify breast cancer patients using their biomarkers test results. We used balanced error rate (BER), demographic parity (DP), equalized odds (EOD), and equal opportunity (EOP) to assess bias. Results: We found differences in predictive accuracy between registries, with the highest accuracy in the registry that contributed the most data (Seattle Registry, BER ratios for all registries >1.25). BER showed no significant algorithmic bias in extracting biomarkers (estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2) for race, Hispanic ethnicity, age at diagnosis, or socioeconomic subgroups (BER ratio <1.25). DP, EOD, and EOP all showed insignificant results. Discussion: We observed significant differences in BER by registry, but no significant bias using the DP, EOD, and EOP metrics for socio-demographic or racial categories. This highlights the importance of employing a diverse set of metrics for a comprehensive evaluation of model fairness. Conclusion: A thorough evaluation of algorithmic biases that may affect equality in clinical care is a critical step before deploying algorithms in the real world. We found little evidence of algorithmic bias in our biomarker classification tool. Artificial intelligence tools to expedite information extraction from clinical records could accelerate clinical trial matching and improve care.

60 APPLIED LIFE SCIENCES↗

Quantum Krylov subspace algorithms for ground- and excited-state energy estimation

Quantum Krylov subspace diagonalization (QKSD) algorithms provide a low-cost alternative to the conventional quantum phase estimation algorithm for estimating the ground- and excited-state energies of a quantum many-body system. While QKSD algorithms typically rely on using the Hadamard test for estimating Krylov subspace matrix elements of the form $\langle \phi_i|e^{-\widehat{H}τ}|\phi_j\rangle$, the associated quantum circuits require an ancilla qubit with controlled multiqubit gates that can be quite costly for near-term quantum hardware. In this paper, we show that a wide class of Hamiltonians relevant to condensed-matter physics and quantum chemistry contain symmetries that can be exploited to avoid the use of the Hadamard test. We propose a multifidelity estimation protocol that can be used to compute such quantities, showing that our approach, when combined with efficient single-fidelity estimation protocols, provides a substantial reduction in circuit depth. In addition, here we develop a unified theory of quantum Krylov subspace algorithms and present three quantum-classical algorithms for the ground- and excited-state energy estimation problems, where each algorithm provides various advantages and disadvantages in terms of total number of calls to the quantum computer, gate depth, classical complexity, and stability of the generalized eigenvalue problem within the Krylov subspace.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Automatic Code Generation for High-Performance Graph Algorithms

Graph problems are common across fields of scientific computing and social sciences. However, despite their importance, implementing graph algorithms effectively on modern computing systems is a challenging task that requires significant programming effort and generally results in customized implementations. Current computing and memory hierarchies are not architected for irregular computations resulting in challenges for graph algorithms to achieve high performance on those architectures. In this paper, we present GraphX, a novel compiler framework and DSL designed to simplify the development of efficient graph algorithms and achieve high performance on modern computing systems. GraphX consists of a DSL for efficient implementation of graph algorithms, various optimizations, such as support for sparse linear algebra and workspace transformations, optimized graph primitives, including semiring and masking, and a high-performance code generation engine. Using GraphX, users can implement graph algorithms using a semantically-rich language with graph-oriented operators. GraphX uses these semantics to automatically generate efficient code for target architectures, increasing performance and portability across architectures. The composable nature of GraphX makes it possible to extend the set of optimizations and architectures without modifying the source code. We demonstrate GraphX outperforms state-of-the-art graph libraries, such as LAGraph, up to $3.7 speedup in semiring operations, $2.19 speedup in an important sparse computational kernel, and $9.05 speedup in graph processing algorithms.

compiler, graph algorithms, semiring, masking, wor↗

Accelerated Voltage Regulation in Multi-Phase Distribution Networks Based on Hierarchical Distributed Algorithm

Here, we introduce a hierarchical distributed algorithm to solve optimal power flow (OPF) problems that aim at dispatching controllable distributed energy resources (DERs) for voltage regulation at minimum cost. The proposed algorithm highlights unprecedented scalability to large multi-phase distribution networks by jointly exploring the tree/subtrees structure of a large radial distribution network and the structure of the linearized distribution power flow (LinDistFlow) model to derive a hierarchical, distributed implementation of the primal-dual gradient algorithm that solves OPF. The proposed implementation significantly reduces the computation loads compared to the centrally coordinated implementation of the same primal-dual algorithm without compromising optimality. Numerical findings on a 4,521-node test feeder show that the designed algorithm achieves more than 10-fold acceleration in the speed of convergence compared to the centrally coordinated primal-dual algorithm through reducing and distributing computational loads.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Algorithmic scatter correction in dual-energy digital mammography

Purpose: Small calcifications are often the earliest and the main indicator of breast cancer. Dual-energy digital mammography (DEDM) has been considered as a promising technique to improve the detectability of calcifications since it can be used to suppress the contrast between adipose and glandular tissues of the breast. X-ray scatter leads to erroneous calculations of the DEDM image. Although the pinhole-array interpolation method can estimate scattered radiations, it requires extra exposures to measure the scatter and apply the correction. The purpose of this work is to design an algorithmic method for scatter correction in DEDM without extra exposures.Methods: In this paper, a scatter correction method for DEDM was developed based on the knowledge that scattered radiation has small spatial variation and that the majority of pixels in a mammogram are noncalcification pixels. The scatter fraction was estimated in the DEDM calculation and the measured scatter fraction was used to remove scatter from the image. The scatter correction method was implemented on a commercial full-field digital mammography system with breast tissue equivalent phantom and calcification phantom. The authors also implemented the pinhole-array interpolation scatter correction method on the system. Phantom results for both methods are presented and discussed. The authors compared the background DE calcification signals and the contrast-to-noise ratio (CNR) of calcifications in the three DE calcification images: image without scatter correction, image with scatter correction using pinhole-array interpolation method, and image with scatter correction using the authors' algorithmic method.Results: The authors' results show that the resultant background DE calcification signal can be reduced. The root-mean-square of background DE calcification signal of 1962 μm with scatter-uncorrected data was reduced to 194 μm after scatter correction using the authors' algorithmic method. The range of background DE calcification signals using scatter-uncorrected data was reduced by 58% with scatter-corrected data by algorithmic method. With the scatter-correction algorithm and denoising, the minimum visible calcification size can be reduced from 380 to 280 μm.Conclusions: When applying the proposed algorithmic scatter correction to images, the resultant background DE calcification signals can be reduced and the CNR of calcifications can be improved. This method has similar or even better performance than pinhole-array interpolation method in scatter correction for DEDM; moreover, this method is convenient and requires no extra exposure to the patient. Although the proposed scatter correction method is effective, it is validated by a 5-cm-thick phantom with calcifications and homogeneous background. The method should be tested on structured backgrounds to more accurately gauge effectiveness.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

SU-E-T-313: The Accuracy of the Acuros XB Advanced Dose Calculation Algorithm for IMRT Dose Distributions in Head and Neck

Purpose: To investigate the accuracy of the Acuros XB version 11 (AXB11) advanced dose calculation algorithm by comparing with Monte Caro (MC) calculations. The comparisons were performed with dose distributions for a virtual inhomogeneity phantom and intensity-modulated radiotherapy (IMRT) in head and neck. Methods: Recently, AXB based on Linear Boltzmann Transport Equation has been installed in the Eclipse treatment planning system (Varian Medical Oncology System, USA). The dose calculation accuracy of AXB11 was tested by the EGSnrc-MC calculations. In additions, AXB version 10 (AXB10) and Analytical Anisotropic Algorithm (AAA) were also used. First the accuracy of an inhomogeneity correction for AXB and AAA algorithms was evaluated by comparing with MC-calculated dose distributions for a virtual inhomogeneity phantom that includes water, bone, air, adipose, muscle, and aluminum. Next the IMRT dose distributions for head and neck were compared with the AXB and AAA algorithms and MC by means of dose volume histograms and three dimensional gamma analysis for each structure (CTV, OAR, etc.). Results: For dose distributions with the virtual inhomogeneity phantom, AXB was in good agreement with those of MC, except the dose in air region. The dose in air region decreased in order of MC<AXB11<AXB10. This may be caused by the difference of the electron cut-off energy for algorithms, ie: 0.700 MeV for MC, 0.711 MeV for AXB11, and 1.011 MeV for AXB 10. Since the AAA algorithm is based on the dose kernel of water, the doses in regions for air, bone, and aluminum considerably became higher than those of AXB and MC. The pass rates of the gamma analysis for IMRT dose distributions in head and neck were similar to those of MC in order of AXB11<AXB10<AAA. Conclusion: The dose calculation accuracy of AXB11 was almost equivalent to the MC dose calculation.

60 APPLIED LIFE SCIENCES↗

SU-E-T-577: Commissioning of a Deterministic Algorithm for External Photon Beams

Purpose: We report commissioning results for a deterministic algorithm for external photon beam treatment planning. A deterministic algorithm solves the radiation transport equations directly using a finite difference method, thus improve the accuracy of dose calculation, particularly under heterogeneous conditions with results similar to that of Monte Carlo (MC) simulation. Methods: Commissioning data for photon energies 6 – 15 MV includes the percentage depth dose (PDD) measured at SSD = 90 cm and output ratio in water (Spc), both normalized to 10 cm depth, for field sizes between 2 and 40 cm and depths between 0 and 40 cm. Off-axis ratio (OAR) for the same set of field sizes was used at 5 depths (dmax, 5, 10, 20, 30 cm). The final model was compared with the commissioning data as well as additional benchmark data. The benchmark data includes dose per MU determined for 17 points for SSD between 80 and 110 cm, depth between 5 and 20 cm, and lateral offset of up to 16.5 cm. Relative comparisons were made in a heterogeneous phantom made of cork and solid water. Results: Compared to the commissioning beam data, the agreement are generally better than 2% with large errors (up to 13%) observed in the buildup regions of the FDD and penumbra regions of the OAR profiles. The overall mean standard deviation is 0.04% when all data are taken into account. Compared to the benchmark data, the agreements are generally better than 2%. Relative comparison in heterogeneous phantom is in general better than 4%. Conclusion: A commercial deterministic algorithm was commissioned for megavoltage photon beams. In a homogeneous medium, the agreement between the algorithm and measurement at the benchmark points is generally better than 2%. The dose accuracy for a deterministic algorithm is better than a convolution algorithm in heterogeneous medium.

07 ISOTOPE AND RADIATION SOURCES↗

Gradient maintenance: A new algorithm for fast online replanning

Purpose: Clinical use of online adaptive replanning has been hampered by the unpractically long time required to delineate volumes based on the image of the day. The authors propose a new replanning algorithm, named gradient maintenance (GM), which does not require the delineation of organs at risk (OARs), and can enhance automation, drastically reducing planning time and improving consistency and throughput of online replanning. Methods: The proposed GM algorithm is based on the hypothesis that if the dose gradient toward each OAR in daily anatomy can be maintained the same as that in the original plan, the intended plan quality of the original plan would be preserved in the adaptive plan. The algorithm requires a series of partial concentric rings (PCRs) to be automatically generated around the target toward each OAR on the planning and the daily images. The PCRs are used in the daily optimization objective function. The PCR dose constraints are generated with dose–volume data extracted from the original plan. To demonstrate this idea, GM plans generated using daily images acquired using an in-room CT were compared to regular optimization and image guided radiation therapy repositioning plans for representative prostate and pancreatic cancer cases. Results: The adaptive replanning using the GM algorithm, requiring only the target contour from the CT of the day, can be completed within 5 min without using high-power hardware. The obtained adaptive plans were almost as good as the regular optimization plans and were better than the repositioning plans for the cases studied. Conclusions: The newly proposed GM replanning algorithm, requiring only target delineation, not full delineation of OARs, substantially increased planning speed for online adaptive replanning. The preliminary results indicate that the GM algorithm may be a solution to improve the ability for automation and may be especially suitable for sites with small-to-medium size targets surrounded by several critical structures.

60 APPLIED LIFE SCIENCES↗

SU-E-T-339: Dosimetric Verification of Acuros XB Dose Calculation Algorithm On An Air Cavity for 6-MV Flattening Filter-Free Beam

Purpose: This study was to verify the accuracy of Acuros XB (AXB) dose calculation algorithm on an air cavity for a single radiation field using 6-MV flattening filter-free (FFF) beam. Methods: A rectangular slab phantom containing an air cavity was made for this study. The CT images of the phantom for dose calculation were scanned with and without film at measurement depths (4.5, 5.5, 6.5 and 7.5 cm). The central axis doses (CADs) and the off-axis doses (OADs) were measured by film and calculated with Analytical Anisotropic Algorithm (AAA) and AXB for field sizes ranging from 2 Χ 2 to 5 Χ 5 cm{sup 2} of 6-MV FFF beams. Both algorithms were divided into AXB-w and AAA -w when included the film in phantom for dose calculation, and AXB-w/o and AAA-w/o in calculation without film. The calculated OADs for both algorithms were compared with the measured OADs and difference values were determined using root means squares error (RMSE) and gamma evaluation. Results: The percentage differences (%Diffs) between the measured and calculated CAD for AXB-w was most agreement than others. Compared to the %Diff with and without film, the %Diffs with film were decreased than without within both algorithms. The %Diffs for both algorithms were reduced with increasing field size and increased relative to the depth increment. RMSEs of CAD for AXB-w were within 10.32% for both inner-profile and penumbra, while the corresponding values of AAA-w appeared to 96.50%. Conclusion: This study demonstrated that the dose calculation with AXB within air cavity shows more accurate than with AAA compared to the measured dose. Furthermore, we found that the AXB-w was superior to AXB-w/o in this region when compared against the measurements.

60 APPLIED LIFE SCIENCES↗

SU-F-SPS-01: Accuracy of the Small Field Dosimetry Using Acuros XB and AAA Dose Calculation Algorithms of Eclipse Treatment Planning System Within and Beyond Heterogeneous Media for Trubeam 2.0 Unit

Purpose: In this study, the comparison of dosimetric accuracy of Acuros XB and AAA algorithms were investigated for small radiation fields incident on homogeneous and heterogeneous geometries Methods: Small open fields of Truebeam 2.0 unit (1×1, 2×2, 3×3, 4×4 fields) were used for this study. The fields were incident on homogeneous phantom and in house phantom containing lung, air, and bone inhomogeneities. Using the same film batch, the net OD to dose calibration curve was obtaine dusing Trubeam 2.0 for 6 MV, 6 FFF, 10 MV, 10 FFF, 15 MV energies by delivering 0- 800 cGy. Films were scanned 48 hours after irradiation using an Epson 1000XL flatbed scanner. The dosimetric accuracy of Acuros XB and AAA algorithms in the presence of the inhomogeneities was compared against EBT3 film dosimetry Results: Open field tests in a homogeneous phantom showed good agreement betweent wo algorithms and measurement. For Acuros XB, minimum gamma analysis passin grates between measured and calculated dose distributions were 99.3% and 98.1% for homogeneousand inhomogeneous fields in thecase of lung and bone respectively. For AAA, minimum gamma analysis passingrates were 99.1% and 96.5% for homogeneous and inhomogeneous fields respectively for all used energies and field sizes.In the case of the air heterogeneity, the differences were larger for both calculations algorithms. Over all, when compared to measurement, theAcuros XB had beter agreement than AAA. Conclusion: The Acuros XB calculation algorithm in the TPS is an improvemen tover theexisting AAA algorithm. Dose discrepancies were observed for in the presence of air inhomogeneities.

60 APPLIED LIFE SCIENCES↗

SU-F-SPS-05: Experimental Validation of Peripheral Dose Distribution of Electron Beams for Eclipse Electron Monte Carlo Algorithm

Purpose: In this study, the two available calculation algorithms of the Varian Eclipse treatment planning system(TPS), the electron Monte Carlo(eMC) and General Gaussian Pencil Beam(GGPB) algorithms were used to compare measured and calculated peripheral dose distribution of electron beams. Methods: Peripheral dose measurements were carried out for 6, 9, 12, 15, 18 and 22 MeV electron beams of Varian Triology machine using parallel plate ionization chamber and EBT3 films in the slab phantom. Measurements were performed for 6×6, 10×10 and 25×25cm{sup 2} cone sizes at dmax of each energy up to 20cm beyond the field edges. Using the same film batch, the net OD to dose calibration curve was obtained for each energy. Films were scanned 48 hours after irradiation using an Epson 1000XL flatbed scanner. Dose distribution measured using parallel plate ionization chamber and EBT3 film and calculated by eMC and GGPB algorithms were compared. The measured and calculated data were then compared to find which algorithm calculates peripheral dose distribution more accurately. Results: The agreement between measurement and eMC was better than GGPB. The TPS underestimated the out of field doses. The difference between measured and calculated doses increase with the cone size. The largest deviation between calculated and parallel plate ionization chamber measured dose is less than 4.93% for eMC, but it can increase up to 7.51% for GGPB. For film measurement, the minimum gamma analysis passing rates between measured and calculated dose distributions were 98.2% and 92.7% for eMC and GGPB respectively for all field sizes and energies. Conclusion: Our results show that the Monte Carlo algorithm for electron planning in Eclipse is more accurate than previous algorithms for peripheral dose distributions. It must be emphasized that the use of GGPB for planning large field treatments with 6 MeV could lead to inaccuracies of clinical significance.

60 APPLIED LIFE SCIENCES↗

SU-F-I-09: Improvement of Image Registration Using Total-Variation Based Noise Reduction Algorithms for Low-Dose CBCT

Purpose: To study the effect of total-variation based noise reduction algorithms to improve the image registration of low-dose CBCT for patient positioning in radiation therapy. Methods: In low-dose CBCT, the reconstructed image is degraded by excessive quantum noise. In this study, we developed a total-variation based noise reduction algorithm and studied the effect of the algorithm on noise reduction and image registration accuracy. To study the effect of noise reduction, we have calculated the peak signal-to-noise ratio (PSNR). To study the improvement of image registration, we performed image registration between volumetric CT and MV- CBCT images of different head-and-neck patients and calculated the mutual information (MI) and Pearson correlation coefficient (PCC) as a similarity metric. The PSNR, MI and PCC were calculated for both the noisy and noise-reduced CBCT images. Results: The algorithms were shown to be effective in reducing the noise level and improving the MI and PCC for the low-dose CBCT images tested. For the different head-and-neck patients, a maximum improvement of PSNR of 10 dB with respect to the noisy image was calculated. The improvement of MI and PCC was 9% and 2% respectively. Conclusion: Total-variation based noise reduction algorithm was studied to improve the image registration between CT and low-dose CBCT. The algorithm had shown promising results in reducing the noise from low-dose CBCT images and improving the similarity metric in terms of MI and PCC.

60 APPLIED LIFE SCIENCES↗

SU-F-J-77: Variations in the Displacement Vector Fields Calculated by Different Deformable Image Registration Algorithms Used in Helical, Axial and Cone-Beam CT Images of a Mobile

Purpose: To investigate quantitatively the displacement-vector-fields (DVF) obtained from different deformable image registration algorithms (DIR) in helical (HCT), axial (ACT) and cone-beam CT (CBCT) to register CT images of a mobile phantom and its correlation with motion amplitudes and frequencies. Methods: HCT, ACT and CBCT are used to image a mobile phantom which includes three targets with different sizes that are manufactured from water-equivalent material and embedded in low density foam. The phantom is moved with controlled motion patterns where a range of motion amplitudes (0–40mm) and frequencies (0.125–0.5Hz) are used. The CT images obtained from scanning of the mobile phantom are registered with the stationary CT-images using four deformable image registration algorithms including demons, fast-demons, Horn-Schunk and Locas-Kanade from DIRART software. Results: The DVF calculated by the different algorithms correlate well with the motion amplitudes that are applied on the mobile phantom where maximal DVF increase linearly with the motion amplitudes of the mobile phantom in CBCT. Similarly in HCT, DVF increase linearly with motion amplitude, however, its correlation is weaker than CBCT. In ACT, the DVF’s do not correlate well with the motion amplitudes where motion induces strong image artifacts and DIR algorithms are not able to deform the ACT image of the mobile targets to the stationary targets. Three DIR-algorithms produce comparable values and patterns of the DVF for certain CT imaging modality. However, DVF from fast-demons deviated strongly from other algorithms at large motion amplitudes. Conclusion: In CBCT and HCT, the DVF correlate well with the motion amplitude of the mobile phantom. However, in ACT, DVF do not correlate with motion amplitudes. Correlations of DVF with motion amplitude as in CBCT and HCT imaging techniques can provide information about unknown motion parameters of the mobile organs in real patients as demonstrated in this phantom visibility study.

60 APPLIED LIFE SCIENCES↗

SU-F-T-600: Influence of Acuros XB and AAA Dose Calculation Algorithms On Plan Quality Metrics and Normal Lung Doses in Lung SBRT

Purpose: To study the influence of superposition-beam model (AAA) and determinant-photon transport-solver (Acuros XB) dose calculation algorithms on the treatment plan quality metrics and on normal lung dose in Lung SBRT. Methods: Treatment plans of 10 Lung SBRT patients were randomly selected. Patients were prescribed to a total dose of 50-54Gy in 3–5 fractions (10?5 or 18?3). Doses were optimized accomplished with 6-MV using 2-arcs (VMAT). Doses were calculated using AAA algorithm with heterogeneity correction. For each plan, plan quality metrics in the categories- coverage, homogeneity, conformity and gradient were quantified. Repeat dosimetry for these AAA treatment plans was performed using AXB algorithm with heterogeneity correction for same beam and MU parameters. Plan quality metrics were again evaluated and compared with AAA plan metrics. For normal lung dose, V{sub 20} and V{sub 5} to (Total lung- GTV) were evaluated. Results: The results are summarized in Supplemental Table 1. PTV volume was mean 11.4 (±3.3) cm{sup 3}. Comparing RTOG 0813 protocol criteria for conformality, AXB plans yielded on average, similar PITV ratio (individual PITV ratio differences varied from −9 to +15%), reduced target coverage (−1.6%) and increased R50% (+2.6%). Comparing normal lung doses, the lung V{sub 20} (+3.1%) and V{sub 5} (+1.5%) were slightly higher for AXB plans compared to AAA plans. High-dose spillage ((V105%PD - PTV)/ PTV) was slightly lower for AXB plans but the % low dose spillage (D2cm) was similar between the two calculation algorithms. Conclusion: AAA algorithm overestimates lung target dose. Routinely adapting to AXB for dose calculations in Lung SBRT planning may improve dose calculation accuracy, as AXB based calculations have been shown to be closer to Monte Carlo based dose predictions in accuracy and with relatively faster computational time. For clinical practice, revisiting dose-fractionation in Lung SBRT to correct for dose overestimates attributable to algorithm may very well be warranted.

60 APPLIED LIFE SCIENCES↗

SU-G-201-09: Evaluation of a Novel Machine-Learning Algorithm for Permanent Prostate Brachytherapy Treatment Planning

Purpose: A novel, automated, algorithm for permanent prostate brachytherapy (PPB) treatment planning has been developed. The novel approach uses machine-learning (ML), a form of artificial intelligence, to substantially decrease planning time while simultaneously retaining the clinical intuition of plans created by radiation oncologists. This study seeks to compare the ML algorithm against expert-planned PPB plans to evaluate the equivalency of dosimetric and clinical plan quality. Methods: Plan features were computed from historical high-quality PPB treatments (N = 100) and stored in a relational database (RDB). The ML algorithm matched new PPB features to a highly similar case in the RDB; this initial plan configuration was then further optimized using a stochastic search algorithm. PPB pre-plans (N = 30) generated using the ML algorithm were compared to plan variants created by an expert dosimetrist (RT), and radiation oncologist (MD). Planning time and pre-plan dosimetry were evaluated using a one-way Student’s t-test and ANOVA, respectively (significance level = 0.05). Clinical implant quality was evaluated by expert PPB radiation oncologists as part of a qualitative study. Results: Average planning time was 0.44 ± 0.42 min compared to 17.88 ± 8.76 min for the ML algorithm and RT, respectively, a significant advantage [t(9), p = 0.01]. A post-hoc ANOVA [F(2,87) = 6.59, p = 0.002] using Tukey-Kramer criteria showed a significantly lower mean prostate V150% for the ML plans (52.9%) compared to the RT (57.3%), and MD (56.2%) plans. Preliminary qualitative study results indicate comparable clinical implant quality between RT and ML plans with a trend towards preference for ML plans. Conclusion: PPB pre-treatment plans highly comparable to those of an expert radiation oncologist can be created using a novel ML planning model. The use of an ML-based planning approach is expected to translate into improved PPB accessibility and plan uniformity.

60 APPLIED LIFE SCIENCES↗

WE-AB-209-06: Dynamic Collimator Trajectory Algorithm for Use in VMAT Treatment Deliveries

Purpose: To develop advanced dynamic collimator positioning algorithms for optimal beam’s-eye-view (BEV) fitting of targets in VMAT procedures, including multiple metastases stereotactic radiosurgery procedures. Methods: A trajectory algorithm was developed, which can dynamically modify the angle of the collimator as a function of VMAT control point to provide optimized collimation of target volume(s). Central to this algorithm is a concept denoted “whitespace”, defined as area within the jaw-defined BEV field, outside of the PTV, and not shielded by the MLC when fit to the PTV. Calculating whitespace at all collimator angles and every control point, a two-dimensional topographical map depicting the tightness-of-fit of the MLC was generated. A variety of novel searching algorithms identified a number of candidate trajectories of continuous collimator motion. Ranking these candidate trajectories according to their accrued whitespace value produced an optimal solution for navigation of this map. Results: All trajectories were normalized to minimum possible (i.e. calculated without consideration of collimator motion constraints) accrued whitespace. On an acoustic neuroma case, a random walk algorithm generated a trajectory with 151% whitespace; random walk including a mandatory anchor point improved this to 148%; gradient search produced a trajectory with 137%; and bi-directional gradient search generated a trajectory with 130% whitespace. For comparison, a fixed collimator angle of 30° and 330° accumulated 272% and 228% of whitespace, respectively. The algorithm was tested on a clinical case with two metastases (single isocentre) and identified collimator angles that allow for simultaneous irradiation of the PTVs while minimizing normal tissue irradiation. Conclusion: Dynamic collimator trajectories have the potential to improve VMAT deliveries through increased efficiency and reduced normal tissue dose, especially in treatment of multiple cranial metastases, without significant safety concerns that hinder immediate clinical implementation.

60 APPLIED LIFE SCIENCES↗

S-OPT: A Points Selection Algorithm for Hyper-Reduction in Reduced Order Models

While projection-based reduced order models can reduce the dimension of full order solutions, the resulting reduced models may still contain terms that scale with the full order dimension. Hyper-reduction techniques are sampling-based methods that further reduce this computational complexity by approximating such terms with a much smaller dimension. The goal of this work is to introduce the points selection algorithm developed by Shin and Xiu as a hyper-reduction method. The selection algorithm was originally proposed as a stochastic collocation method for uncertainty quantification. Since the algorithm aims at maximizing a quantity $\mathcal{S}$ that measures both the column orthogonality and the determinant, we refer to the algorithm as S-OPT. Numerical examples are provided to demonstrate the performance of S-OPT and to compare its performance with a gappy proper orthogonal decomposition (POD) algorithm. Here, we found that using the S-OPT algorithm is shown to predict the full order solutions with higher accuracy than gappy POD especially when the number of sampling points is small, although we note that S-OPT shows slow asymptotic convergence with respect to the number of samples for some applications, e.g., Lagrangian hydrodynamics.

97 MATHEMATICS AND COMPUTING↗