I El Naqa

A radiomics model from joint FDG-PET and MRI texture features for the prediction of lung metastases in soft-tissue sarcomas of the extremities

This study aims at developing a joint FDG-PET and MRI texture-based model for the early evaluation of lung metastasis risk in soft-tissue sarcomas (STSs). We investigate if the creation of new composite textures from the combination of FDG-PET and MR imaging information could better identify aggressive tumours. Towards this goal, a cohort of 51 patients with histologically proven STSs of the extremities was retrospectively evaluated. All patients had pre-treatment FDG-PET and MRI scans comprised of T1-weighted and T2-weighted fat-suppression sequences (T2FS). Nine non-texture features (SUV metrics and shape features) and forty-one texture features were extracted from the tumour region of separate (FDG-PET, T1 and T2FS) and fused (FDG-PET/T1 and FDG-PET/T2FS) scans. Volume fusion of the FDG-PET and MRI scans was implemented using the wavelet transform. The influence of six different extraction parameters on the predictive value of textures was investigated. The incorporation of features into multivariable models was performed using logistic regression. The multivariable modeling strategy involved imbalance-adjusted bootstrap resampling in the following four steps leading to final prediction model construction: (1) feature set reduction; (2) feature selection; (3) prediction performance estimation; and (4) computation of model coefficients. Univariate analysis showed that the isotropic voxel size at which texture features were extracted had the most impact on predictive value. In multivariable analysis, texture features extracted from fused scans significantly outperformed those from separate scans in terms of lung metastases prediction estimates. The best performance was obtained using a combination of four texture features extracted from FDG-PET/T1 and FDG-PET/T2FS scans. This model reached an area under the receiver-operating characteristic curve of 0.984 ± 0.002, a sensitivity of 0.955 ± 0.006, and a specificity of 0.926 ± 0.004 in bootstrapping evaluations. Ultimately, lung metastasis risk assessment at diagnosis of STSs could improve patient outcomes by allowing better treatment adaptation.

Sci—Thur AM: YIS ‐ 02: Radiogenomic Modeling of Normal Tissue Toxicities in Prostate Cancer Patients Receiving Hypofractionated Radiotherapy

Inter-patient radiation sensitivity variability has recently been shown to have a genetic component. This genetic component may play a key role in explaining the fluctuating rates of radiation-induced toxicities (RITs). Single nucleotide polymorphisms (SNPs) have thus far yielded inconsistent results in delineating RITs while copy number variations (CNVs) have not yet been investigated for such purposes. We explore a radiogenomic modeling approach to investigate the association of CNVs and SNPs, along with clinical and dosimetric variables, in radiation induced rectal bleeding (RB) and erectile dysfunction (ED) in prostate cancer patients treated with curative hypofractionated irradiation. A cohort of 62 prostate cancer patients who underwent hypofractionated radiotherapy (66 Gy in 22 fractions) between 2002 to 2010 were retrospectively genotyped for CNV and SNP rs5489 in the xrcc1 DNA repair gene. Late toxicity rates for RB grade 2 & 3 and grade 3 alone were 29.0% and 12.9%, respectively. ED toxicity was found to be 62.9%. Radiogenomic model performance was evaluated using receiver operating characteristic area under the curve (AUC) and resampling by cross-validation. Binary variables were evaluated using Chi-squared contingency table analysis and multivariate models by Spearmans rank correlation coefficient (rs). Ten patients were found to have three copies of xrcc1 CNV (RB: χ2=14.6, p<0.001 and ED: χ2=4.88, p=0.0272) and twelve had heterozygous rs25489 SNP (RB: χ2=0.278, p=0.599 and ED: χ2=0.112, p=0.732). Radiogenomic modeling yielded significant, cross-validated NTCP models for RB (AUC=0.665) and ED (AUC=0.754). These results indicate that CNVs may be potential predictive biomarkers of both late ED and RB.

SU‐E‐J‐69: An Anomaly Detector for Radiotherapy Quality Assurance Using Machine Learning

Purpose: To investigate a new approach and automated quality assurance (QA) tool for detectingradiotherapy errors using machine learning. The proposed datamining approach utilizes anomaly detection based on one‐class estimation to overcome computational challenges of detecting rare events encountered in currently existing techniques. Methods: The proposed anomaly detection approach captures regions in the input space of radiotherapy data where the safe class probability density lives and estimate errors as outliers that reside outside this support region. To model nonlinear support regions, we used a support vector machine (SVM) formalism, in which the QA data is mapped into higher dimensional space using kernel functions to achieve maximal separability and is denoted QA‐SVM detector. We demonstrated our method using forty‐three treatments plans from patients who received stereotactic body radiation therapy(SBRT) for lungcancer. Results: Features related to monitor units, beam energies, number of beams, number of fractions, in addition to V20 were extracted from the DICOM files and used to train the QA‐SVM detector. Using principle component analysis, 5 features were selected and used subsequently. For testing, we used a combination of cases that were considered “safe” with class label “+1” and simulated cases, which were considered “risky” with class label “‐1“. A radial basis function was used as kernel and false positive limit was set at 10%. Our preliminary results indicate a training accuracy of 84% on cross‐validation and testing accuracy of 80% with 100% positive predictive value and 80% negative predictive value. Conclusions: We presented a new approach and tool based on machine learning to overcome the problem of direct modeling of QA errors and rare events in radiotherapy. The tool will be very valuable for automated QA and safety management for patients who undergo radiotherapytreatment.

MO‐FG‐303‐02: BEST IN PHYSICS (THERAPY): Cherenkov Emission Dosimetry: Feasibility for Electron Radiotherapy

Purpose: To investigate from first principles, corroborated by Monte Carlo simulations and experimental measurements, the feasibility of developing a relative Cherenkov emission (CE) dosimetry protocol for electron beam radiotherapy. Methods: Monte Carlo (MC) simulations of mono-energetic electrons incident on water were carried out in Geant4. Percent depth Cherenkov emission (PDCE) and dose (PDD) distributions were scored for incidence energies of 4, 6, 9, 12, 15, and 18 MeV. PDCE-to-PDD analytical conversion models were developed from least-squares data fits generated for PDD as a function of PDCE at the same depth and at different depths. Experimental techniques for validation of these models are examined. Results: Same-depth PDD versus PDCE data fits indicate that although the relationship is linear to first order (correlation r > 0.9 for all energies), it is much more accurately approximated by separate linear and quadratic models for the build-up and drop-off regions, respectively (r > 0.999), which is theoretically underpinned. To understand the source of this relationship and its basis for developing robust conversion models, an approximate quadratic first-principles model was derived and found in agreement with MC/measured data (20% deviation at worst). Conversely, data fits of PDD versus different-depth PDCE unveiled a depth-invariant effective point of measurement of 1.5–2.1 mm downstream with 4–18 MeV incidence, respectively (r > 0.999 in the drop-off region). We present an analytical first-principles justification for this shift. This method led to errors of <1% in drop-off region PDD (<2% for PDD<20% with 4 MeV incidence) and <0.2 mm in practical range prediction. Conclusion: We present robust quantitative prediction models, derived from first-principles and supported by simulation and measurement, for relative dose from Cherenkov emission by high-energy electrons. This constitutes a major step towards development of protocols for routine clinical quality assurance as well as real-time in vivo Cherenkov dosimetry in radiotherapy. The authors acknowledge partial support by Fonds de recherche du Quebec - Nature et technologies (FRQNT), CREATE Medical Physics Research Training Network grant of the Natural Sciences and Engineering Research Council of Canada (NSERC), CREATE Integrated Sensor Systems grant of NSERC, the Canadian Institutes of Health Research (CIHR), and NSERC.

MO-DE-BRA-04: The CREATE Medical Physics Research Training Network: Training of New Generation Innovators

Purpose: Over the past century, physicists have played a major role in transforming scientific discovery into everyday clinical applications. However, with the increasingly stringent requirements to regulate medical physics as a health profession, the role of physicists as scientists and innovators has become at serious risk of erosion. These challenges trigger the need for a new, revolutionized training program at the graduate level that respects scientific rigor, attention for medical physics-relevant developments in basic sciences, innovation and entrepreneurship. Methods: A grant proposal was funded by the Collaborative REsearch and Training Experience program (CREATE) of the Natural Sciences and Engineering Research Council (NSERC) of Canada. This enabled the creation of the Medical Physics Research Training Network (MPRTN) around two CAMPEP-accredited medical physics programs. Members of the network consist of medical device companies, government (research and regulatory) and academia. The MPRTN/CREATE program proposes a curriculum with three main themes: (1) radiation physics, (2) imaging & image processing and (3) radiation response, outcomes and modeling. Results: The MPRTN was created mid 2013 (mprtn.com) and features (1) four new basic Ph.D. courses; (2) industry participation in research projects; (3) formal job-readiness training with involvement of guest faculty from academia, government and industry. MPRTN activities since 2013 include 22 conferences; 7 workshops and 4 exchange travels. Three patents were filed or issued, nine awards/best papers were won. Fifteen journal publications were accepted/published, 102 conference abstracts. There are now 13 industry partners. Conclusion: A medical physics research training network has been set up with the goal to harness graduate student’s job-readiness for industry, government and academia in addition to the conventional clinical role. Two years after inception, significant successes have been booked, but the true challenge will be to demonstrate that with this training philosophy CREATE scholars gain access to a much broader job market. Supported by the Natural Sciences and Engineering Research Council (NSERC) Canada

Poster — Thur Eve — 18: Cherenkov Emission By High-Energy Radiation Therapy Beams: A Characterization Study

We investigate Cherenkov emission (CE) by radiotherapy beams via radiation dose-versus-CE correlation analyses, CE detection optimization by means of a spectral shift towards the near-infrared (NIR) window of biological tissue, and comparison of CE to on-board MV imaging. Dose-CE correlation was investigated via simulation and experiment. A Monte Carlo (MC) CE simulator was designed using Geant4. Experimental phantoms include: water; tissue-simulating phantom composed of water, Intralipid®, and beef blood; plastic phantom with solid water insert. The detector system comprises an optical fiber and diffraction-grating spectrometer incorporating a front/back-illuminated CCD. The NIR shift was carried out with CdSe/ZnS quantum dots (QDs), emitting at (650±10) nm. CE and MV images were acquired with a CMOS camera and electronic portal imaging device. MC and experimental studies indicate a strong linear dose-CE correlation (Pearson coefficient > 0.99). CE by an 18-MeV beam was effectively NIR-shifted in water and a tissue-simulating phantom, exhibiting a significant increase at 650 nm for QD depths up to 10 mm. CE images exhibited relative contrast superior to MV images by a factor of 30. Our work supports the potential for application of CE in radiotherapy online imaging for patient setup and treatment verification, since CE is intrinsic to the beam and non-ionizing and QDs can be used to improve CE detectability, potentially yielding image quality superior to MV imaging for the case of low-density-variability, low-optical-attenuation materials (ex: breast/oropharynx). Ongoing work involves microenvironment functionalization of QDs and application of multi-channel spectrometry for simultaneous acquisition of dosimetric and tumor oxygenation signals.

SU‐E‐I‐04: Implementation of a Fast Monte Carlo Scatter Correction for Cone‐ Beam Computed Tomography

PURPOSE To improve image quality in cone-beam computed tomography (CBCT) scans by implementation of a fast and accurate MC-based scatter correction algorithm. METHODS A Solid WaterTM phantom was imaged on a Varian OBI CBCT scanner using the standard-dose head protocol (100 kVp, 151 mAs, partial-angle). A fast Monte Carlo simulation developed in the EGSnrc framework was used to transport photons through the uncorrected CBCT scan. From the simulation output, the contribution from both primary and scattered photons for each projection image was estimated. Using these estimates, a subtractive scatter correction was performed on the CBCT projection data. This correction procedure was repeated iteratively, using the previous scatter corrected scan as input to the Monte Carlo simulation. RESULTS Implementation of the scatter correction algorithm on real CBCT data was shown to help mitigate scatter-induced artifacts, such as cupping and streaking. The scatter corrected images were also shown to have improved accuracy in reconstructed attenuation coefficient values. In a region of interest centered on the Solid Water phantom, the number of voxels agreeing to within 10% of the theoretical attenuation coefficient increased from 46% to 97% after two iterations of the scatter correction. CONCLUSIONS These results suggest that the proposed scatter correction algorithm is successful in improving image quality in real CBCT images. The accuracy of the attenuation coefficients extracted from the corrected CBCT scan renders the data suitable for on-the-fly dose recalculations, as well as vastly improved image registration.

Sci—Fri PM: Delivery — 12: Scatter‐B‐Gon: Implementing a fast Monte Carlo cone‐beam computed tomography scatter correction on real data

A fast and accurate MC-based scatter correction algorithm was implemented on real cone-beam computed tomography (CBCT) data. An ACR CT accreditation phantom was imaged on a Varian OBI CBCT scanner using the standard-dose head protocol (100 kVp, 151 mAs, partial-angle). A fast Monte Carlo simulation developed in the EGSnrc framework was used to transport photons through the uncorrected CBCT scan. From the simulation output, the contribution from both primary and scattered photons for each projection image was estimated. Using these estimates, a subtractive scatter correction was performed on the CBCT projection data. Implementation of the scatter correction algorithm on real CBCT data was shown to help mitigate scatter-induced artifacts, such as cupping and streaking. The scatter corrected images were also shown to have improved accuracy in reconstructed attenuation coefficient values. In three regions of interest centered on material inserts in the ACR phantom, the reconstructed CT numbers agreed with clinical CT scan data to within 35 Hounsfield units after scatter correction. These results suggest that the proposed scatter correction algorithm is successful in improving image quality in real CBCT images. The accuracy of the attenuation coefficients extracted from the corrected CBCT scan renders the data suitable for adaptive on the fly dose calculations on individual fractions, as well as vastly improved image registration.

Cherenkov emission dosimetry for electron beam radiotherapy: a Monte Carlo feasibility study of absolute dose prediction

Current electron beam dosimeters face two ma- jor challenges. They are not water/tissue equivalent, and there- fore require conversion to dose to water/tissue. Moreover, they must be placed in the radiation beam, which results in beam perturbation and dose averaging. These challenges limit their spatial resolution for intensity-modulated delivery or in vivo dosimetry. Yet, Cherenkov radiation by high-energy charged particles is emitted in water and in tissue, can be detected outside the beam, and is inherent to all high-energy radiotherapy beams. Despite these advantages, Cherenkov emission has yet to be implemented for clinical dosimetry. The work presented here investigates via first-principles derivation and Monte Carlo simulation the feasibility of absolute Cherenkov dosimetry for electron beam radiotherapy. A quantitative model for predicting absolute dose from Cherenkov intensity in a phantom was derived from first principles for high- energy charged particles of known incident energy with the assumption that all collisional energy loss is absorbed and all radiative energy loss escapes. The model was validated via simulation of 260 keV - 18 MeV electrons incident on water. Monte Carlo simulations were carried out in Geant4. The absolute Cherenkov dosimetry model presented here was able to predict absolute dose from Cherenkov intensity to within a clinically viable uncertainty of less than 3%.

Development of a multi-modality 4D biomechanical phantom for evaluation of simultaneous registration/segmentation algorithms

A package for evaluating joint registration/ segmentation (“regmentation” algorithms) and motion prediction systems was developed using a pair of preserved swine lungs pneumatically controlled with a custom-built respirator. The phantom is MRI, CT and PET compatible and moves in a realistic 4D non-rigid fashion. The segmentation and registration ground truths are provided by a dual compartment mock tumor and a bifurcation tracking pipeline. The mock tumor consists of two vacuum-sealed sea sponges in separate compartments, allowing for the injection of radiotracer to approximate an active tumor and surrounding healthy tissue. Injection is possible through catheters connected to each compartment. The boundary of the inner compartment, segmented post-extraction, provides a ground truth for the boundary of the tumor region. Bifurcations of the bronchi of the lungs were used as anatomical landmarks, providing a registration ground truth between two sets of images of the lungs using in-house bifurcation detection and matching software. The accuracy of the bifurcation tracking pipeline was found to be on the order of a voxel width for human and swine lung datasets, with tracking capabilities up to deformations of 7.8 and 3.4 cm respectively. Accuracy was evaluated using a known virtual deformation. The computer-controlled respirator was found capable of mimicking human breathing traces using the swine lungs within a maximum error of ±2.2% and an average error of ±0.5%. PET/CT and MRI scans of the lungs were acquired for various levels of image noise.