Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Image Segmentation”

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 1,675 records · Page 93

LUVOIR Primary Mirror Segment Alignment Control With Joint Laser Metrology and Segment Edge Sensing

An approach is developed for the alignment and stability maintenance of the LUVOIR segmented primary mirror using a segment state estimation and wavefront control method based on a hybrid segment motion sensing architecture of laser truss metrology and segment edge sensors. Our current computer model was generated for LUVOIR Architecture Option A with a 15-meter aperture, 120segment primary mirror. The methodology and simulation results will be presented and analyzed. JPL has a long history of technology development in laser metrology and edge sensors, including work in SIM [7], Keck and TMT [8], CCAT [3] and LUVOIR [1]. We will discuss our current efforts of LUVOIR laser metrology and edge-sensor models development, showing sensitivities of sensor measurements to various mirror eigenmodes, removing global modes and strengthening weak modes by performing joint (hybrid) laser-metrology and edge sensing. We will define and derive an important performance metric called wavefront error multiplier (WEM), and show that WEM provides a simple link between sensor errors and the closed-loop (controlled) system wavefront error. We will show WEM values for several hybrid sensor configuration options studied. We will discuss an algorithm for mirror shape control and maintenance through segment state and wavefront estimations using joint edge-metrology sensing. We will compare simulated performance of mirror state estimation, wavefront estimation and wavefront control based on joint edge-metrology sensing among several sensor configurations, and show the impact of sensor error distributions on the segmented mirror alignment performance. Mirror shape control performance will be also be evaluated in terms of imaging contrast between inner working angles (IWA) and outer working angles (OWA) of a LUVOIR coronagraph. The research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. LUVOIR is a mission concept under study for the next astronomy Decadal Survey—the information presented in this paper is pre-decisional and is provided for planning and discussion purposes only.

Shelton, Chris↗

Thin Shell, Segmented X-Ray Mirrors

Thin foil mirrors were introduced as a means of achieving high throughput in an X-ray astronomical imaging system in applications for which high angular resolution were not necessary. Since their introduction, their high filling factor, modest mass, relative ease of construction, and modest cost have led to their use in numerous X-ray observatories, including the Broad Band X-ray Telescope, ASCA, and Suzaku. The introduction of key innovations, including epoxy replicated surfaces, multilayer coatings, and glass mirror substrates, has led to performance improvements, and in their becoming widely used for X-ray astronomical imaging at energies above 10 keV. The use of glass substrates has also led to substantial improvement in angular resolution, and thus their incorporation into the NASA concept for the International X-ray Observatory with a planned 3 in diameter aperture. This paper traces the development of foil mirrors from their inception in the 1970's through their current and anticipated future applications.

Petre, Robert↗

Development of 5- and 10-year-old pediatric phantoms based on polygon mesh surfaces

Purpose: The purpose of this study is the development of reference pediatric phantoms for 5- and 10-year-old children to be used for the calculation of organ and tissue equivalent doses in radiation protection. Methods: The study proposes a method for developing anatomically highly sophisticated pediatric phantoms without using medical images. The 5- and 10-year-old male and female phantoms presented here were developed using 3D modeling software applied to anatomical information taken from atlases and textbooks. The method uses polygon mesh surfaces to model body contours, the shape of organs as well as their positions, and orientations in the human body. Organ and tissue masses comply with the corresponding data given by the International Commission on Radiological Protection (ICRP) for the 5- and 10-year-old reference children. Bones were segmented into cortical bone, spongiosa, medullary marrow, and cartilage to allow for the use of micro computer tomographic ({mu}CT) images of trabecular bone for skeletal dosimetry. Results: The four phantoms, a male and a female for each age, and their organs are presented in 3D images and their organ and tissue masses in tables which show the compliance of the ICRP reference values. Dosimetric data, calculated for the reference pediatric phantoms by Monte Carlo methods were compared with corresponding data from adult mesh phantoms and pediatric stylized phantoms. The comparisons show reasonable agreement if the anatomical differences between the phantoms are properly taken into account. Conclusions: Pediatric phantoms were developed without using medical images of patients or volunteers for the first time. The models are reference phantoms, suitable for regulatory dosimetry, however, the 3D modeling method can also be applied to medical images to develop patient-specific phantoms.

61 RADIATION PROTECTION AND DOSIMETRY↗

The CLAS12 Ring Imaging Cherenkov Detector

A ring imaging Cherenkov (RICH) detector has been installed in the CLAS12 spectrometer at Jefferson Laboratory (JLab) to provide kaon identification in the momentum range between 3 GeV/c and 8 GeV/c. The detector adopts a hybrid optics solution with aerogel radiator, light planar and spherical mirrors, and highly segmented photon detectors. We report here on the design, construction, and initial performance of the RICH during the commissioning of the detector and the first physics data taking period

PID detectors↗

Analysis of Texture Using the Fractal Model

Properties such as the fractal dimension (FD) can be used for feature extraction and classification of regions within an image. The FD measures the degree of roughness of a surface, so this number is used to characterize a particular region, in order to differentiate it from another. There are two basic approaches discussed in the literature to measure FD: the blanket method, and the box counting method. Both attempt to measure FD by estimating the change in surface area with respect to the change in resolution. We tested both methods but box counting resulted computationally faster and gave better results. Differential Box Counting (DBC) was used to segment a collage containing three textures. The FD is independent of directionality and brightness so five features were used derived from the original image to account for directionality and gray level biases. FD can not be measured on a point, so we use a window that slides across the image giving values of FD to the pixel on the center of the window. Windowing blurs the boundaries of adjacent classes, so an edge-preserving, feature-smoothing algorithm is used to improve classification within segments and to make the boundaries sharper. Segmentation using DBC was 90.8910 accurate.

Navas, William↗

SU-E-J-129: Atlas Development for Cardiac Automatic Contouring Using Multi-Atlas Segmentation

Purpose: To develop a set of atlases for automatic contouring of cardiac structures to determine heart radiation dose and the associated toxicity. Methods: Six thoracic cancer patients with both contrast and non-contrast CT images were acquired for this study. Eight radiation oncologists manually and independently delineated cardiac contours on the non-contrast CT by referring to the fused contrast CT and following the RTOG 1106 atlas contouring guideline. Fifteen regions of interest (ROIs) were delineated, including heart, four chambers, four coronary arteries, pulmonary artery and vein, inferior and superior vena cava, and ascending and descending aorta. Individual expert contours were fused using the simultaneous truth and performance level estimation (STAPLE) algorithm for each ROI and each patient. The fused contours became atlases for an in-house multi-atlas segmentation. Using leave-one-out test, we generated auto-segmented contours for each ROI and each patient. The auto-segmented contours were compared with the fused contours using the Dice similarity coefficient (DSC) and the mean surface distance (MSD). Results: Inter-observer variability was not obvious for heart, chambers, and aorta but was large for other structures that were not clearly distinguishable on CT image. The average DSC between individual expert contours and the fused contours were less than 50% for coronary arteries and pulmonary vein, and the average MSD were greater than 4.0 mm. The largest MSD of expert contours deviating from the fused contours was 2.5 cm. The mean DSC and MSD of auto-segmented contours were within one standard deviation of expert contouring variability except the right coronary artery. The coronary arteries, vena cava, and pulmonary vein had DSC<70% and MSD>3.0 mm. Conclusion: A set of cardiac atlases was created for cardiac automatic contouring, the accuracy of which was comparable to the variability in expert contouring. However, substantial modification may need for auto-segmented contours of indistinguishable small structures.

60 APPLIED LIFE SCIENCES↗

High-Resolution, Wide-Field-of-View Scanning Telescope

A proposed telescope would afford high resolution over a narrow field of view (<0.10 ) while scanning over a total field of view nominally 16 wide without need to slew the entire massive telescope structure. The telescope design enables resolution of a 1-m-wide object in a 50- km-wide area of the surface of the Earth as part of a 200-km-wide area field of view monitored from an orbit at an altitude of 700 km. The conceptual design of this telescope could also be adapted to other applications both terrestrial and extraterrestrial in which there are requirements for telescopes that afford both wide- and narrow-field capabilities. In the proposed telescope, the scanning would be effected according to a principle similar to that of the Arecibo radio telescope, in which the primary mirror is stationary with respect to the ground and a receiver is moved across the focal surface of the primary mirror. The proposed telescope would comprise (1) a large spherical primary mirror that would afford high resolution over a narrow field of view and (2) a small displaceable optical relay segment that would be pivoted about the center of an aperture stop to effect the required scanning (see figure). Taken together, both comprise a scanning narrow-angle telescope that does not require slewing the telescope structure. In normal operation, the massive telescope structure would stare at a fixed location on the ground. The inner moveable relay optic would be pivoted to scan the narrower field of view over the wider one, making it possible to retain a fixed telescope orientation, while obtaining high-resolution images over multiple target areas during an interval of 3 to 4 minutes in the intended orbit. The pivoting relay segment of the narrow-angle telescope would include refractive and reflective optical elements, including two aspherical mirrors, to counteract the spherical aberration of the primary mirror. Overall, the combination of the primary mirror and the smaller relay optic would provide narrow-angle, diffraction-limited high resolution at a wavelength of 500 nm.

Sepulveda, Cesar↗

LABQ3: Bayesian method for quantification of mineral compositions and nano-scale elemental mapping of 3D synchrotron XCT data

Quantitative analysis of mineral compositions is essential in understanding geochemical, mineralogical and environmental processes. Fine-resolution 3D imaging is widely done using synchrotron X-ray computed tomography (XCT), but existing analyses are limited to visualization and segmentation. This paper presents a new method, Linear Attenuation Bayesian Quantitative 3D-mapper (LABQ3), based on the linearity of X-ray attenuation with respect to elemental concentrations. To address the random variability in attenuation measurements, LABQ3 employs Bayesian decision theory to minimize classification error, using reference attenuation distributions from scans of pure mineral standards. To demonstrate LABQ3 and test its performance, we studied precipitated carbonate samples. XCT scans were done at multiple energies using the transmission X-ray microscope (TXM) at beamline 32-ID-C of the Advanced Photon Source at Argonne National Laboratory. The reconstructed 3D images have a voxel size of 20 nm. Analyses revealed rich nano-scale compositional heterogeneity within individual particles. A mixture of calcium and cadmium produced an overall stoichiometric composition of (Ca 0.78 ,Cd 0.22 )CO 3 , with some voxels containing nearly pure CdCO 3 . The addition of zinc led to an overall stoichiometric composition of 33% Ca, 28% Cd, 39% Zn, with a nearly pure CaCO 3 core and compositional zonation through the rim. These compositional gradients are related to temporal sequences of carbonate mineral formation where Cd precipitated at the beginning in (Ca,Cd)CO 3 , while Cd and Zn precipitated at the end in (Ca, Cd,Zn)CO 3 . Results differ from bulk analyses using Inductively Coupled Plasma-Mass Spectrometry (ICP-MS), showing that LABQ3 provides particle-specific insights. LABQ3 distinguishes itself by quantifying chemical compositions along a continuum, making it different from XCT analyses based on segmentation. LABQ3 allows simultaneous acquisition of morphology and chemical composition in 3D, facilitating the interpretation of chemical gradients of trace elements, quantification of solid solution compositions, inferences about temporal sequences of mineral precipitation, and addressing other concerns about solid-phase chemistry.

58 GEOSCIENCES↗

Clinical Evaluation of Soft Tissue Organ Boundary Visualization on Cone-Beam Computed Tomographic Imaging

Purpose: Cone-beam computed tomographic images (CBCTs) are increasingly used for setup correction, soft tissue targeting, and image-guided adaptive radiotherapy. However, CBCT image quality is limited by low contrast and imaging artifacts. This analysis investigates the detectability of soft tissue boundaries in CBCT by performing a multiple-observer segmentation study. Methods and Materials: In four prostate cancer patients prostate, bladder and rectum were repeatedly delineated by five observers on CBCTs and fan-beam CTs (FBCTs). A volumetric analysis of contouring variations was performed by calculating coefficients of variation (COV: standard deviation/average volume). The topographical distribution of contouring variations was analyzed using an average surface mesh-based method. Results: Observer- and patient-averaged COVs for FBCT/CBCT were 0.09/0.19 for prostate, 0.05/0.08 for bladder, and 0.09/0.08 for rectum. Contouring variations on FBCT were significantly smaller than on CBCT for prostate (p < 0.03) and bladder (p < 0.04), but not for rectum (p < 0.37; intermodality differences). Intraobserver variations from repeated contouring of the same image set were not significant for either FBCT or CBCT (p < 0.05). Average standard deviations of individual observers' contour differences from average surface meshes on FBCT vs. CBCT were 1.5 vs. 2.1 mm for prostate, 0.7 vs. 1.4 mm for bladder, and 1.3 vs. 1.5 mm for rectum. The topographical distribution of contouring variations was similar for FBCT and CBCT. Conclusion: Contouring variations were larger on CBCT than FBCT, except for rectum. Given the well-documented uncertainty in soft tissue contouring in the pelvis, improvement of CBCT image quality and establishment of well-defined soft tissue identification rules are desirable for image-guided radiotherapy.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

High-Performance Scanning Acousto-Ultrasonic System

A high-performance scanning acousto-ultrasonic system, now undergoing development, is designed to afford enhanced capabilities for imaging microstructural features, including flaws, inside plate specimens of materials. The system is expected to be especially helpful in analyzing defects that contribute to failures in polymer- and ceramic-matrix composite materials, which are difficult to characterize by conventional scanning ultrasonic techniques and other conventional nondestructive testing techniques. Selected aspects of the acousto-ultrasonic method have been described in several NASA Tech Briefs articles in recent years. Summarizing briefly: The acousto-ultrasonic method involves the use of an apparatus like the one depicted in the figure (or an apparatus of similar functionality). Pulses are excited at one location on a surface of a plate specimen by use of a broadband transmitting ultrasonic transducer. The stress waves associated with these pulses propagate along the specimen to a receiving transducer at a different location on the same surface. Along the way, the stress waves interact with the microstructure and flaws present between the transducers. The received signal is analyzed to evaluate the microstructure and flaws. The specific variant of the acousto-ultrasonic method implemented in the present developmental system goes beyond the basic principle described above to include the following major additional features: Computer-controlled motorized translation stages are used to automatically position the transducers at specified locations. Scanning is performed in the sense that the measurement, data-acquisition, and data-analysis processes are repeated at different specified transducer locations in an array that spans the specimen surface (or a specified portion of the surface). A pneumatic actuator with a load cell is used to apply a controlled contact force. In analyzing the measurement data for each pair of transducer locations in the scan, the total (multimode) acousto-ultrasonic response of the specimen is utilized. The analysis is performed by custom software that extracts parameters of signals in the time and frequency domains. The computer hardware and software provide both real-time and postscan processing and display options. For example, oscilloscope displays of waveforms and power spectral densities are available in real time. Images can be computed while scanning continues. Signals can be digitally preprocessed and/or post-processed by filtering, windowing, time-segmenting, and running-waveform-averaging algorithms. In addition, the software affords options for off-line simulation of the waveform-data-acquisition and scanning processes. In tests, the system has been shown to be capable of characterizing microstructural changes and defects in SiC/SiC and C/SiC ceramic-matrix composites. Delaminations, variations in density, microstructural changes attributable to infiltration by silicon, and crack-space indications (defined in the next sentence) have been revealed in images formed from several time- and frequency-domain parameters of scanning acousto-ultrasonic signals. The crack-space indications were image features that were not revealed by other nondestructive testing methods and are so named because they turned out to mark locations where cracking eventually occurred.

Roth, Don↗

A Study of Pseudoprogression After Spine Stereotactic Body Radiation Therapy

Purpose: To determine the incidence of pseudoprogression (PP) after spine stereotactic body radiation therapy based on a detailed and quantitative assessment of magnetic resonance imaging (MRI) morphologic tumor alterations, and to identify predictive factors distinguishing PP from local recurrence (LR). Methods and Materials: A retrospective analysis of 35 patients with 49 spinal segments treated with spine stereotactic body radiation therapy, from 2009 to 2014, was conducted. The median number of follow-up MRI studies was 4 (range, 2-7). The gross tumor volumes (GTVs) within each of the 49 spinal segments were contoured on the pretreatment and each subsequent follow-up T1- and T2-weighted MRI sagittal sequence. T2 signal intensity was reported as the mean intensity of voxels constituting each volume. LR was defined as persistent GTV enlargement on ≥2 serial MRI studies for ≥6 months or on pathologic confirmation. PP was defined as a GTV enlargement followed by stability or regression on subsequent imaging within 6 months. Kaplan-Meier analysis was used for estimation of actuarial local control, disease-free survival, and overall survival. Results: The median follow-up was 23 months (range, 1-39 months). PP was identified in 18% of treated segments (9 of 49) and LR in 29% (14 of 49). Earlier volume enlargement (5 months for PP vs 15 months for LR, P=.005), greater GTV to reference nonirradiated vertebral body T2 intensity ratio (+30% for PP vs −10% for LR, P=.005), and growth confined to 80% of the prescription isodose line (80% IDL) (8 of 9 PP cases vs 1 of 14 LR cases, P=.002) were associated with PP on univariate analysis. Multivariate analysis confirmed an earlier time to volume enlargement and growth within the 80% IDL as significant predictors of PP. LR involved the epidural space in all but 1 lesion, whereas PP was confined to the vertebral body in 7 of 9 cases. Conclusions: PP was observed in 18% of treated spinal segments. Tumor growth confined to the 80% IDL and earlier time to tumor enlargement were predictive for PP.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Deep Neural Network Algorithm for CMC Microstructure Characterization and Variability Quantification

Microstructure characterization and variability quantification are crucial for understanding ceramic matrix composites (CMCs) mechanical behavior and deformation mechanisms across length scales. Traditionally, analyses of the micrographs obtained from microscopy are labor-intensive. However, with the vast improvement in computer vision (CV) and deep learning (DL), an automated algorithm can be designed to extract essential microstructure variability from micrographs which can then be used to construct a statistically representative volume element (SRVE). The DL-based algorithm spans the taxonomy of microstructure analyses, including semantic segmentation of microstructure constituents, secondary phases, matrix/fiber interface, and defects, and quantifying the microstructure variability in terms of probability distributions. In this work, C/SiNC and SiC/SiNC CMCs microstructures are semantically segmented through a deep convolutional neural network, followed by variability quantification through the implementation of a fully connected regression layer, hence forming a deep regression network. The deep regression network operates in a feedforward regime, in which the neuron output signal traverses through the network in a unidirectional manner. The weight tensor associated with each layer is updated through a backpropagation stochastic gradient descent approach. The input gray-scale image obtained through in-house scanning electron microscope and confocal microscope micrographs is augmented through affine transformations to increase the training set size, which is then processed through four strided convolutional layers. This compresses the image resolution by half at each layer while increasing the image depth by applying different filters (image encoding). The class activation maps (CAMs) corresponding to the applied filters highlight the key architectural features and assist with the semantic segmentation of the microstructure.

Hamza, Mohamed H.↗

SU-G-JeP2-01: A New Approach for MR-Only Treatment Planning: Tissue Segmentation-Based Pseudo-CT Generation Using T1-Weighted MRI

Purpose: To evaluate MR-only treatment planning for brain Stereotactic Ablative Radiotherapy (SABR) based on pseudo-CT (pCT) generation using one set of T1-weighted MRI. Methods: T1-weighted MR and CT images from 12 patients who were eligible for brain SABR were retrospectively acquired for this study. MR-based pCT was generated by using a newly in-house developed algorithm based on MR tissue segmentation and voxel-based electron density (ED) assignment (pCTv). pCTs using bulk density assignment (pCTb where bone and soft tissue were assigned 800HU and 0HU,respectively), and water density assignment (pCTw where all tissues were assigned 0HU) were generated for comparison of ED assignment techniques. The pCTs were registered with CTs and contours of radiation targets and Organs-at-Risk (OARs) from clinical CT-based plans were copied to co-registered pCTs. Volumetric-Modulated-Arc-Therapy(VMAT) plans were independently created for pCTv and CT using the same optimization settings and a prescription (50Gy/10 fractions) to planning-target-volume (PTV) mean dose. pCTv-based plans and CT-based plans were compared with dosimetry parameters and monitor units (MUs). Beam fluence maps of CT-based plans were transferred to co-registered pCTs, and dose was recalculated on pCTs. Dose distribution agreement between pCTs and CT plans were quantified using Gamma analysis (2%/2mm, 1%/1mm with a 10% cut-off threshold) in axial, coronal and sagittal planes across PTV. Results: The average differences of PTV mean and maximum doses, and monitor units between independently created pCTv-based and CT-based plans were 0.5%, 1.5% and 1.1%, respectively. Gamma analysis of dose distributions of the pCTs and the CT calculated using the same fluence map resulted in average agreements of 92.6%/79.1%/52.6% with 1%/1mm criterion, and 98.7%/97.4%/71.5% with 2%/2mm criterion, for pCTv/CT, pCTb/CT and pCTw/CT, respectively. Conclusion: Plans produced on Voxel-based pCT is dosimetrically more similar to CT plans than bulk assignment-based pCTs. MR-only treatment planning using voxel-based pCT generated from T1-wieghted MRI may be feasible.

60 APPLIED LIFE SCIENCES↗

Scattering and the Point Spread Function of the New Generation Space Telescope

Preliminary design work on the New Generation Space Telescope (NGST) is currently under way. This telescope is envisioned as a lightweight, deployable Cassegrain reflector with an aperture of 8 meters, and an effective focal length of 80 meters. It is to be folded into a small-diameter package for launch by an Atlas booster, and unfolded in orbit. The primary is to consist of an octagon with a hole at the center, and with eight segments arranged in a flower petal configuration about the octagon. The comers of the petal-shaped segments are to be trimmed so that the package will fit atop the Atlas booster. This mirror, along with its secondary will focus the light from a point source into an image which is spread from a point by diffraction effects, figure errors, and scattering of light from the surface. The distribution of light in the image of a point source is called a point spread function (PSF). The obstruction of the incident light by the secondary mirror and its support structure, the trimmed corners of the petals, and the grooves between the segments all cause the diffraction pattern characterizing an ideal point spread function to be changed, with the trimmed comers causing the rings of the Airy pattern to become broken up, and the linear grooves causing diffraction spikes running radially away from the central spot, or Airy disk. Any figure errors the mirror segments may have, or any errors in aligning the petals with the central octagon will also spread the light out from the ideal point spread function. A point spread function for a mirror the size of the NGST and having an incident wavelength of 900 nm is considered. Most of the light is confined in a circle with a diameter of 0.05 arc seconds. The ring pattern ranges in intensity from 10(exp -2) near the center to 10(exp -6) near the edge of the plotted field, and can be clearly discerned in a log plot of the intensity. The total fraction of the light scattered from this point spread function is called the total integrated scattering (TIS), and the fraction remaining is called the Strehl ratio. The angular distribution of the scattered light is called the angle resolved scattering (ARS), and it shows a strong spike centered on a scattering angle of zero, and a broad , less intense distribution at larger angles. It is this scattered light, and its effect on the point spread function which is the focus of this study.

Schreur, Julian J.↗

TU-G-303-00: Radiomics: Advances in the Use of Quantitative Imaging Used for Predictive Modeling

‘Radiomics’ refers to studies that extract a large amount of quantitative information from medical imaging studies as a basis for characterizing a specific aspect of patient health. Radiomics models can be built to address a wide range of outcome predictions, clinical decisions, basic cancer biology, etc. For example, radiomics models can be built to predict the aggressiveness of an imaged cancer, cancer gene expression characteristics (radiogenomics), radiation therapy treatment response, etc. Technically, radiomics brings together quantitative imaging, computer vision/image processing, and machine learning. In this symposium, speakers will discuss approaches to radiomics investigations, including: longitudinal radiomics, radiomics combined with other biomarkers (‘pan-omics’), radiomics for various imaging modalities (CT, MRI, and PET), and the use of registered multi-modality imaging datasets as a basis for radiomics. There are many challenges to the eventual use of radiomics-derived methods in clinical practice, including: standardization and robustness of selected metrics, accruing the data required, building and validating the resulting models, registering longitudinal data that often involve significant patient changes, reliable automated cancer segmentation tools, etc. Despite the hurdles, results achieved so far indicate the tremendous potential of this general approach to quantifying and using data from medical images. Specific applications of radiomics to be presented in this symposium will include: the longitudinal analysis of patients with low-grade gliomas; automatic detection and assessment of patients with metastatic bone lesions; image-based monitoring of patients with growing lymph nodes; predicting radiotherapy outcomes using multi-modality radiomics; and studies relating radiomics with genomics in lung cancer and glioblastoma. Learning Objectives: Understanding the basic image features that are often used in radiomic models. Understanding requirements for reliable radiomic models, including robustness of metrics, adequate predictive accuracy, and generalizability. Understanding the methodology behind radiomic-genomic (’radiogenomics’) correlations. Research supported by NIH (US), CIHR (Canada), and NSERC (Canada)

60 APPLIED LIFE SCIENCES↗

TU-G-303-01: Radiomics: Quantitative Imaging in the Service of Improved Treatment Decision Making

‘Radiomics’ refers to studies that extract a large amount of quantitative information from medical imaging studies as a basis for characterizing a specific aspect of patient health. Radiomics models can be built to address a wide range of outcome predictions, clinical decisions, basic cancer biology, etc. For example, radiomics models can be built to predict the aggressiveness of an imaged cancer, cancer gene expression characteristics (radiogenomics), radiation therapy treatment response, etc. Technically, radiomics brings together quantitative imaging, computer vision/image processing, and machine learning. In this symposium, speakers will discuss approaches to radiomics investigations, including: longitudinal radiomics, radiomics combined with other biomarkers (‘pan-omics’), radiomics for various imaging modalities (CT, MRI, and PET), and the use of registered multi-modality imaging datasets as a basis for radiomics. There are many challenges to the eventual use of radiomics-derived methods in clinical practice, including: standardization and robustness of selected metrics, accruing the data required, building and validating the resulting models, registering longitudinal data that often involve significant patient changes, reliable automated cancer segmentation tools, etc. Despite the hurdles, results achieved so far indicate the tremendous potential of this general approach to quantifying and using data from medical images. Specific applications of radiomics to be presented in this symposium will include: the longitudinal analysis of patients with low-grade gliomas; automatic detection and assessment of patients with metastatic bone lesions; image-based monitoring of patients with growing lymph nodes; predicting radiotherapy outcomes using multi-modality radiomics; and studies relating radiomics with genomics in lung cancer and glioblastoma. Learning Objectives: Understanding the basic image features that are often used in radiomic models. Understanding requirements for reliable radiomic models, including robustness of metrics, adequate predictive accuracy, and generalizability. Understanding the methodology behind radiomic-genomic (’radiogenomics’) correlations. Research supported by NIH (US), CIHR (Canada), and NSERC (Canada)

60 APPLIED LIFE SCIENCES↗

TU-G-303-02: Robust Radiomics Methods for PET and CT Imaging

‘Radiomics’ refers to studies that extract a large amount of quantitative information from medical imaging studies as a basis for characterizing a specific aspect of patient health. Radiomics models can be built to address a wide range of outcome predictions, clinical decisions, basic cancer biology, etc. For example, radiomics models can be built to predict the aggressiveness of an imaged cancer, cancer gene expression characteristics (radiogenomics), radiation therapy treatment response, etc. Technically, radiomics brings together quantitative imaging, computer vision/image processing, and machine learning. In this symposium, speakers will discuss approaches to radiomics investigations, including: longitudinal radiomics, radiomics combined with other biomarkers (‘pan-omics’), radiomics for various imaging modalities (CT, MRI, and PET), and the use of registered multi-modality imaging datasets as a basis for radiomics. There are many challenges to the eventual use of radiomics-derived methods in clinical practice, including: standardization and robustness of selected metrics, accruing the data required, building and validating the resulting models, registering longitudinal data that often involve significant patient changes, reliable automated cancer segmentation tools, etc. Despite the hurdles, results achieved so far indicate the tremendous potential of this general approach to quantifying and using data from medical images. Specific applications of radiomics to be presented in this symposium will include: the longitudinal analysis of patients with low-grade gliomas; automatic detection and assessment of patients with metastatic bone lesions; image-based monitoring of patients with growing lymph nodes; predicting radiotherapy outcomes using multi-modality radiomics; and studies relating radiomics with genomics in lung cancer and glioblastoma. Learning Objectives: Understanding the basic image features that are often used in radiomic models. Understanding requirements for reliable radiomic models, including robustness of metrics, adequate predictive accuracy, and generalizability. Understanding the methodology behind radiomic-genomic (’radiogenomics’) correlations. Research supported by NIH (US), CIHR (Canada), and NSERC (Canada)

60 APPLIED LIFE SCIENCES↗

TU-G-303-03: Machine Learning to Improve Human Learning From Longitudinal Image Sets

‘Radiomics’ refers to studies that extract a large amount of quantitative information from medical imaging studies as a basis for characterizing a specific aspect of patient health. Radiomics models can be built to address a wide range of outcome predictions, clinical decisions, basic cancer biology, etc. For example, radiomics models can be built to predict the aggressiveness of an imaged cancer, cancer gene expression characteristics (radiogenomics), radiation therapy treatment response, etc. Technically, radiomics brings together quantitative imaging, computer vision/image processing, and machine learning. In this symposium, speakers will discuss approaches to radiomics investigations, including: longitudinal radiomics, radiomics combined with other biomarkers (‘pan-omics’), radiomics for various imaging modalities (CT, MRI, and PET), and the use of registered multi-modality imaging datasets as a basis for radiomics. There are many challenges to the eventual use of radiomics-derived methods in clinical practice, including: standardization and robustness of selected metrics, accruing the data required, building and validating the resulting models, registering longitudinal data that often involve significant patient changes, reliable automated cancer segmentation tools, etc. Despite the hurdles, results achieved so far indicate the tremendous potential of this general approach to quantifying and using data from medical images. Specific applications of radiomics to be presented in this symposium will include: the longitudinal analysis of patients with low-grade gliomas; automatic detection and assessment of patients with metastatic bone lesions; image-based monitoring of patients with growing lymph nodes; predicting radiotherapy outcomes using multi-modality radiomics; and studies relating radiomics with genomics in lung cancer and glioblastoma. Learning Objectives: Understanding the basic image features that are often used in radiomic models. Understanding requirements for reliable radiomic models, including robustness of metrics, adequate predictive accuracy, and generalizability. Understanding the methodology behind radiomic-genomic (’radiogenomics’) correlations. Research supported by NIH (US), CIHR (Canada), and NSERC (Canada)

60 APPLIED LIFE SCIENCES↗