Derek Gerstmann

In vivo three-dimensional optical coherence elastography

Abstract: We present the first three-dimensional (3D) data sets recorded using optical coherence elastography (OCE). Uni-axial strain rate was measured on human skin in vivo using a spectral-domain optical coherence tomography (OCT) system providing >450 times higher line rate than previously reported for in vivo OCE imaging. Mechanical excitation was applied at a frequency of 125 Hz using a ring actuator sample arm with, for the first time in OCE measurements, a controlled static preload. We performed 3D-OCE, processed in 2D and displayed in 3D, on normal and hydrated skin and observed a more elastic response of the stratum corneum in the hydrated case.

Predictions for ASKAP neutral hydrogen surveys

The Australian Square Kilometre Array Pathfinder (ASKAP) will revolutionize our knowledge of gas-rich galaxies in the Universe. Here we present predictions for two proposed extragalactic ASKAP neutral hydrogen (H I) emission-line surveys, based on semi-analytic models applied to cosmological N-body simulations. The ASKAP H I All-Sky Survey, known as Widefield ASKAP L-band Legacy All-sky Blind surveY (WALLABY), is a shallow 3π survey (z = 0–0.26) which will probe the mass and dynamics of over 6 × 10 5 galaxies. A much deeper small-area H I survey, called Deep Investigation of Neutral Gas Origins (DINGO), aims to trace the evolution of H I from z = 0 to 0.43, a cosmological volume of 4 × 10 7 Mpc 3 , detecting potentially 10 5 galaxies. The high-sensitivity 30 antenna ASKAP core (diameter ∼2 km) will provide an angular resolution of 30 arcsec (at z = 0). Our simulations show that the majority of galaxies detected in WALLABY (87.5 per cent) will be resolved. About 5000 galaxies will be well resolved, i.e. more than five beams (2.5 arcmin) across the major axis, enabling kinematic studies of their gaseous discs. This number would rise to 1.6 × 10 5 galaxies if all 36 ASKAP antennas could be used; the additional six antennas provide baselines up to 6 km, resulting in an angular resolution of 10 arcsec. For DINGO this increased resolution is highly desirable to minimize source confusion, reducing confusion rates from a maximum of 10 per cent of sources at the survey edge to 3 per cent. We estimate that the sources detected by WALLABY and DINGO will span four orders of magnitude in total halo mass (from 10 11 to 10 15 M� ) and nearly seven orders of magnitude in stellar mass (from 10 5 to 10 12 M� ), allowing us to

A performance-oriented data parallel virtual machine for GPUs

Existing GPU programming interfaces require applications to adopt a graphics-centric programming model exported by a device driver tuned for real-time graphics and games. This programming model, however, hinders the development and performance of non-graphics applications by imposing a graphics policy for program execution and hiding hardware resources. We present a new virtual machine abstraction for GPUs that provides policy-free, low-level access to the hardware and is designed for high-performance, data-parallel applications.

Imaging true 3D endoscopic anatomy by incorporating magnetic tracking with optical coherence tomography: proof-of-principle for airways

Endoscopic imaging using optical coherence tomography (OCT) has been demonstrated as clinically useful in the assessment of human airways. These airways have a complex 3D structure, bending, tapering and bifurcating. Previously published 3D OCT reconstructions have not accounted for changes in the orientation and trajectory of the endoscopic probe as it moves through the airway during imaging. We propose a novel endoscopic setup incorporating a magnetic tracking system that accounts for these changes, yielding reconstructions that reveal the true 3D nature of the imaged anatomy. We characterize the accuracy of the system, and present the first published magnetic tracker-assisted endoscopic OCT reconstructions using a phantom airway.

Integrating HPC into Radio-Astronomical data reduction

The construction of the SKA and its pathfinders puts significant computational and power consumption constraints on the design of the computer hardware. However, hardware is only one aspect of the problem space, even if extremely fast and power efficient computers were available today, we still could not process the data coming from the SKA due to its volume and the time critical nature of the results. This talk outlines the challenges in the areas of: data flow; storage design and optimisation; database integration into HPC; and low-latency scheduling for ultrascale visualisation and analysis on HPC systems.

A practical visualization strategy for large-scale supernovae CFD simulations

Simulating the expansion of a Type II supernova using an adaptive computational fluid dynamics (CFD) engine yields a complex mixture of turbulent flow with dozens of physical properties. The dataset shown in this sketch was initially simulated on iVECs EPIC supercomputer (a 9600 core Linux cluster) using FLASH [Fryxell et al. 2000] to model the thermonuclear explosion, and later post-processed using a novel integration technique to derive the radio frequency emission spectra of the expanding shock-wave front [Potter et al. 2011]. Model parameters have been chosen to simulate the asymmetric properties of the SN 1987A remnant [Potter et al. 2009].

Physical and graphical effects in OpenCL by example: (Copyright restrictions prevent ACM from providing the full text for this article)

There are strong indications that the future of interactive graphics involves a more flexible programming model than todays OpenGL/Direct3D pipelines. That means that graphics developers will need a basic understanding of how to combine emerging parallel-programming techniques with the traditional interactive rendering pipeline. This course provides an introduction to parallel-programming architectures and environments for interactive graphics, and demonstrates how to combine traditional rendering APIs with advanced parallel computation. It presents several studies of how developers combine traditional rendering API techniques with advanced parallel computation. Each case study includes a live demo and discusses the mix of parallel-programming constructs used, details of the graphics algorithm, and how the rendering pipeline and computation interact to achieve the technical goals. All computation is done in OpenCL. A combination of DirectX and OpenGL is used for the rendering.