HPCS 2012 Paper Presentations – Applications


Dr James Chow, Radiation Medicine Program, Princess Margaret Hospital, University Health Network, and Department of Radiation Oncology, University of Toronto, ON, Canada

Dosimetric corrections for a 4D–computed tomography dataset using the Free-Form deformation algorithm
A Free-Form Deformable (FFD) image registration algorithm in conjunction with 4D Computed Tomography (CT) images was implemented within a graphical user interface (GUI) for dosimetric calculations. The algorithm was developed using the cubic-B-spline method with smoothness corrections and registration point assistance to mark fudicials. A GUI called Free Form Deformation Four Dimensional (FFD4D) was created using GUIDE in MATLAB to encapsulate the deformation algorithm. Validation of the FFD algorithm was performed using the QUASAR Respiratory Motion Phantom. The phantom’s motion had an amplitude of 4 cm in the superior-inferior direction mimicking lung motion due to breathing. 4D-CT image sets at 10 phases were acquired. Using FFD, nine phases of motion were registered to the 30% phase and evaluated. The average geometric differences for each phase were in the range of 1.7 – 4.3 mm with an average error of 3.2 mm. The software has also been demonstrated using 4D-CT images of a lung cancer patient.

Peter Tieleman, Professor, Department of Biological Sciences and Institute for Biocomplexity and Informatics, University of Calgary

Computer Simulations of Large-Scale Membrane Transformations

In molecular dynamics simulations powerful computers follow the motions of a collection of atoms over time to create a movie that gives information about the dynamics and thermodynamics of the system that is simulated. We are interested in applying such simulations to models of biological cell membranes, the important walls that separate the inside of a cell from the outside. Certain proteins cause bending of cell membranes, which is a key step in several processes including the release of nearly mature Influenza viruses from the host cell that produced them. The length and time scale of this budding process and the complex mixture of molecules involved complicate simulations but there are few experimental alternatives. I will discuss progress in simulations at this scale and challenges we are encountering in analyzing the large amount of data we obtain from our simulations.

Soltan Alharbi, PhD Candidate, ECE Department, University of Victoria

High Performance Proactive Digital Forensics
In this presentation, we will illustrate the importance of High Performance Computing in Digital Forensic (DF) investigation, especially in the next generation DF tools. We will then discuss a new high performance automated proactive digital forensic system. The main components of the system are described from a design and an implementation perspectives. The most expensive component, namely proactive analysis, implements a new parallel information-based outlier analysis to proactively and forensically handle suspicious activities. To analyze a large number of targets and events and continuously do so (to capture the dynamics of the system), we rely on a multi-resolution approach to explore the digital forensic space and carry out analysis only when and where needed.


Individual Presentations

Intended Audience 

General: Faculty, staff, students

Session Duration

60 minutes

Session Schedule

Wednesday, May 2nd
3:00 – 4:00 p.m.
Canfor Policy Room