PhD Showcase II

Date:
Tuesday 16 March 2010

Time:
6.30pm (light refreshments) for 7.00pm

Venue:
The Foresight Centre, University of Liverpool, 1 Brownlow Street, Liverpool, L69 3GL | Map and directions (PDF)
There will be a small charge for parking.

Cost:
Free

Merseyside Branch has arranged for several PhD students at Liverpool University to give a short (10 minutes) description of their research. Four talks have been scheduled. After each talk there will be 5 minutes for questions.

This event is free and open to all.

The talks are:-

1. Ashraf El-Sayed - Processing MRI Brain Scan Images

2. Ross McFarlane - Cardiac Simulations

3. Dan Cartwright - Analysing Public Opinion

4. Robert Piro - Languages for the Semantic Web

This will be an excellent opportunity to hear something about cutting edge research in several areas of IT.

Booking is not essential, but to plan catering please email Jennie McWilliam if you wish to attend.

Ashraf El-Sayed
Region Of Interest Based Image Classification Using Time Series Analysis
 
An approach to Region Of Interest Based Image Classification (ROIBIC), based on a time series analysis approach, is described.  The focus of the approach is the classification of MRI brain scan data according to the nature of the corpus callosum (a feature within such scans), however the approach also has general applicability.  The advocated approach combines a number of image processing techniques combined with time series analysis, specifically dynamic time warping.  Of note is the mechanism used to generate the desired time series.  The application of the time series based ROIBIC demonstrates that the proposed approach performs both efficiently and effectively, obtaining a classification accuracy of over 98% in the case of the given application.  Comparisons are also presented with a graph based ROIBIC approach.

Ross McFarlane
Beatbox - High Performance Computing for Computational Biology of the Heart

Despite over a century's study, the trigger mechanisms of cardiac arrhythmias are poorly understood.  Even modern experimental methods do not provide sufficient temporal and spacial resolution to trace the development of fibrillation in samples of cardiac tissue, not to mention the heart in vivo.  Advances in human genetics provide information on the impact of certain genes on cellular activity, but do not explain the resultant mechanisms by which fibrillation arises.  Thus, for some genetic cardiac diseases, the first presenting symptom is death.

Computer simulations of electrical activity in cardiac tissue offer increasingly detailed insight into these phenomena, providing a view of cellular-level activity on the scale of a whole tissue wall.  Already, advances in this field have led to developments in our understanding of heart fibrillation and sudden cardiac death and their impact is expected to increase significantly as we approach the ultimate goal of whole-heart modelling.

Modelling the propagation of Action Potential through cardiac tissue is computationally expensive due to the huge number of equations per cell and the vast spacial and temporal scales required.  The complexity of the problem encompasses the description of ionic currents underlying excitation of a single cell through the inhomogeneity of the tissue to the complex geometry of the whole heart.  The timely running of computational models of cardiac tissue is increasingly dependant on the effective use of High Performance Computing (HPC), i.e. systems with parallel processors.

In this talk I will present Beatbox - cardiac computer simulation environment.  Beatbox aims to abstract much of the complexity associated with HPC away from users and model developers and  provides an intuitive means of describing complex simulations.  Users with little programming knowledge will be able to configure and run simulations, while proficient software developers will be able to extend Beatbox via robust and flexible interfaces.  I shall discuss the feature set and workflow of Beatbox and some of the challenges faced in its development, in particular its parallel implementation.

Dan Cartwright
Computationally Gathering and Analysing Public Opinion

 
A wide variety of tools exist on the internet for gathering and analysing public opinion.  The tools available typically fall into one of two categories: Tools that are expressive, but do not allow for easy in-depth analysis of responses, and tools that facilitate analysis but do not allow for much expressivity.  My reseach is two fold, firstly focusing on the development of an opinion-gathering tool named Parmenides, which attempts to provide a trade-off between expressiveness and the ability to analyse responses.  Secondly, I am developing and formalising the theories of computational argumentation on which Parmenides is based.

Robert Piro
Understanding Ontology Languages

Formalising ontologies and languages is a recent development in Computer Science.  OWL2, the recent W3C recommendation for web ontology languages, specifies six languages of distinct expressivity and complexity.  While SQL, the the main language for querying relational databases, is well understood, the characterisation of the expressivity of the OWL2 languages poses new and exciting challenges. The aim of this talk is to present a survey of recent progress in this direction.