Keywords: HPCN, volume visualisation, parallel processing, medical imaging, environmental applications, decision support
Start Date: 1 March 95 / Duration: 30 months
[ participants / contact]
The SMART project aims to develop a decision support environment for 3 applications domains: medical imaging, pollution dispersion modelling and remote visualisation for the two former applications. To do this, the proposers and their collaborators will build upon a volume visualisation toolkit. Some of the visualisation algorithms are computationally intensive and parallel versions of the algorithm have already been developed. The proposers are working closely with end-users thus ensuring that the software developed meets their requirements. The aim will be to hide the vagaries of the underlying HPC architecture from the end-user, and, further, to provide an algorithm migration scheme, whereby the computationally intensive part may be executed by other nodes of the network. Applications will be built to help with the modelling of pollution dispersion in the Amazon rain forest, and with visualisation of medical data for remote diagnosis.
The above applications have a common requirement: visualisation & HPC. Those requirements will be provided by the Parallel Volume Visualisation Toolkit (PVV). The PVV toolkit is a basic framework to develop high performance volume visualisation applications based on the recent advances in high performance computing tools such as the MPI (message passing interface) library, visualisation and remote processing.
On September 1996 the PVV 1.5 release will be made available for Beta tests running on a Meiko Computer Surface II distributed parallel computer with 10 Hyperspare 100 Mhz nodes and 384 Mbytes of RAM memory. Since May 1996 the MEIKO CS-2 is installed and available for the project partners. Two applications are being developed: VV and Xslicer. These tools will be available in the second quarter of 1997. VV is a general purpose volume manipulation and rendering tool. It includes tools and techniques for slicing, filtering, segmentation, data classification, compression and rendering of cartesian volumes. Xslicer is a interactive volume navigation tool for clinical purposes. Xslicer includes facilities for real time slicing, local operations such as zooming, segmentation, metrics (distance, area, volume) and 3D volume display.
A WWW site has been established at: http://www.lsi.usp.br/~pvv. Results achieved so far have been published in international journals and presented at international conferences. A workshop and two training weeks are being planned to take place in 1997.
Escola Politecnica da USP, Sao Paulo
Laboratorio de Sistemas
Integraveis
av. Prof. Luciano Gualberto N. 158 Trav. 3
05508-060
Sao Paulo, BRZ
EU Partners
University of Manchester, UK
Non-EU Partners
Escola Politécnica da Universidada de Sao Paulo,
BRZ
National Institute for Space Research (INPE), BRZ
The Children
Institute at Hospital das Clinicas da
Universidade de Sao Paulo,
BRZ
Instituto de Informatica da Universidade
Federal do Rio Grande
do Sul, BRZ
Mr. Marcelo Knorich Zuffo
Tel: +55 11 818 5254
Fax: +55 11 818
5664
E-mail: mkzuffo@lsi.usp.br
SMART - ITDC-225, May 1997
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook