HPC technology optimizes parallel PET reconstruction methods for accurate diagnostic analysis

London 07 September 1998 Medical and computer specialists face a tough job generating high quality images out of positron emission tomography (PET) data. The Parapet project is trying to meet the challenge through the development of a PARAllel PETscan system. Both clinical partners and algorithm designers are united in this Esprit funded initiative to create and apply new advanced parallel algorithms, which have to run on high performance hardware, in order to reduce costs. The enhanced processing rates will produce a superior volume reconstruction quality, very suitable for accurate diagnostics and improved clinical procedures.

Advertisement

Medical and computer specialists face a tough job generating high quality images out of positron emission tomography (PET) data. The Parapet project is trying to meet the challenge through the development of a PARAllel PETscan system. Both clinical partners and algorithm designers are united in this Esprit funded initiative to create and apply new advanced parallel algorithms, which have to run on high performance hardware, in order to reduce costs. The enhanced processing rates will produce a superior volume reconstruction quality, very suitable for accurate diagnostics and improved clinical procedures.

The three year project has just passed its halfway stage and the partners are currently preparing the first results for the end-user forum, to be organized at the Parsytec offices in Aachen during the second week of December. Three institutions are performing the algorithm development and the software parallelisation processing task, which are the Brunel University in London; the Technion Optimization Laboratory at the Facutly of Industrial Engineering and Management of the Israel Institute of Technology in Haifa; and the German Parsytec Company which also provides the powerful CC-12 multiprocessor machine.

Together, these computer experts have concentrated on the creation of iterative techniques, in order to compare analytic methods through the use of simulated and real patient data. Three hospitals with a large experience in PET techniques, namely the English Hammersmith Hospital, the Ospedale San Rafaele in Milan, and the Geneva University Hospital, will evaluate the algorithms via the application of real patient data for the establishment of an effective benchmarking set to provide criteria for objective measurements. In a next step, the six partners will have to decide about the adaptation and implementation of the generated image reconstruction algorithms to real life scanner tomographs by a scanner manufacturer.

As an ultimate goal, the Parapet project partners strive to demonstrate the use of embedded high performance computing technology to perform the detailed reconstruction of PET data in a reasonable timeframe within a clinical environment. The results of Parapet will serve to study physiological, biochemical and pharmaco-kinetic processes in healthy and diseased tissue in vivo. The use of iterative techniques will lead to the reduction in elapsed time needed to produce clinically useful, high quality reconstruction. It is of great technical importance to integrate the high-performance systems with the scanner systems at the hospital sites for an optimum result. The Elgems Company, based in Haifa, will monitor this procedure by closely following the application of algorithms in tomograph reconstruction at an early stage.

The project management, executed by Drs. Francis Wray and David Jeffery of Wray Consultancy in Cambridge, will keep a watchful eye on both the industrial and clinical relevance of Parapet. European industry will benefit from the project in three different ways. First, clinicians as well as the pharmaceutical industry will be highly stimulated to apply new diagnostic techniques and to test a great variety of advanced therapeutic compounds. Second, European scanner manufacturers will be urged to offer competitive systems thanks to the use of embedded HPCN. In third place, the market for developers of embedded supercomputer systems undoubtedly will grow if the project enjoys a successful outcome. For up to date information, we refer to the Parapet Web site.


Leslie Versweyveld

[Medical IT News][Calendar][Virtual Medical Worlds Community][News on Advanced IT]