Docstoc

Application of multiprocessor and GRID technology in medical image processing

Document Sample
Application of multiprocessor and GRID technology in medical image processing Powered By Docstoc
					Application of multiprocessor and GRID technology in medical image processing
IKTA5-00153/2002

1

Consortium
University of Debrecen, Medical and Health Science Center, PET Center, Debrecen
Home page:< http://www.pet.dote.hu>

Service Center for Informatics of University of Debrecen, Debrecen
Home page: <http://www .diszk.unideb.hu>

Institute of Nuclear Research of the Hungarian Academy of Sciences, Debrecen
Home page: <http://wwwpet.atomki.hu>

MEDISO Medical Imaging System Ltd., Budapest
Home page: <http://www.mediso.hu>

2

Phase 1
Consideration & specification of tasks
1) review of recent proceedings in the field of the project :
• study present medical image processing and reconstruction algorythms, select the best for parallelisation, collect the most up-to-date methods in the field of parallel data acquisition, review of standard GRID implementations (eg. Globus), pick the one most suitable for the project, contact other GRID projects.

• • •

2) hardware specification, cluster implementation

3

Phase 2
UPS

Cluster building, planning
switch RAID
file server
node 0

LAN

node 1
node 2

• • • • • •

UPS switch 100 MBit /1GBit IDE-RAID 200-600 GByte fileserver node node 4 – front-end machine
DAQ

node N-1 node N

4

Phase 3

Implementation I.
• creation of a C++ based volumetric software library in parallel environment • USB, firewire,DIO or Gigabit Ethernet based high speed data acquisition nodes in the cluster • cluster integration (logical, physical)

5

Phase 3
Internet

Integration of clusters

Cluster A

metacluster
Cluster B

6

Phase 4

Implementation II.

• optimisation of the parallel volumetric software library using the developed test programs • testing data acquisition nodes, implementing parallel event processing algorithms • cluster integration, GRID implementation

7

Phase 5

GRID integration & optimisation

• running visualization and other image processing algorithms in cluster and meta-cluster environments, optimising visualization, • optimising data acquisition programs • GRID integration, forming a Virtual Organization, the system can appear in the Hungarian GRID

8

Phase 6

complex integration, quality monitoring

running the programs in environments with different resources (dedicated cluster, cluster with dynamically changing number of nodes, meta-cluster, GRID) Using clusters in commercial medical image processing facilities?

9

Data acquisition – Hardware Specification
Gigabit Ethernet

cluster
switch

detector system

PC1 : data acquisition

Gigabit Ethernet, Myrinet

PC2 : data acquisition
PC3 : calculation

USB?

PCN : calculation
1 GBit

LAN
100 MBit

FRONT-END PC

Gigabit switch

fileserver (1TB) UPS

10

Meta-cluster based data acquisition
detector cluster switch data acquisition cluster
PC1 : data acquisition

detector system

Gigabit Ethernet, Myrinet

Fast Ethernet Gigabit Ethernet

PC2 : data acquisition
PC3 : calculation

USB?

PCN : calculation
1 GBit

LAN
100 MBit

FRONT-END PC

Gigabit switch

fileserver (1TB) UPS

11

Meta-cluster based data acquisition
detector cluster switch

Linux data acquisition cluster
PC1 : data acquisition

detector system

Gigabit Ethernet, Myrinet

Fast Ethernet Gigabit Ethernet

PC2 : data acquisition
PC3 : calculation

Linux
USB?

PCN : calculation
1 GBit

LAN

100 MBit

FRONT-END PC

Gigabit switch

fileserver (1TB) UPS

Linux or/and Windows

12

Software components
• Data acquisition
LM data collection and sorting, sinogram generation

• Hardware control
Electronics controlling and programming

• Reconstruction
reconstruction and correction

• Quality control
utilities for hardware and detector developers and for operators

• Image processing
visualization, documentation, registration, fusion, tracer kinetic modelling, ...

13

Software Development Environment
complex development environment

GUI

Visualization Library

prog 1

prog2

Demo3

Volumetric Library

Image, LM-data, sinogram database

14

Software Development Environment
complex development environment

single processor

GUI

Visualization Library

prog 1

prog2

Demo3

Volumetric Library

Image, LM-data, sinogram database

15

Software Development Environment
complex development environment

single processor

GUI

multi thread

Visualization Library

prog 1

prog2

Demo3

Volumetric Library

Image, LM-data, sinogram database

16

Software Development Environment
complex development environment

single processor

GUI

multi thread
parallelised

Visualization Library

prog 1

prog2

Demo3

Volumetric Library

Image, LM-data, sinogram database

17

Software libraries
GUI Qt 3.x: Tools for professional application development on
MS Windows, Linux, Unix and Mac. (www.trolltech.com)

Commersional: Qt/X11 (linux, unix) Non-commersional: Qt/Windows, Qt/Mac

Visualization Library MDL : OpenGL based structured visualization library.

binaries and headers are available for partners

Volumetric Library VPP: C++ based data modeling and algorithm library.

Full source and documentation will be free under GPL.

18

Other applications
basic- and applied research

brain research:
analysis of functional connections
segmentation image warping techniques using perfusion examinations

detector design:

Monte Carlo simulation

19


				
DOCUMENT INFO
Shared By:
Stats:
views:115
posted:4/18/2008
language:English
pages:19