Medical Imaging Application Layer for
Infrastructures using Grids
A Cyber-Infrastructure
for Medical Imaging in
Oncology
Ignacio Blanquer
Universidad Politécnica de Valencia, ITACA
High-Performance Computing and Networking Group
www.grycap.upv.es
INFSO-RI-508833
Contents
Grupo de Redes y Computación de Altas Prestaciones
• Challenges of Medical Imaging.
• A Valencian Cyber-Infrastructure for Medical Imaging
on Oncology (CVIMO).
• Data Integration on CVIMO
–
–
–
–
Semantic Integration and Indexation.
Scalable Distributed Storages.
Security Requirements.
High-Performance Processing.
• SHARE - Future Trends.
• Conclusions.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
2 / 19
Advantages of Medical Imaging
Grupo de Redes y Computación de Altas Prestaciones
•
•
•
•
Medical Images Constitutes a Main Information Source for
Diagnosis and Therapy.
Medical Images are Used to Identify Trauma, Organ
Malfunction, Tumours, Surgery Planning, etc.
Medical Images are Present in all Medical Disciplines.
They are also Used for the Quantitative Evaluation of
Masses, Flows, Injures, etc.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
3 / 19
Grupo de Redes y Computación de Altas Prestaciones
Challenges in
Medical Imaging
• Integration of Digital Medical Imaging
in Clinical Practice is a Well-Known
Problem Successfully Addressed by
Many Commercial Solutions
– Archiving in Hospitals.
– Post-Processing in Radiology
Worksations.
– Reporting and Workflows.
• However, Research is Introducing New Requirements Which
are Not Covered
–
–
–
–
Inter-Organisation Data Federation and Sharing.
Computational Demand of Advanced Post-Processing Tools.
Semantic Integration of Repositories.
Epidemiological Exploitation of Data.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
4 / 19
Grupo de Redes y Computación de Altas Prestaciones
•
Challenges in
Medical Imaging
Inter-Organisation Data Federation and Semantic Integration
– Training is Mainly Based on Evidence.
– Research on Rare Pathologies Require Collecting a Large Number of
Cases, Possibly From Different Centres.
– Privacy Must be Preserved.
– Challenge: A Secure Infrastructure to Share Selected Interesting
Cases Among Different Federated Centres.
•
Computational Demand of Advanced Post-Processing Tools and
Epidemiological Exploitation of Data
– Size of Medical Images are Increasing and New Postprocessing Tools
are Computationally Intensive, Exceeding in Many Cases the
Resources of Hospitals.
– Advanced Statatistical Methods and Data Mining Techniques Over
Large Repositories Will Produce Important Information for Public
Health Planning.
– Challenge : An Infrastructure to Deploy High-Performance and HighThroughput Processing Services Over a Federated Repository.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
5 / 19
Grupo de Redes y Computación de Altas Prestaciones
Valencian Cyber-infrastructure for
Oncological Medical Imaging (CVIMO):
Semantic Structuring of Distributed
Medical Image Repositories
INFSO-RI-508833
CVIMO
Grupo de Redes y Computación de Altas Prestaciones
• Valencian Cyber-infrastructure for Oncological Medical Imaging
(CVIMO)
• It is an Environment that is Being Developed under the Project
“Creation of a Cyber-infrastructure for Learning, Research and
Epidemiology Study of Cancer Using Medical Images”.
• 7 Hospitals from the Land of Valencia are Participating in the
Creation of an Advanced Repository for Medical Images and
Radiology Reports on Three Major Areas (Central Nervous
System, Lung and Liver).
• The Platform Organises the Studies According to their Relevance
to Different Virtual Communities Expert on Different Topographies
and Morphologies.
• It Provides Tools for Creating,
Comparing and Searching Studies
From Structured Reports, Along with
High-Performance Post-Processing
Services.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
7 / 19
CVIMO Technology: TRENCADÍS
Grupo de Redes y Computación de Altas Prestaciones
• Towards a Grid Environment for Processing
and Sharing DICOM Objects
– TRENCADIS Aims at the Development of a
Middleware to Create Virtual Repositories
of DICOM Images and Reports.
– It Uses a Semantic Model to Organise the Data.
– Data is Encrypted and Decrypted to Ensure
Privacy Protection.
– High-Performance Services are Included with the System.
– Architecture Totally Based on WSRF.
• Objective: Creation of Virtual Shared Repositories of Medical
Images.
–
–
–
–
–
Complementary to PACS.
Intended Mainly for Research and Training.
Multicentric and Multiuser.
Data to be Shared is Explicitly Selected.
Data is Pseudoanonimised Before Entering in the System.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
8 / 19
TRENCADIS: Data Indexation
Grupo de Redes y Computación de Altas Prestaciones
• Semantic Organisation
View:
and 22 Years
Years
View:(e.g.
(e.g.Patients
Patients Between
between 1 and
with
the Front)
Front)
withFindings
Findings on the
Experiment:
Experiment:
(e.g.Neuroblastoma)
Neuroblastoma)
(e.g
User Community
Comunity
User
(e.g.Paediatric
Paediatric Neuroimaging)
Neuroimaging)
(e.g.
Global Database:
Database: All
All the Images and
Global
and
Reports
Shared
Reports Shared
– Users Organise Themselves on VOs.
– From the Studies Available, Only
Those Matching the Selection
Criteria of the VO Profile are
Accessible.
– From the Images Available to a
Virtual Community, a User Can
Create an Experiment with the
Studies Matching a Set of
Restrictions.
– From this Experiment, More Detailed
Views can be Obtained.
– The Criteria for the Selection of the
Relevant Information Relies on the
DICOM Tags of the Image and the
Structured Report.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
9 / 19
Structured Reporting
Grupo de Redes y Computación de Altas Prestaciones
• Seven Templates Have Been Generated By the Experts
– Report for the Staging of Malignant Liver Neoplasia, Small and NonSmall Cell Lung Cancer and Intraaxial Tumours of Central Nervous
System.
– Follow-up Reports for Liver Metastasis, Lung Carcinoma and Intraaxial
Tumours of Central Nervous System.
• The Reports are Structured and Coded Using the Rules of DICOMSR.
• Standard Coding (Mainly DICOM) Has Been Used When Possible,
Following the DICOM-SR Rules To Introduce New
Coding Schemas.
• The Reports Generate Automatically the TNM
Staging Code (From the Radiological Information)
in the Cases of Liver and Lung.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
10 / 19
TRENCADIS: Security
Grupo de Redes y Computación de Altas Prestaciones
• Authentication and Authorisation
– Users are Authenticated Through X.509 Certificates in an “Single
Sign-on” Procedure (Using Proxies).
– Roles of the Users (And Though the Virtual Community and the
Access Permissions) are Managed Through VOMS proxies.
• Privacy
– All Transactions are Based on
Secure Protocols.
– Data is Encrypted on the Grid
Storage to Avoid The Access of
Users with Privileges.
– Keys are Split and Shared
Through the VO Group.
Fragmento de Clave
Clave
Credenciales
Repositorio
Encriptado
Fragmento de Clave
Datos Desencriptados
Credenciales
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
12 / 19
Deployment
Grupo de Redes y Computación de Altas Prestaciones
Provider
Provider
Client
Provider
Estación RM
CT
Workstation
PACS
DICOM
Q/R
DICOM
Q/R
DICOM
Q/R
Internet
Https, ssl & GridFTP
Ports: (80)80, (84)43, 2911).
UPV
Code
Manager
Web
Server
Storage & Index
Server
Oncology
Server
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
Processing
Servers
13 / 19
Deployment
Grupo de Redes y Computación de Altas Prestaciones
Provider
Provider
Client
Provider
Estación RM
CT
Workstation
PACS
DICOM
Q/R
DICOM
Q/R
DICOM
Q/R
Internet
Https, ssl & GridFTP
Ports: (80)80, (84)43, 2911).
UPV
Code
Manager
Web
Server
Storage & Index
Server
Oncology
Server
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
Processing
Servers
14 / 19
Interest on Schizophrenia
Grupo de Redes y Computación de Altas Prestaciones
• Research on Schizophrenia Considers Several Data Sources
– Medical Imaging Sources
 MRI
• Functional MR in the Provocation and Analysis of Schizophrenic Episodes,
such as Hearing Hallucination.
• Diffusion Analysis. Analysis of the Permeability of Cellular Barriers in
Neurons.
• Morphometry. Quantitative Measurement of Regions, Functional Areas and
Lesions
 Spectroscopy
• Analysis of the Metabolomics.
– Genetics.
– Clinical Records.
• Semantic Integration of Data is Complex but
Very Important in Areas where the Diseases
are Based on the Combination of Multiple
Factors.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
15 / 19
SHARE
SHARE is an European Project (www.eu-share.org) Which Mission is
to Precise the Target, the Current Situation, the Key Gaps, Barriers
and Opportunities, Short Term Achievements and Key Elements and
Actors to Reach the HealthGrid Vision.
The Main Result of SHARE is a Roadmap
The Roadmap will Cover Issues Regarding Networks,
Infrastructure Deployment, Grid Operating Systems, Services to
End Users, Standards Requirements, Security, Legislative
development and Economic Issues.
Roadmaps
Destination, Transportation, “Whistle Stops”, Starting Point, Travellers and
Difficulties in the Trip.
Analysis of the Current Situation, Requirements, Actors, Needs,
Opportunities and Barriers from the Technological, Legal, Ethical and EndUser’s Point of View.
Use Cases
Epidemiology. Important Issue on Data Integration.
Innovative Medicine. Main Focus on Large-Scale Computing.
A Roadmap for Data Integration in e-Health
16
Conclusions
Grupo de Redes y Computación de Altas Prestaciones
• Medical Imaging is a Key Application for Exploiting the
Benefits of Grids.
• There are Successful Examples of Using Grids to
Provide the Computing Needs or to Structure and
Consolidate Distributed Databases.
• Key Aspects in eHealth such as Security and
Communication Efficiency are Essential Part of Grids
and Should not be a Barrier.
• A Key Factor is the Involvement of Users. The Creation
of Platforms Integrating Hospitals and Research
Centres is a Challenge.
• Although Research is the
Most Sensible Aim, Clinical
Production could Also Benefit on
the Long Term.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
17 / 19
Conclusions-II
Grupo de Redes y Computación de Altas Prestaciones
• Data Integration Issues on Medical Imaging
– There are Several Technological Barriers to Overcome:
 Network Infrastructure of Hospital Centres is Generally Insufficient
for the Integration of Large Data.
 Privacy and Security Barriers Make Computer Services of Hospitals
Reluctant to Integration.
– Although They are Not the Most Important Barriers …
 Data Must be Carefully Structured for an Efficient and MachineEnabled Integration. Ensuring High Data Quality is a Must.
 Sharing is Always Difficult when Competence is Very High.
 Medical Users Must Participate Since the Very First Moment. The
Human Factor is the Most Difficult One.
 Data Integration Must be Encouraged with
Added-Value Tools. Post-Processing
Services, Automatic Classification, etc.
INFSO-RI-508833
Challenges on Imaging · CVIMO · TRENCADIS Technology · SHARE · Conclusions
18 / 19
Grupo de Redes y Computación de Altas Prestaciones
INFSO-RI-508833
Contact
Grupo de Redes y Computación de Altas Prestaciones
Vicente Hernández / Ignacio Blanquer
Universidad Politécnica de Valencia
Camino de Vera s/n
46022 Valencia, Spain
Tel: +34-963879743
Fax. +34-963877274
E-mail: vhernand@dsic.upv.es
iblanque@dsic.upv.es
INFSO-RI-508833