MammoGrid European federated mammogram database implemented on a GRID infrastructure presented by Salvator Roberto Amendolia / CERN on behalf of the MammoGrid.

MammoGridEuropean federated

mammogram database implemented on a GRID

infrastructure

presented by Salvator Roberto Amendolia / CERN

on behalf of the MammoGrid ConsortiumHealthGrid Forum

Brussels, 20th September, 2002

20 September 2002 HealthGrid Forum - Brussels MammoGrid Project 2

MammoGrid Consortium• CERN (Technical Coordinator)

– Vitamib (France) - subcontractor Finance/Admin

• Mirada Solutions (UK) – Medical Image Analysis S/W• Univ of Oxford (UK) – Medical Vision Laboratory• Univ of Pisa (I) – Medical Physics section• Univ oF Sassari (I) – Maths & Physics Dept• Univ West of England (UK) – Computing Research• Univ of Cambridge (UK) – Addenbrookes Hospital• Univ Hospital of Udine (I) – Inst of Diagnostic Imaging

– Ospedale Valdese Torino (I) - Breast Screening Unit - subcontr.

– Zybert Computing Ltd. (UK) - Subcontractor for GRIDserver

Project kick-off meeting CERN 18th-19th September 2002


MammoGrid Objectives1. To evaluate current Grids technologies and determine

the requirements for Grid-compliance in a pan-European mammography database.

2. To implement the MammoGrid database, using novel Grid-compliant and Federated-Database technologies that will provide improved access to distributed data and will allow rapid deployment of software packages to operate on locally stored information.

3. To deploy enhanced versions of a standardization system that enables comparison of mammograms in terms of intrinsic tissue properties independently of scanner settings, and to explore its place in the context of medical image formats (DICOM).

4. To develop software tools to automatically extract image information that can be used to perform quality controls on the acquisition process of participating centres (e.g. average brightness, contrast).


MammoGrid Objectives (cont.)

5. To develop software tools to automatically extract tissue information that can be used to perform clinical studies (e.g. breast density, presence, number and location of micro-calcifications) in order to increase the performance of breast cancer screening programs.

6. To use the annotated information and the images in the database to benchmark the performance of the software described in points 3, 4 and 5.

7. To exploit the MammoGrid database and the algorithms to propose initial pan-European quality controls on mammographic acquisition and ultimately to provide a benchmarking system to third party algorithms.


Mammography Diagnosis and GRID.

GRID opens up the possibility of addressing important clinical needs in radiology:

- Data sharing among clinicians for second review diagnosis.

- Image Based epidemiological studies- Computer Aided Quality Control in Acquisition and Diagnosis- Validation of Computer Aided Diagnosis Systems


A GRID INFRASTRUCTURE IS IDEALA GRID INFRASTRUCTURE IS IDEALThe databases to statistically validate image based clinical hypothesis are:

Populated by large number of cases Contain large files (1 mammogram

10Mb+) Geographically distributed repositories Heterogenous database formats Need to be accessible to co-workers

Development and validation of medical image analysis solutions demands:

Computationally expensive simulations. Repeated runs for optimal parameter tuning. Statistical test rigs. Remote execution and maintenance

Services (e.g. security) must be system-resident, invisible, generic


Mammo GRID Philosophy• Project concentrates on applying emerging GRID technology rather

on developing it.• It plans to implement a ‘lightweight’ (but fully functional) GRID and

study its usage in hospitals• It will draw heavily on other Grids projects e.g. DataGrid• It will deliver a prototype federated database of mammograms in

hospitals in the UK and Italy• It will investigate :

– the role of Grids-based meta-data for resolving queries– the use of standardised mammogram to resolve image and

population variability – Health data security using a novel ‘Grid box’– the infrastructure needed for CADe

• It will provide rapid feedback from the Hospital community• And should inform the next generation of HealthGrids developments


HEP vs Mammogrid

• Similarities– Large number of big files – Files can be sensibly organized in directory tree– Need to replicate and move file copies between sites– Need to execute commands on the node which hosts

data locally

• Difficulties– Complexity of co-working in medical environment– Lack of trained IT personnel – Confidentiality


Problem • Typical next generation HEP experiment

– Large scale simulation & reconstruction effort– Heavily distributed processing and event storage

• ~1000 scientists in ~100 of institutions– Complex analyses of distributed data– Large files (one event up to 2GB)

10^9 files/year (x n, n>2)2 PB/year

• Experiment lifetime – 20-25 years

• GRID – Widely accepted as a solution


The challenge in HEP

Can we provide, building on top of available public domain and open source components and standards, a functional distributed computing infrastructure to the community of our users which will remain operational even if underlying technologies keep changing?


One HEP solution• AliEn framework

– Lightweight, simplified but fully functional GRID implementation

– Distributed file catalogue with support for replication– Strong (certificate) based authentication – Resource broker – Possibility to submit and execute commands in the system

• It makes extensive use of Open Source components and the latest internet standards (SOAP, Web services, OpenSSL, OpenLDAP, Globus, MySQL, perl, CPAN)

• AliEn provides coherent interface and shields user from rapid changes in underlying technology

• On mid to long term, ALICE experiment remains committed to integrate AliEn with DataGRID solutions as they become available

• Given the worldwide nature of ALICE computing, AliEn will be interfaced to other GRID solutions (U.S., Asia, Japan..)


AliEn and Open Source

Benefits of development based on OpenSource components are more than obvious…

1%

99%

AliEn

Open SourceModules


Some AliEn featuresAuthentication module which supports various authentication methods (including Globus/GSI)Distributed file catalogue built on top of RDBMS with user interface that mimics the file system Secure file transport and replication ServiceTask queue which holds commands to be executed in the system (commands, inputs and outputs are all registered in catalogue)Computing and Storage elementsMetadata catalogue Monitoring frameworkC/C++/perl APIWeb portal

EDG compatible authentication and JDL


GRID of GRIDs

AliEn User Interface

AliEn stackiVDGL stack EDG stack


Federated System SolutionHospital Italy

Healthcare Institute

University Database

Hospital UK

Shared meta-data

Analysis-specific data

•Knowledge is stored alongside data•Active (meta-)objects manage various versions of data and algorithms•Small network bandwidth required

Clinician’s Workstations

QueryResult

LocalQuery

LocalAnalysis

LocalAnalysis

LocalAnalysis

LocalAnalysis

Massively distributed dataAND distributed analyses

GRIDLocalQuery

LocalQuery

LocalQuery


MammoGrid Workpackages• WP 1 Project Management• WP 2 User Requirements Specifications• WP 3 Information System Architecture and Grid

Compliance• WP 4 Local Node Implementation• WP 5 Integration Environment• WP 6 Standardisation Software• WP 7 Acquisition Control Software for a Grid deployment

scenario.• WP 8 Software for CAD Diagnosis and QC in a Grid deployment

scenario.• WP 9 Pilot Study 1: Breast Density Measurements in a Grid

deployment Scenario• WP10 Pilot Study 2: CAD for Quality Control in a Grid deployment

scenario

• WP11 Dissemination & Exploitation


MammoGrid Implementation

Use case/validation

UserReq’s

& Specs

GRID/DBinfrastructure

H/Wlocal node implem.

StandardisationS/W.

Application S/W

Dissemination & Exploitation

Project Management

WP 2CERN/UWE

Hospitals

WP 3 - CERN/UWE

WP 4 - Mirada

WP 6 - Mirada

Integrationtest bed

WP 5 - CERN

WP 7&8 - Oxford,Pisa/Sassari

WP 9&10Cambridge

Udine

WP 11 - All

WP 1 - CERN (Vitamib)

spec

ific

atio

ns

Info

rmat

ion

infr

astr

uctu

re


Main Deliverables/milestones

• User Requirements Specification and Technical System Specification (months 3, 6)

• Prototype GRID-compliant database and information infrastructure (first release m. 18, final rel. m. 36)

• Packaged medical imaging workstation with interface to GRID, secure GRID box, (month 12)

• Grid compliant SMF software (month 12)• Application software (CADe etc.) (months 12, 24,

36) • Clinical Trial results (month 24, 36)


Dissemination & Clustering• “GRID related” dissemination efforts

– CERN/UWE members of Global Grid Forum, GGF and of the Object Management Group, OMG

– CERN already in GRIDSTART for dissemination and closely working with DataGrid. UWE to join.

– Pursue relationship with EU funded GRID projects (e.g. CROSSGRID, BIOGRID, GEMSS)

– Develop relationship with NDMA project (USA)

• “Clinical dissemination”– Advisory Group (Oxford, Torino, IMIM & GEIE-LINC) to give

visibility to future MammoGrid partners.

• “Medical Image Analysis dissemination”– Academic dissemination through targeted conferences and

journals in computer science & medical informatics


Exploitation• Mirada

– Establish SMF as a standard for breast density measurement.– Establish SMF as a standard for mammogram data exchange.– Prototype SMF based review workstation for CADiagnosis.

• CERN/UWE – Study the spinning-out of GRID/database technologies to address

needs in healthcare

• Oxford– Develop patentable technologies for medical image analysis

products.– Transfer of technology agreement with Mirada.

• Pisa/Sassari– Develop patentable technologies for medical image analysis

products.


MammoGrid Added ValuePrevious EC projects MammoGrid

• Distributed network of medical images DB (Medimedia)

• GroupWare collaborative platform (Horizon)

• Remote consultation/diagnostic, (Europath)

• Case-by-case security implementations

• Point-to-point connectivity on the Web

• Multiple federated mammogram databases

• Clinicians tele- and co-working in new and innovative groupings (‘virtual organisations’)

• Distributed and ubiquitous analysis and diagnosis

• Security handled by services’ on the Grid’

• Massive connectivity and datasets with massive available compute power


eDiamond (UK) and GP-CALMA (I)

• similar approach, one UK based, one Italy based, one Europe wide

• synergy!!• Different areas of application

• Teaching & CPD eDiamond• Tele-diagnosis GP-CALMA• Quality control MammoGrid• Epidemiology MammoGrid• Algorithm development: data mining

eDiamond• CADe development GP-CALMA


GRID - which one?• One year ago GRID projects were still in their infancy

• Globus Toolkit™ – Open source toolkit for building GRID infrastructure and

applications– APIs, SDKs, and tools which implement Grid protocols & services

• Components Globus Toolkit™ are used in many current GRID (including EU DataGrid) projects but…– toolkit is only a toolkit– someone has to do integration work

• Emerging new technologies and standards– Web services, W3 standard protocols..– B2B solution, not specifically designed to support massive

distributed computing


“Web Services”• Geneva, May 2001:

– Instead of using Globus toolkit or waiting for DataGRID to deliver re-packaged version of Globus, we decided to try different path and use Web Services and related standards as a backbone of our GRID implementation

• Web Services - components– WSDL: Web Services Description Language

• Interface Definition Language for Web services– SOAP: Simple Object Access Protocol

• XML-based RPC protocol; common WSDL target– UDDI: Universal Desc., Discovery, & Integration

• Directory for Web services

GGF in Toronto, February 2002 : Web Services declared as a key element of new OGSA (Open Grid Services Architecture) initiative


I

OGSA


AliEn Open Source Components

• SASL/OpenSSL/OpenCA as authentication protocol• Globus/GSS as an implementation of authentication

compatible with other Grid projects• ClassAds language for job description (compatible with

EU DataGrid) • OpenLDAP for configuration management• Apache for Web Portal• MySQL as relational database backend


Conclusions on AliEn

• After just one year of development with limited resources, AliEn has become a lightweight, simplified but fully functional GRID implementation

• Adding AliEn interface between our application and external GRID infrastructure – Allows us early prototyping of GRID technology– Enables massive distributed production– Protects users from rapid changes in technology

• It makes extensive use of Open Source components and the latest internet standards (SOAP, OpenSSL, OpenLDAP, Globus, MySQL,

• These components and modules are interchangeable and easily replaceable by other (possibly non-OpenSource) components offering the same functionality


(…)

mamogrid.cern.ch

Region#1 Region#2 Region#3

Hospital#1 Hospital#2

Town#1 Town#2 Town#3

Hospital#1 Hopsital#2 Hospital#3

User

It is possible to define an abstract interface (API to be used by Mirada Workstation) for which we will provide an implementation based on AliEn.

If there is clear benefit for our project, we could consider using

WebSphere as UDDI service

DB2 as relational DB backend

However, this has not been planned and would require additional effort.

In our current picture, GridBox will act as adapter between GRID (services) and Mirada Workstation

MammoGrid European federated mammogram database implemented on a GRID infrastructure presented by Salvator Roberto Amendolia / CERN on behalf of the MammoGrid.

Documents

mammogrid database

image information

database technologies

uk subcontractor

software tools

tissue information

cmpda grid infrastructure

novel gridcompliant