Standardization Efforts of Digital Pathology in Europe EURO-TELEPATH. Telepathology Network in Europe

COST Action IC0604 Marcial García Rojo, MD, PhD

COST is supported by the EU Framework Programme

Chair. Pathology Department Hospital General Universitario de Ciudad Real / Spain

ESF provides the COST Office through a European Commission contract

Scientific context and objectives (1/2) • Background / Problem statement: Adequate technological framework for the management of multimedia electronic healthcare records • • • •

Informatics applied to Anatomic Pathology Research on standards of digital medical images World wide search engine Collaborate with standardisation bodies

• Main challenges are educational, technological, and complexity of health information systems. • Brief objectives: Evaluate and validate the common

technological framework and communication standards required to access, transmit and manage digital medical records by pathologists and other medical specialties in a networked environment. 2

Scientific context and objectives (2/2) •

Research directions: 1. Automation procedures in Pathology. Best technology available and under research. 2. Scanning solutions for Pathology microscopic slides. 3. Technological solutions for compression and storage problems with large image files. 4. Virtual slide standard viewer specifications which allow efficient reviewing of pathology images. 5. International standards (DICOM, HL7, SNOMED, CEN) and IHE initiative. 6. Model for pathology and other hospital information systems 7. An European-scope telepathology network 8. Collection of interesting and typical samples, and clinicopathological sessions 3

Industry involvement and standardization • • • • • • • •

Tribvn, France. SME Philips Healthcare, The Netherlands SlidePath, Ireland, SME VICOMTech, Spain, SME Aurora MSC, SME SATEC, Spain Leica Microsystems, Germany COST Action IC0604 is highly involved with all standardization bodies related to medical informatics and Pathology, namely DICOM (images), JPEG (compression), HL7 (messages), IHTSDO-SNOMED CT (terminology), and IHE (how to use standards). 4

Significant Highlights in Science or Networking (1/2) 2st Nottingham Image Analysis Training School • 27-29th June, 2010. Dubingiai, Lithuania • 38 participants from 10 different countries • A successful multidisciplinary course with 18 medical doctors, 4 biologists, 16 engineers • Related STSM months later, also related with image analysis. • Third: June 23-24, 2011. Udine, Italy in dates close to the 25th Adriatic Society of Pathology Congress

5

Highlights

6

Significant Highlights in Science or Networking (2/2) • 10th European Congress on Telepathology and 4th International Congress on Virtual Microscopy to be held in Vilnius, Lithuania 1-3 July, 2010. • I.T. engineers, bioinformaticians, biologists, pathologists • A productive image analysis algorithms contest • Congress proceedings published in the journal "Diagnostic Pathology“ • Next Congress:

Venice, Italy 2012! 7

Highlights

8

Working groups 4 Working Groups

Coordinators

WG1. Pathology Business Modelling

Thomas Schrader (Germany) and Luis Gonçalves (Portugal)

e.g. Consensus study of existing workflows in pathology departments

WG2 – Informatics Standards in Pathology e.g. Participation in covered standards bodies and initiatives.

WG3 – Images: Analysis, Processing, Retrieval and Management

Christel Le Bozec (France) and Bernd Blobel (Germany) Janina Slodkowska (Poland) and EPFL (Switzerland)

e.g. study on the image analysis models (methods, systems, tools, fields of pathology).

WG4 – Technology and Automation in Pathology

Marcial García Rojo (Spain) and Bernd Blobel (Germany)

e.g. analysis of microscope brands. 9

WG 1 Business modeling in Pathology

• Comparison study between different notations in business modeling
Business Process Modeling Notation – BPMN
Event Process Chain – EPC
UML Activity Diagram

10

Results • BPMN is the modeling notation which is clearer and more understandable to pathologists. • EPC is harder for pathologists to read and understand. It is probably because EPC is not as popular as BPMN and UML AD, so pathologists are not so familiar with the concept of event-driven process. 11

BPMN diagram elements

12

Challenges • The number of young female researchers has been increasing, but the number of STSM (short visits) is low in our COST Action. • Terminology standardization research has become a topic of high interest, after a pathology image standard (DICOM Supp. 145) has been achieved • Image scanning technology is evolving slower that expected in speed and image quality. • Standardization in automatic image analysis and implementing a distributed pathology digital slides European database are our main next goals. 13

Hospital processes

14

Process Map Diagram

Terminology standardization! Author: Thomas Schrader 15

16

17

SNOMED CT

18

WG2 – Informatics Standards in Pathology

• IHE/HL7 Anatomic Pathology • Anatomic Pathology Structured Reports • Extending the scope? • Mapping PathLex to SNOMED CT • Joint IHE AP/DICOM WG26 initiative • Better use of DICOM in Anatomic Pathology workflow (DICOM Modality Work list) Dr Christel DANIEL (IHE Anatomic Pathology co-chair) Jacques KLOSSA (DICOM WG26 co-chair) Bernd BLOBEL 19

Informatics Standards in Pathology: IHE • Intra hospital integration profiles • Anatomic Pathology Workflow (APW) • Ordering and performing anatomic pathology examinations • Community • Anatomic Reporting for Public Health (ARPH) • Sending anatomic pathology reports to public health organizations • Anatomic Pathology Structured Report (APSR) • Sharing/exchanging structured anatomic pathology reports as CDA documents

20

IHE Technical Framework 2010

Revision 2.0 July 23, 2010 Draft for Trial Implementation

2011

2010 & 2011 Supplements for Trial Implementation 21

AP observation (Coded Descriptor) • Breast-Specimen Collection Procedure (OID 1.3.6.1.4.1.19376.1.8.1.4.28)

• Value set OID: 1.3.6.1.4.1.19376.1.8.5.2 displayName Breast excision without wire-guided localization Breast excision with wire-guided localization Total mastectomy (including nipple and skin)

CodeSystem: PathLex – OID: 1.3.6.1.4.1.19376.1.8.2.1 – Admitted exceptions : Unknown, Asked but unknown, Other

22

Results. Implementation of SNOMED CT. Hierarchies of precoordinated terms

• Postcoordination needed in 19% (3 conceptID needed only rarely) (81% precoordinated) • Morphology – Abnormal body structure – Only 44% of the codes!

• Clinical findings (disorders) – 48% of the codes

• Qualifiers, normal anatomic structures, procedures, physical object, substance, organism – 10% of the codes

Terminology standards (1/2) • Observable entities. Some of the Pathlex-IHE Structured report terms (mainly cancer) could not be mapped to SNOMED CT • Some daily practice Morphology codes (histological findings) cannot be mapped to SNOMED CT morphologically abnormal structure subhierarchy. • Create a subset of histological findings! 24

Terminology standards (1/2) • Pathology reports diagnosis can be coded using “morphologically abnormal structure subhierarchy” or “clinical findings hierarchy”? • Defining links between morphology codes and clinical findings (for instance a link between a granulomatous reaction changes in a pathology specimen and tuberculosis disease • Grading, scales and stages. Consider including version when available (TNM) • Post coordination needed in histological findings (combining 2 or more morphological codes)

25

WG3 Image compression o About 50 TB/year compressed images only for pathology images o An efficient compression method is needed for very large medical images, that can be used with standard image viewers using standards formats like DICOM o JPEG has to be avoided for certain image analysis applications o JPEG2000 is very efficient but client performance can be difficult to be optimized and it is not used by current web browsers. 26

Image compression o A new compression method, JPEG XR, mainly aimed at those digital photography applications where JPEG 2000 penetration has been limited due to complexity reasons o Evaluating Advanced Image Coding (AIC) in medical images. o Image Search: Images can be stored in JPSearch - Part 4 compliant format containing metadata EPFL, in collaboration with JPEG committee. 27

Serendipia project 2008-2011(Spain) From theory to practice INFORMATION SYSTEMS IMAGE INTEGRATION

• Starting point was an Regional Pathology Information System (LIS) in all the hospitals of the Castilla-La Mancha region, avoiding complex integrations. • Image acquisition. Gross station, slide scanners  Integration in PACS

COLLABORATION • Growing need of specialists  New collaboration tools with experts in PATHOLOGISTS Castilla-La Mancha, and other national and international experts. EDUCATION DIAGNOSIS QUALITY

• Improving relationship between hospital, public heatlh service and university by using telemedicine and on line diagnosis • High quality images, collaborative tools, and education are all aimed to improve quality in diagnosis, treatment and prognosis evaluation.

STANDARDIZATION • Using IT standards is mandatory to generate a flexible and replication of

the solution: IHE Pathology, HL7, DICOM, PACS, SNOMED CT

Conclusions • Lack of experience with DICOM Sup 145. Current solutions based on JPEG2000. A role for AIC? • IHE technical framework successfully tested • SNOMED CT needs to be improved (in morphology hierarchy, postcoordination,…) • Consensus is needed in image analysis data, TMA data exchange • Pathologists need to understand better how IT standards work 29