1. Imaging as a biomarker: standards for change measurements in therapy NIST, Gaithersburg, MD

2. Workshop « Imaging as Biomarker »  General issue: Imaging use for Clinical trials  Organized by NIST  Supported by:  Sponsoring organization: NIH, NCI (Larry Clarke)  FDA  Pharmaceutical industry: PhRMA, big labs  Normalization structures: NIST, NEMA  Professional societies: RSNA, AAPM, SNM, ISMRM, SPIE  Vendors: Siemens, GE, Philips  General issue: Imaging use for Clinical trials  Organized by NIST  Supported by:  Sponsoring organization: NIH, NCI (Larry Clarke)  FDA  Pharmaceutical industry: PhRMA, big labs  Normalization structures: NIST, NEMA  Professional societies: RSNA, AAPM, SNM, ISMRM, SPIE  Vendors: Siemens, GE, Philips

3. The issues (1/2)  Use of imaging to evaluate new treatment efficiency:  Not only a subjective reading of images  But, more and more objective measurements with “quantitative” methods  Examples : Alzheimer desease, lung cancer, arthrosis…  Low efficiency statement of clinical trials  Longs, expensives, and often… no positive result!  …  Use of imaging to evaluate new treatment efficiency:  Not only a subjective reading of images  But, more and more objective measurements with “quantitative” methods  Examples : Alzheimer desease, lung cancer, arthrosis…  Low efficiency statement of clinical trials  Longs, expensives, and often… no positive result!  …

4. The issues (2/2)  Main reasons:  Too often, mediocre quality of image data  No quantitative  Reproducibility issues (intra system and inter systems)  Acquisition protocols non respected → lot of cases are non usable  Data processing differs between different site ( → “site” effect)  Too subjective  Not enough automated  Insufficient quality of clinical data associated  “Local” semantic for clinical information  Resulting bias amplify intrinsic variability of the phenomena  → increase of the necessary number of cases  Data management cost  Because support tools needs to be rebuild case by case  Data cannot be re-used  Despite the politic of sponsoring organizations (dissemination plan)  No sharing infrastructure  Main reasons:  Too often, mediocre quality of image data  No quantitative  Reproducibility issues (intra system and inter systems)  Acquisition protocols non respected → lot of cases are non usable  Data processing differs between different site ( → “site” effect)  Too subjective  Not enough automated  Insufficient quality of clinical data associated  “Local” semantic for clinical information  Resulting bias amplify intrinsic variability of the phenomena  → increase of the necessary number of cases  Data management cost  Because support tools needs to be rebuild case by case  Data cannot be re-used  Despite the politic of sponsoring organizations (dissemination plan)  No sharing infrastructure

5. Solution components, standard based (1/2)  Standard phantoms to calibrate imaging systems  Standard study protocols  To be really identical between machines (patient preparation, image acquisition, pre-processing and applied corrections, reconstruction…)  Standard procedures (guidelines) to manage clinical trials  good comprehension of protocols, physician and technologists education, quality assurance  Meta data and clinical data standardization  Need the use of normalized terminologies, ontologies…  Standard phantoms to calibrate imaging systems  Standard study protocols  To be really identical between machines (patient preparation, image acquisition, pre-processing and applied corrections, reconstruction…)  Standard procedures (guidelines) to manage clinical trials  good comprehension of protocols, physician and technologists education, quality assurance  Meta data and clinical data standardization  Need the use of normalized terminologies, ontologies…

6. Solution components, standard based (2/2)  Management tools adapted to the data and processing management  Scientific calculation  Step sequencing (including quality assurance)  Providing the required traceability  Compatibility issues with current PACS (regarding image management)  Reference implementation (open source shared tools)  Standard processing tools (that can be shared)  Open source  Plug-in or services remotely executable  Standard data  Anatomical atlases  Reference data (e.g. healthy populations )  Management tools adapted to the data and processing management  Scientific calculation  Step sequencing (including quality assurance)  Providing the required traceability  Compatibility issues with current PACS (regarding image management)  Reference implementation (open source shared tools)  Standard processing tools (that can be shared)  Open source  Plug-in or services remotely executable  Standard data  Anatomical atlases  Reference data (e.g. healthy populations )

7. Needs for the future  Medical  No more and less than tomorrow treatments for numerous pathologies  Vendors  Pharmaceutical industry, decrease of validation costs for new medication  Imaging systems industry  New generations of systems. Economical model.  Sharing / interoperability of processing tools  Scientific  Validation of processing and image quantification tools  Diffusion and re-use of tools  Medical  No more and less than tomorrow treatments for numerous pathologies  Vendors  Pharmaceutical industry, decrease of validation costs for new medication  Imaging systems industry  New generations of systems. Economical model.  Sharing / interoperability of processing tools  Scientific  Validation of processing and image quantification tools  Diffusion and re-use of tools

8. What happened since the NIST workshop ?  The Gil Jost (RSNA), Larry Clarke (NCI) and Michael Vannier (University of Chicago) propositions were well received in France and Europe  SFR and ESR decided to join RSNA and NCI in an international initiative on biomarkers  At RSNA meeting, Guy Frija (ESR) and Philippe Grenier (SFR) met Gil Jost to discuss on this topic.  SFR will launch a biomarkers working group.  The Gil Jost (RSNA), Larry Clarke (NCI) and Michael Vannier (University of Chicago) propositions were well received in France and Europe  SFR and ESR decided to join RSNA and NCI in an international initiative on biomarkers  At RSNA meeting, Guy Frija (ESR) and Philippe Grenier (SFR) met Gil Jost to discuss on this topic.  SFR will launch a biomarkers working group.