Thursday, March 5, 2015

ECR 2015 - Imaging Biobanks

A. The biobanks: genomic, moleculomic and proteomic - which link to radiomics? - M. Borra
B. Radiomic: report from the ESR WG on Imaging Biobanks. - H.U. Kauczor
C. Existing imaging biobanks. - A. Jackson
D. Extraction and analysis of biomarkers from medical images. - B. Gibaud

This session discussed the importance of establishment of imaging biobanks and integrating them into conventional biobanks. The imaging can have a large impact on these biobanks and the development of imaging biomarkers. The ESR has established a workgroup on this topic that is currently working on the first whitepaper on this topic to be published in insights into imaging.

One of the main challenges is in the standardization of imaging and postprocessing protocols. This also requires more automatic postprocessing tools. Validation, standardization and integration are the keywords in the developement of imaging biobanks.

Most current imaging biobanks are for research and clinical reference and disease oriented. The number of cases is extremely variable but mainly less then 1000 with mostly CT and MR images and restricted to a specific group of researchers.

One of the first real imaging biobanks is the UK biobank imaging initiative. They started data collection in 2014. Three dedicated imaging centers will be established scanning 21 pts per day to reach the planned participants in the coming six years. A full workup takes four hours for the volunteers in which a number of imaging exams are done. 

Imaging Biomarkers play an important role in imaging biobanks. They can be measurable quantities, measurements instruments, and decision support instruments. Again, standardization and validation are crucial to establish proper imaging biomarkers. Imaging biomarkers should be ontology based (e.g. QIBO, BiomRKRS) but the current ontologies for imaging biomarkers are not yet applicable widely and development of a proper ontology is required. Using the ontologies the implementeration of integrated federated queries to multiple databass is feasible.

No comments:

Post a Comment