View source for Abstract

Daniel Rubin (Stanford)

'''Imaging Big Data: Talking About Images'''
 
:Driving use cases for imaging in the era of Big Data
:The central role of imaging ontology
:Example projects using ontologies

:The Big Data revolution in the biomedical sciences is being driven by the growth in databases of primary scientific data as well as the published literature. Most of these data--comprising molecular and clinical data—are being annotated with ontologies as part of the curation process, and analysis of these annotations is enabling scientific research. However, these activities generally lack imaging data, because images are complex, unstructured objects whose semantic content are not explicit. Nonetheless, incorporating images into analyses of biomedical data is increasingly important for discovery in the Big Data era. Many researchers are now attempting to integrate imaging data with molecular and clinical data in large scale to discover non-invasive imaging biomarkers of disease subtypes and indicators of their response to treatment. At the same time, there is increasing interest in ontologies in imaging domains, and using them for structuring description of imaging results in a structured reporting format. In this talk I will discuss recent activities that are leveraging imaging in large-scale discovery efforts and the central role of imaging ontology for enabling this research. I will describe some of the current ontology developments in imaging, emerging standard information models for making imaging results explicit, and tools for producing semantic annotations on images during the image viewing workflow. I will conclude with thoughts on challenges for imaging ontology and future directions.