Our research group uses computational methods to leverage the information in radiology images to enable biomedical discovery and to guide physicians in personalized care. Just as biology has been revolutionized by online genetic data, our goal is to advance radiology by making the content in images computable and to electronically correlate images with other clinical data such as pathology and mol
ecular data. Our work develops and translates basic biomedical informatics methods to improve radiology practice and decision making in several areas: tools to efficiently and thoroughly capture the semantic terms radiologists use to describe lesions; standardized terminologies to enable radiologists to describe lesions comprehensively and consistently; image processing methods to characterize the shape of lesions; content-based image retrieval with structured image information to enable radiologists to find similar images; methods to enable physicians to quantitatively and reproducibly assess tumor burden in images and to more effectively monitor treatment response in cancer treatment; natural language techniques to enable uniform indexing, searching, and retrieval of radiology information resources such as radiology reports; and decision support applications that relate radiology findings to diagnoses to improve diagnostic accuracy. We collaborate with a variety of investigators at Stanford both in Radiology and Oncology as well as with investigators outside Stanford. We are engaged in scientific collaborations with the National Center for Biomedical Ontology, and we participate in a national working group that is developing imaging informatics infrastructure for the cancer Biomedical Informatics Grid program at National Cancer Institute. Our ultimate goal is to bridge the divide between radiological knowledge and practice--for all radiological knowledge and research data to be structured, accessed, and processed by computers so that we can create and deploy decision support applications in image workstations to improve radiologist clinical effectiveness.
