Sandy Napel, PhD,
Professor of Radiology,
Co-Division-Chief, Integrative Biomedical Imaging Informatics at Stanford (IBIIS)
Co-Director, Radiology 3DQ Lab

Abstract:
Radiomics is a process that extracts a feature vector of quantitative metrics to describe an image or a part of one. The vector may contain tens through thousands of elements, each describing an aspect of size, shape, edge-sharpness, histogram, or texture. Once obtained, this vector may be integrated with other data, including elements from the electronic medical record, including demographic information and results of other assays, e.g., genomic sequencing.  In this talk, I will discuss the background leading up to the development of radiomics and how it is being used in research today. I will also describe the radiomics workflow for cancer, which includes tumor identification, semantic annotation with a controlled vocabulary, (possibly) segmentation, integration with other data, and predictive modeling. I will describe the Stanford Quantitative Imaging Feature Pipeline (QIFP) and how it is being used in cancer imaging, and discuss remaining challenges.