Abstract: Biomedical researchers work with many different types of image and signal data. My students have worked with 1D brain signals, 2D mouse eye images, 3D CT scans of human skulls, 3D meshes from six-camera stereo of children's heads, 4D functional MRI (fMRI) brain scans, and other modalities. When studying diseases and genetic disorders, medical researchers would like to quantify the amount of presence of a condition, rather than just identifying the condition (which they already know). This leads to the need for similarity-based retrieval.
The objective of our work is to develop a unified methodology for organization and retrieval of biomedical data from scientific experiments. We have proposed and developed a first prototype system that combines the similarity-based retrieval methodology of content-based image retrieval systems and image indexing with the efficiency of relational database systems. In this talk, I will discuss the motivating biomedical applications, the design of the system, and the GUI for the first prototype system. I will then talk about the new biomedical applications for which we are designing similarity measures and the more general patient retrieval system we are developing.