Core7

Core 7
Informatics Resource Core
(Director: R.W. McColl, Ph.D.)

Dr. McColl's primary responsibilities are the design and development of novel algorithms and techniques for medical image processing, including the development of software for image handling and communication, and the integration of multi-modal images for visualization/analysis. His recent efforts have been applied toward functional imaging, in particular fMRI and SPECT studies of brain function, and to the provision of access to radiological information via the Internet. Dr. McColl will have responsibility for ongoing data management and analysis. Specific tasks include Image Acquisition, Access, Storage and Retrieval.

All clinical modalities available to those Cores utilizing image-gathering devices (CT, MRI, PET, SPECT, Ultrasound, X-Ray) throughout the Institution, currently support the DICOM (ANSI standard protocol) for electronic communication of radiological and associated demographic information between computer platforms with appropriate client/server software. The main method by which images will be transmitted from the modalities to the Core will be via this mechanism. With a combination of the http/CGI mechanism and an SQL-compliant database, primary access to image data for the interested researcher will be via a Web browser. Researchers will be able to access only those studies to which they have authorization, thereby protecting patient privacy (we will comply specifically with HIPAA guidelines) and reducing confusion/search time requirements of the user. A simple image display interface is available to the user for qualitative comparison of images. An additional interface allows the user to download all or part of the study in its original data format for local use. Actual disk files containing the captured image data will be archived and removed as space is needed (at least two months of data will be online at all times) with a record of the original entry added to a separate, Web-accessible SQL database for historical auditing and data mining purposes.

Co-registration and Image Fusion will be needed for the analysis of longitudinal or multi-modal information from a given patient or from an animal study. Voxel-by-voxel analysis of time course signals will also be of necessity. Researchers may pursue sequential and/or multi-modal (structure/function) studies, wherein structural information obtained via MRI or CT, is combined with functional information such as nuclear medicine or fMRI studies of vascular or metabolic processes, or of gene expression or receptor binding. The available information provided by each modality will determine whether automated or interactive registration tools. A series of extensible image handling environments have been created by Dr. McColl, using a variety of public domain and commercial application development platforms, to provide the capabilities of co-registration, image fusion, and voxel-based analyses for qualitative and quantitative information gathering. It is envisaged that these approaches may be drawn together to provide all users with the necessary tools to manipulate and analyze their data without heavy dependence on Core personnel. The role of the Core in data analysis is foreseen as threefold: 1) provide or recommend tools for users to obtain and analyze their own data; 2) offer an analysis service based on a single platform/software tool; 3) create the mechanism to support meta-analysis (data mining) of the ensemble database as needs for such information emerge. In particular, the MEDx package (Sensor Systems) offers many of the image processing and analysis features described above and allows for the creation of additional plugins (locally created components). We propose to use MEDx as the main software platform for development of novel algorithms for co-registration, fusion and time-course signal analysis.

Return To Program's Home Page

Thursday, September 20, 2001

Southwestern In Vivo Cellular and Molecular Imaging Program