2005 Scientific Session Abstracts
Towards Image Globalization and Sharing with a Refined Knowledge-based Image Retrieval System
Jannie Woo, PhD (wooj@upstate.edu), Robert Schelper, MD, Ph.D. Steve K Landas, MD, Department of Pathology, SUNY Upstate Medical University, Syracuse, NY
Context: Locating suitable images for pathology education is a daunting task. In our department we have the accumulation of glass slides and projection images in departmental archives as well as personal faculty collections, multiple image repositories with inconsistent and largely non-standardized indexing criteria. This process is time consuming and seldom productive. The presentation/retrieval of digitized medical images is an ongoing challenge. Textual retrieval using patient demographics is inadequate since the interpretation of medical images is inherently knowledge-based, necessitating a clinical image report following image examination.
Technology: We developed a Web-based relational database with use of FileMaker Pro to consolidate images from our Department. Images from glass slides and Kodachromes are digitized using a Nikon scanner and touched-up using PhotoShop.
Design: Each image in our database is identified by a unique alphanumeric code and is categorized by image type, diagnosis, case number, organ type, and staining method if microscopic; a brief annotation is provided by a pathology faculty or resident. For image input, a blank Web-based template with a unique alphanumeric code allows input of image-attributes from the contributor; the code becoming the image name once the image is uploaded to the server.
Results: The module is password protected to provide security. After signing on, the user may employ single or multiple search criteria to retrieve general or specific lists of images. Each image may be reviewed with its categorized description and downloaded for PowerPoint lecture use. To date over 3200 high quality images have been incorporated into the module, which has proven invaluable to pathology residents and teaching faculty.
Conclusion: Our system has the potential of consolidating and sharing images across the globe. It is capable of acquiring, indexing, editing and annotating images on a Web-based template, and is able to retrieve images for display from decentralized repositories.