Wednesday, January 16, 2008

Selective Scanning for Faster Prostate Pathology

In new work from the UCSD Jacobs School of Engineering and the UCSD Medical Center, computer vision and machine learning techniques are being used to create an automated system to analyze low magnification images of prostate tissue sections and select those parts of the specimen that should be scanned at high magnification for examination by human diagnosticians. Pathologists hardly ever view entire images at high resolution, but rather concentrate on a few relevant parts. The new system selects the parts that warrant scanning at high magnification.

The very large sizes of image files generated by digital microscopes with a scanning mechanism can make it hard for pathologists at different locations to consult with each other by sharing files through the Internet. The new approach can decrease the space required for storage and time for transmission by a factor of ten.

The new collaborative work from UCSD electrical engineering Ph.D. student Mayank Kabra, computer science professor Yoav Freund and UCSD Medical Center clinical pathology professor Steve Baird will be presented on Thursday 17 January 2008 at the Workshop on Bio-Image Informatics: Biological Imaging, Computer Vision and Data Mining, 2008, hosted by UC Santa Barbara’s Center for Bio-Image Informatics.

Read the full abstract here: Selective Scanning for faster Prostate Pathology
Mayank Kabra ECE, UC San Diego; Yoav Freund CSE, UC San Diego; Steve Baird School of Medicine, UC San Diego