The investigators propose to test the feasibility of developing a distributed system for computer aided diagnosis of mammographic lesions, as further described by their abstract: "The availability of high resolution film digitizers and the promise of new digital detector technologies have led to the development of CAD (computer-aided diagnostic) tools to assist the radiologist in early detection and characterization of breast lesions. CAD tools have been found to reduce misdiagnosis and have attracted significant commercial interest. "The goal of the proposed work is to develop a mechanism to provide standardized, consistent, and state-of-the-art computer-aided diagnostic (CAD) processing for all digital mammograms within a single hospital system. The development of this system will: (a) enable CAD processing that is independent of the mammographic units or computer systems employed at the individual breast imaging clinics within the system, (b) allow digitized mammograms to be worked up in satellite clinics using exactly the same procedures as those in the main institution but without the need for significant local computing resources, and (c) reduce the cost of providing CAD tools to satellite clinics. "Phase I will focus on developing an environment for CAD processing to be accessed on any computing platform, determining how to allocate network computing resources to allow CAD processing in resource-limited situations, and implementing a software distribution network to ensure all clinics access to the same most updated CAD processing tools. "Phase I will develop a proof-of-principle CAD delivery system that will be deployed at the University of Michigan Department of Radiology and one of its satellite clinics. Phase II will significantly extend this concept to operate in an Internet environment (as opposed to the simple intranet in this application). PROPOSED COMMERCIAL APPLICATION: This project will develop a low-cost solution for implementing a distributed system for computer aided diagnosis of mammographic lesions. Potential commercial applications may include a CAD software package that can be deployed in a medium-size hospital and other software by- products that enable more efficient computing resource distribution and allocation.
