The long term objective of this phase is to provide a means for reducing inter- and intra- reader variability in diagnosing interstitial lung diseases in chest radiographs through a computer-based system for analyzing digital images. The Computer-assisted Chest Radiograph Reader System (CARRS) applies recognized principles in the psychophysics of human vision, incorporates neural network-based image analysis and integrates these with a graphical user interface. Advances in digital image processing, and classification techniques have made CARRS feasible for meeting screening, research arid development, and clinical requirements. CARRS will implement the International Labor Organization (ILO) classification procedures.
The specific aims of this project are to implement enhancements to the CARRS prototype developed in Phase I and to validate the advanced version with several hundred chest radiographs.
Today, there exists a need for an automated chest radiograph diagnostic system to screen the thousands of images collected daily at radiological service centers and hospitals worldwide and throughout the United States. A computer-based system that eliminates or reduces inter- and intra-reader variability significantly is required to improve the management and early diagnosis of the disease. CARRS would be marketed and sold to most radiological services centers and hospitals worldwide and throughout the U.S.
