The goal of this proposal is to improve the effectiveness of transrectal ultrasound (TRUS) in diagnosing and treating prostate diseases. TRUS is a key imaging technology widely used in diagnosis, treatment and follow-up studies. Identification of prostate boundaries is an important aspect of TRUS imaging; boundaries provide the anatomical description and volume of the prostate needed for disease treatment, for planning treatment procedures and for monitoring the progression of these treatments. Delineation of prostate boundaries in TRUS images is highly operator-dependent because of poor contrast and missing boundaries in parts of TRUS images. Computer-aided delineation of prostate boundaries would provide urologists with a useful technique that will decrease both the variability currently associated with manual outlining and the measurement time, resulting in more accurate, cost-effective and efficient diagnosis and disease treatment. We will develop methods for prostate boundary delineation in TRUS images using curvature-driven image processing techniques and deformable models. To establish feasibility of these methods, in Phase I, we will evaluate these techniques on several clinical cases. By improving the accuracy and reducing the time required for measurement of prostate boundaries in TRUS images, we will be able to enhance the usability and effectiveness of TRUS in diagnosis and treatment planning of prostate diseases.
A software for computer aided boundary delineation of TRUS images has applications in all aspects of prostate disease management, diagnosis, treatment, and follow-up. This tool could be commercialized as a stand- alone software, as an add-on software to exiting radiation treatment planning products, or as a built-in feature of ultrasound image acquisition equipment.
