In Phase I we developed a method for automated border detection (ABD) of echocardiographic scans that is feasible for clinical application. The accuracy of our processes provides exceeds Phase I goals and is comparable to interobserver variability in measuring volume and ejection fraction, and in border location. For a diverse set of patients, we have achieved an accuracy of 10 ml for endocardial volume, 4% for ejection fraction, and ,2.0 mm for border position. Our processes operates in 4 min. In Phase II we propose to continue research and development to move our ABD technology closer to clinical user. Our first specific aim is to reduce the amount of manual input required even further.
Our second aim i s to develop a prototype system suitable for clinical evaluation.
Our third aim i s to perform a pilot trial to evaluate the performance of our ABD process, as a preparation for a more formal, multi-center clinical trial planned for Phase III. The proposed research is important because quantitative 3D echo provides greater accuracy and reproducibility and more comprehensive information on cardiac status than currently available imaging techniques. The significant advantages of 3D echo are not currently available for clinical practice because it is impractical without automation.
Automation of echocardiogram border detection enables physicians to obtain accurate, reproducible and comprehensive measurements of the heart's size, shape and function. This technology can be included in ultrasound systems or provided in workstations. The core technology can be applied to other organs and other imaging modalities.
