Computer-Aided Detection: Cost Effectiveness Analysis with Learning Model

Abstract

Computer-aided detection (CAD) has been a promising research area over the last two decades in the medical field. CAD usually supports doctors by marking medical images with potential lesions. This technology has performed many remarkable accomplishments in the laboratory and has been used for some medical cases (e.g., breast cancer and colorectal cancer). However, its health-economic impact (i.e., costeffectiveness) from a societal perspective remains controversial because of the lack of studies. To clarify the cost-effectiveness of CAD, we propose a new methodology, the "learning model," and conduct a simulation based on data reported in previous studies. This model can predict performance improvement due to the accumulation of medical imaging data and the proficiency of CAD for doctors. In this article, we demonstrate the feasibility of our proposed methodology in mammography and computed tomographic colonography CAD. Furthermore, based on the analyzed results, we discuss the potential of CAD.