3-D intracoronary ultrasound surface detection using the "thin-plate" model

Reliable detection of the vessel and lumen surfaces in 3-D intracoronary ultrasound images (ICUS) could significantly enhance both the qualitative and quantitative assessment of coronary anatomy. The authors have developed and implemented a surface estimation method using the "thin-plate" model and compared these results to those obtained using a 2-D method in the individual frames of both synthetic and real image data sets. The 2-D and 3-D estimates in each image frame were compared to the known synthetic image contours using the mean square error (MSE). In these test cases, the MSE in 3-D was less than the MSE in the 2-D estimates indicating an improvement in accuracy over the individual 2-D estimates. This surface estimation strategy is accurate, efficient, and will facilitate the development of methods to more effectively visualize and assess coronary anatomy in three dimensions.