Tyler Chlebowski '16, Mechanical Engineering

In this study, we compared the capabilities of five programs for the segmentation, 3D reconstruction and 3D printing of whole-heart tissue and blood volumes. We based the segmentations on CT images obtained for diagnosis of pediatric congenital heart disease at Geisinger Medical Center. Creation of three-dimensional printed models from serial medical images is becoming more common for diagnosis, operative planning and teaching. The crucial image analysis task required is the segmentation of the tissue of interest from the surrounding tissue. Various software programs and algorithms are available to accomplish this segmentation, which can lead to confusion as to their relative merits. In addition, typical pediatric heart scans often have poor definition of walls and chambers due to the need to limit the use of contrast agent and radiation exposure in this population. Our project was successful in developing metrics and documenting the capabilities and accuracy of the segmentation and reconstruction algorithms. In addition, we showed that the 3D reconstructions can be done much faster than previously thought, which opens more opportunities for use in diagnosis and treatment.

Advisor: Professor Christine Buffinton