Semi-automatic segmentation and surface reconstruction of computed tomography images by using rotoscoping and warping techniques

Abstract

Background: Quick and large-scale segmentation along with three-dimensional (3D) reconstruction is necessary to make precise 3D musculoskeletal models for surface anatomy education, palpation training, medical communication, morphology research, and virtual surgery simulation. However, automatic segmentation of the skin and muscles remain undeveloped. Materials and methods: Therefore, in this study, we developed workflows for semi-automatic segmentation and surface reconstruction, using rotoscoping and warping techniques. Results: The techniques were applied to multi detector computed tomography images, which were optimised to quickly generate surface models of the skin and the anatomical structures underlying the fat tissue. Conclusions: The workflows developed in this study are expected to enable researchers to create segmented images and optimised surface models from any set of serially sectioned images quickly and conveniently. Moreover, these optimised surface models can easily be modified for further application or educational use.