Non-invasive Face Registration for Surgical Navigation

초록

This study aims to propose a precise rigid-body face registration method that does not require an invasive marker attachment for surgical navigation devices. The non-invasive face registration involved the following steps: anatomical feature points such as the eyes and nose are found from the computed tomography (CT) image and the location tracking device attached to the patient's forehead prior to the procedure; an initial registration is attempted on the two previously extracted feature points; secondary registration was performed using a coherent point drift (CPD) algorithm; and using the Powell method, precise registration was performed to minimize the closest point registration error (CPRE). The CPRE co-registration accuracy was measured for the original and sampled phantom data, and the error was found to be 1.07 mm on average for skin coordinate data. The method introduced in this study enabled precise rigid-body face registration in a non-invasive manner using only skin coordinates. © (2022) The Korean Institute of Information Scientists and Engineers.

키워드