Simultaneous Reconstruction and Tracking for Motion Correction in Image Guided Surgery.

Abstract

Optical tracking system (OTS) localizes the position of rigid surgical tools attached with retro-reflective markers in image-guided surgery (IGS). IGS uses marker-based probes to collect points per frame, or attaches a coordinate reference frame (CRF) on a 3D camera to indirectly measure the surface of patient. However, these measurement methods involve compressive deformation errors from probes or calibration errors from 3D camera to OTS. This paper designs a simultaneous reconstruction and tracking framework for the markered tools and markerless surfaces in the indoor environment without calibration between measurement devices and OTS. A synchronized near-infrared (NIR) digital light projector (DLP) is integrated with OTS, and projects phase-shifting fringe patterns to the target surface. Surgical tools and surface are simultaneously tracked in the same coordinate system of OTS. The reconstructed surface is then registered to CT image space and corrects the intraoperative motion for mixed reality guided surgery. The proposed method is compared with conventional measurement methods using calibrated devices in image-patient registration experiments, and they are evaluated with target registration error (TRE). The TRE of the proposed method is 0.805 $\pm$ 0.152 mm in quantitative phantom experiments and 1.328 $\pm$ 0.318 mm in qualitative cadaveric specimen experiments respectively, achieving the best performance among comparison methods in the phantom experiments. It demonstrates that our approach has the ability to enhance the registration accuracy and efficiency of IGS due to the noncontact and uncalibrated tracking.