The X-Ray machine serves as a crucial auxiliary tool in surgical settings, collaborating with surgical navigation systems to facilitate intra-operative 2D fluoroscopic image navigation and provide a reliable foundation for 2D-3D registration navigation technology. The attainment of precise navigation for both 2D fluoroscopic images and 3D preoperative CT/MRI images hinges upon the efficient and accurate calibration of X-Ray machine. Conventional calibration techniques for X-Ray machine, whether conducted online or offline, are characterized by excessively intricate calibration processes, imposing significant time demands during surgery and resulting in diminished surgical efficiency. We propose a novel permanent calibration ( PerCal ) strategy employing a self-developed X-Ray Source Tracker ( TRKR-S ), Image Plate Tracker ( TRKR-IP ) and Portable Calibrator ( CALP ), aiming to streamline surgical procedures and enhance overall surgical efficiency. This strategy reframes the X-Ray machine calibration challenge into the determination of the spatial position of the X-Ray source in relation to TRKR-S and the 2D fluoroscopic image relative to TRKR-IP . In the intra-operative navigation scenario, solely collecting position information of TRKR-S and TRKR-IP under the optical camera during perspective scanning enables real-time and precise intra-operative 2D fluoroscopic image navigation. Meanwhile, the initialization mechanism for projection coordinate frame ( IM-Proj ) has been proposed to ensure the effective solution of the PerCal problem. Furthermore, the Membership Criterion for 2D/3D marker points ( MC-2D/3D ) has been formulated with the aim of expeditiously and precisely determining the mapping correlation between spatial marker points within the 3D calibration scene and their corresponding marker points on the 2D fluoroscopic image. The efficacy of the proposed PerCal strategy combining IM-Proj and MC-2D/3D is validated through ablation experiments using a G-Type Arm, confirming its calibration accuracy and timeliness. The AMSE errors for frontal and lateral perspectives were denoted as 0.332±0.026mm and 0.53±0.023mm, respectively. Notably, it necessitates only one calibration session to attain permanent reuse of calibration data in spinal surgery navigation. The navigation registration time is confirmed not to exceed 10 seconds, affirming the accuracy and efficiency of PerCal .

With the emergence of Virtual and Mixed Reality (XR) devices, eye tracking has received significant attention in the computer vision community. Eye gaze estimation is a crucial component in XR -- enabling energy efficient rendering, multi-focal displays, and effective interaction with content. In head-mounted XR devices, the eyes are imaged off-axis to avoid blocking the field of view. This leads to increased challenges in inferring eye related quantities and simultaneously provides an opportunity to develop accurate and robust learning based approaches. To this end, we present MagicEyes, the first large scale eye dataset collected using real MR devices with comprehensive ground truth labeling. MagicEyes includes $587$ subjects with $80,000$ images of human-labeled ground truth and over $800,000$ images with gaze target labels. We evaluate several state-of-the-art methods on MagicEyes and also propose a new multi-task EyeNet model designed for detecting the cornea, glints and pupil along with eye segmentation in a single forward pass.