Computer Vision News - September 2021

• varies between modalities, and sometimes even within a single modality. Registration aims to match the FOVs captured in the images by cropping one of the images (or both). • Resolution - a resolution discrepancy usually exists between the per- operative CT image and the intraoperative X-ray fluoroscopy image. Registration aims to reach a pixel-to-pixel transformation by matching the resolutions using either down sampling or up sampling of the images. • Dimensionality - the preoperative CT • is a three-dimensional, high- resolution image, while the intraoperative fluoroscopy is two- dimensional with a lower resolution. Registration aims to match dimensionalities by reducing the three-dimensional image to a two- dimensional image. • Orientation change - the orientation of the different bones and tissues captured changes between the preoperative and the intraoperative imaging. Registration aims to match the orientation to the preoperative plan using an approximation of the movement. 2 Summary 14 Medical Imaging Project