Performance of a 2D-3D Image Registration System using (Lossy) Compressed X-ray CT

In 2D-3D Medical Image Registration (MIR), preoperative 3D CT images are registered with intraoperative 2D X-rays images obtained by fluoroscopy or Electronic Portal Imaging Devices (EPID). 2D-3D MIR is established by computing Digitally Reconstructed Radiograph (DRR) images rendered from the volumetric data and is useful in many medical procedures where surgical planning is undertaken using volumetric data prior to treatment, such as in radiation therapy treatment and image guide surgery. The conventional approach used to render DRR images is a computationally expensive process and forms a bottleneck in 2D-3D MIR. This paper evaluates the registration performance of a 2D-3D MIR system which attempts to improve the rendering performance by reducing the number of intersections between the rays cast and internal spaces within the CT volume by using an Octree compression. An Octree compressed CT volume comprises fewer internal spaces, each containing voxels which share similar CT numbers and as such, ray casting through a volume represented as an Octree is potentially computationally simpler. Our work shows that DRR images rendered from an Octree compressed CT volume can be registered with 3 mm accuracy, sufficient for radiotherapy patient setup, even when the volume data is subject to a high degree of compression.