Prior Knowledge in 3D Reconstruction from 2D X-Rays

Johns Hopkins University, Baltimore MD

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): Primarily due to an aging population, the number of orthopedic surgeries performed each year is steadily increasing. Reduction and fixation of fractures due to osteoporosis is now one of the most common surgical procedures performed in the United States. As well, the number of joint replacement surgeries has increased so markedly over the past decade or so that they are now considered relatively routine. As our population continues to age, the ability of our older population to maintain levels of activity associated with satisfactory quality of life as well as to prevent long-term disability and extensive rehabilitations depends on the continued improvement of these orthopaedic surgeries. Minimally invasive surgical techniques, requiring only small incisions, have tremendous potential to decrease operative time, improve patient safety, speed recovery, and yield improved outcomes. Most joint reconstructions and fracture repairs are presently performed entirely without intraoperative imaging except to verify the final result. In minimally invasive surgeries, however, where the opportunity for direct examination is very limited, there is an increased need for intraoperative imaging to verify fracture repair or joint reduction before fixation in order to avoid prolongation of the surgery or the need for a revision surgery at a later time. The goals of the proposed research are to develop novel algorithms for the three-dimensional reconstruction of bone images from mobile C-arm x-ray images and to validate these methods in a clinically realistic setting. The primary novelty of the proposed methods is their exploitation of prior knowledge in the form of prior computed tomography scans when available and a statistical atlas when comPuted tomography scans are not available. Specifically, we propose to: 1) Develop methods to register both prior computed tomography data and a statistical atlas to measured C-arm projections; 2) to develop methods to fuse prior information with observed information for optimal three-dimensional reconstruction; and 3) to build a statistical atlas and validate methods on phantoms and cadavers. The proposed methods will apply immediately on calibrated C-arms that are prevalent in the operating room, and their impact on surgical practice in orthopedics has the potential to be very high.