Abstract
In the past measurement of deformity correction in spinal surgery has been done using measurements made directly from radiographs using a pencil, ruler and goniometer The aim of this paper is to describe a reproducible, accurate and partially automated system that has been developed for measuring x-rays of patients with spinal disorders.
Computer assisted measurement of polyethylene wear in patients with THJR is now well established. Many of the image processing algorithms have been modified to allow identification of the outline of both thoracic and lumbar vertebral bodies on digital images of radiographs made from patients with spinal disorders. The Genetic Algorithm (GA), a branch of Artificial Intelligence, has been adapted to allow the modelling of a four sided figure to each vertebral body, with minimal user input.
The accurate identification of each vertebral body within a spinal radiograph allows measurement of multiple parameters, including Cobb angles, vertebral width, vertebral height and cross sectional area, as well as measurement of average disc height and cross sectional area. The method is 100% reproducible for each digital image. An attempt to measure accuracy has not been made because these are two dimensional measurements of a three-dimensional structure.
Comparison of these measurements between pre and post-operative radiographs for a patient allows accurate and reproducible measurement of reconstructive surgery for scoliosis and other spinal disorders. It may aid in development of a classification system for scoliosis.
The abstracts were prepared by Editorial Secretary Jean-Claude Theis. Correspondence should be addressed to NZOA at Department of Orthopaedic Surgery, Dunedin Hospital, Private Bag 1921, Dunedin, New Zealand.
