Use of Fourier domain filtering and dynamic programming in finding a titanium coil implant in high voltage x-ray images

This paper deals with the problem of finding precise position and orientation of a titanium coil implant in humans. Analysis of high voltage X-rays stereo images are used to determine the true 3D position. High voltage images inherently presents with poor contrast. Various image processing techniques, such as template matching and grey scale morphology operations were tested. A final solution based on filtering in Fourier domain, greyscale morphology and a new shortest path algorithm is implemented as a set of ImageJ macros. The method has been tested on 100 images. The method correctly determined the position of the titanium wire within less than 1 mm of ground truth determined from manual analysis of the images.