Open AccessRegion-of-interest cone-beam computed tomography
Author(s) -
S.G. Azevedo,
P. Rizo,
Pierre Grangeat
Publication year - 1995
Language(s) - English
Resource type - Reports
DOI - 10.2172/125412
Subject(s) - region of interest , computer vision , trajectory , projection (relational algebra) , sampling (signal processing) , artificial intelligence , resolution (logic) , radon transform , image resolution , computer science , beam (structure) , mathematics , optics , physics , algorithm , filter (signal processing) , astronomy
A methodology for solving the general cone-beam region-of-interest (ROI) problem on a circular trajectory is presented using the mathematical framework described by Grangeat. The algorithm, called Radon-ROI, takes scans at two different resolutions-low resolution covering the entire object and high resolution covering only the ROI-and combines the scans in both projection and Radon spaces so that the ROI is reconstructed at high resolution without artifacts from missing-data, under-sampling, or cone-beam errors. A circular source trajectory is assumed and the object must have low spatial frequencies outside the ROI. Simulated and experimental results of the Radon-ROI code show marked improvement on resolution within the ROI