Huanzhou Yu1, Rebecca Fahrig, Norbert J Pelc. 1. Department of Electrical Engineering, Lucas MRS Imaging Center, Stanford University, 1201 Welch Road, Rm. P064, Stanford, CA 94305, USA. hzhyu@stanford.edu
Abstract
PURPOSE: To validate one possible function of a real-time x-ray/MR (XMR) interface in a hybrid XMR system using x-ray images as "scouts" to prescribe the MR slices. MATERIALS AND METHODS: The registration process consists of two steps: 1) calibration, in which the system's geometric parameters are found from fiducial-based registration; and 2) application, in which the x-ray image of a target structure and the estimated geometric parameters are used to prescribe an MR slice to observe the target structure. Errors from the noise in the location of the fiducial markers, and MR gradient nonlinearity were studied. Computer simulations were used to provide guidelines for fiducial marker placement and tolerable error estimation. A least-squares-based correction method was developed to reduce errors from gradient nonlinearity. RESULTS: In simulations with both sources of errors and the correction for gradient nonlinearity, the use of 16 fiducial markers yielded a mean error of about 0.4 mm over a 7200 cm(3) volume. Phantom scans showed that the prescribed target slice hit most of the target line, and that the length visualized was improved with the least-squares correction. CONCLUSION: The use of 16 fiducial markers to co-register XMR FOVs can offer satisfactory accuracy in both simulations and experiments. (c) 2005 Wiley-Liss, Inc.
PURPOSE: To validate one possible function of a real-time x-ray/MR (XMR) interface in a hybrid XMR system using x-ray images as "scouts" to prescribe the MR slices. MATERIALS AND METHODS: The registration process consists of two steps: 1) calibration, in which the system's geometric parameters are found from fiducial-based registration; and 2) application, in which the x-ray image of a target structure and the estimated geometric parameters are used to prescribe an MR slice to observe the target structure. Errors from the noise in the location of the fiducial markers, and MR gradient nonlinearity were studied. Computer simulations were used to provide guidelines for fiducial marker placement and tolerable error estimation. A least-squares-based correction method was developed to reduce errors from gradient nonlinearity. RESULTS: In simulations with both sources of errors and the correction for gradient nonlinearity, the use of 16 fiducial markers yielded a mean error of about 0.4 mm over a 7200 cm(3) volume. Phantom scans showed that the prescribed target slice hit most of the target line, and that the length visualized was improved with the least-squares correction. CONCLUSION: The use of 16 fiducial markers to co-register XMR FOVs can offer satisfactory accuracy in both simulations and experiments. (c) 2005 Wiley-Liss, Inc.
Authors: Aruna V Arujuna; R James Housden; Yingliang Ma; Ronak Rajani; Gang Gao; Niels Nijhof; Pascal Cathier; Roland Bullens; Geert Gijsbers; Victoria Parish; Stamatis Kapetanakis; Jane Hancock; C Aldo Rinaldi; Michael Cooklin; Jaswinder Gill; Martyn Thomas; Mark D O'neill; Reza Razavi; Kawal S Rhode Journal: IEEE J Transl Eng Health Med Date: 2014-01-30 Impact factor: 3.316