Zikuan Chen1, Vince Calhoun2. 1. The Mind Research Network and LBERI, Albuquerque, NM 87106, United States. Electronic address: zchen@mrn.org. 2. The Mind Research Network and LBERI, Albuquerque, NM 87106, United States; University of New Mexico, ECE Department, Albuquerque, NM 87131, United States.
Abstract
BACKGROUND: By solving an inverse problem of T2*-weighted magnetic resonance imaging for a dynamic fMRI study, we reconstruct a 4D magnetic susceptibility source (χ) data space for intrinsic functional mapping. NEW METHODS: A 4D phase dataset is calculated from a 4D complex fMRI dataset. The background field and phase wrapping effect are removed by a Laplacian technique. A 3D χ source map is reconstructed from a 3D phase image by a computed inverse MRI (CIMRI) scheme. A 4D χ data space is reconstructed by repeating the 3D χ source reconstruction for each time point. A functional map is calculated by a temporal correlation between voxel signals in the 4D χ space and the timecourse of the task paradigm. RESULTS: With a finger-tapping experiment, we obtain two 3D functional mappings in the 4D magnitude data space and in the reconstructed 4D χ data space. We find that the χ-based functional mapping reveals co-occurrence of bidirectional responses in a 3D activation map that is different from the conventional magnitude-based mapping. COMPARISON WITH EXISTING METHODS: The χ-based functional mapping can also be achieved by a 3D deconvolution of a phase activation map. Based on a subject experimental comparison, we show that the 4D χ tomography method could produce a similar χ activation map as obtained by the 3D deconvolution method. CONCLUSION: By removing the dipole effect and other fMRI technological contaminations, 4D χ tomography provides a 4D χ data space that allows a more direct and truthful functional mapping of a brain activity. Published by Elsevier B.V.
BACKGROUND: By solving an inverse problem of T2*-weighted magnetic resonance imaging for a dynamic fMRI study, we reconstruct a 4D magnetic susceptibility source (χ) data space for intrinsic functional mapping. NEW METHODS: A 4D phase dataset is calculated from a 4D complex fMRI dataset. The background field and phase wrapping effect are removed by a Laplacian technique. A 3D χ source map is reconstructed from a 3D phase image by a computed inverse MRI (CIMRI) scheme. A 4D χ data space is reconstructed by repeating the 3D χ source reconstruction for each time point. A functional map is calculated by a temporal correlation between voxel signals in the 4D χ space and the timecourse of the task paradigm. RESULTS: With a finger-tapping experiment, we obtain two 3D functional mappings in the 4D magnitude data space and in the reconstructed 4D χ data space. We find that the χ-based functional mapping reveals co-occurrence of bidirectional responses in a 3D activation map that is different from the conventional magnitude-based mapping. COMPARISON WITH EXISTING METHODS: The χ-based functional mapping can also be achieved by a 3D deconvolution of a phase activation map. Based on a subject experimental comparison, we show that the 4D χ tomography method could produce a similar χ activation map as obtained by the 3D deconvolution method. CONCLUSION: By removing the dipole effect and other fMRI technological contaminations, 4D χ tomography provides a 4D χ data space that allows a more direct and truthful functional mapping of a brain activity. Published by Elsevier B.V.
Entities:
Keywords:
4D magnetic susceptibility (χ) tomography; Blood oxygenation level-dependent functional magnetic resonance imaging (BOLD fMRI); Dipole inversion; Intrinsic functional brain mapping; T2* imaging; T2* magnitude and phase images
Authors: Julio Acosta-Cabronero; Carlos Milovic; Hendrik Mattern; Cristian Tejos; Oliver Speck; Martina F Callaghan Journal: Neuroimage Date: 2018-07-31 Impact factor: 6.556