Image-based MRI gradient estimation

Geometric distortion characterization and correction for the 1.0 T Australian MRI‐linac system using an inverse electromagnetic method

“…However, the fifth‐order SH may not work optimally for the area at the edges and/or outside of the DSV . In the Australian MRI‐Linac system, as the scanner has a smaller DSV (30‐cm DSV, ±4.05 ppm) than commercial whole‐body MRI scanners (50‐cm DSV, 1 ppm) because of the split structure, the GNL‐induced distortions for the target anatomy at the periphery or outside the DSV need to be resolved …”

“…16 However, the fifth-order SH may not work optimally for the area at the edges and/or outside of the DSV. 19,20 In the Australian MRI-Linac system, as the scanner has a smaller DSV (30-cm DSV, AE4.05 ppm) than commercial whole-body MRI scanners 21 (50-cm DSV, 1 ppm) because of the split structure, the GNL-induced distortions for the target anatomy at the periphery or outside the DSV need to be resolved. 22 In this study, we combined the phantom measurement 17 and a designated inverse electromagnetic (EM) method [23][24][25] to characterize the gradient field both inside and outside of the DSV for the Australian MRI-Linac system.…”

A fully automated method for accurate measurement of geometrical distortion in magnetic resonance imaging of a 3D-lattice phantom

“…Once the apparent positions of fiducial points and their adjacency relationship is known, it becomes possible to apply the algorithm reported in [1] for estimation of Spherical Harmonic Coefficients. Note that the estimated SHC sets provide a valid field expansion within the volume enclosed by the envelope of the fiducial points whose positions have been determined, in particular also outside of the homogeneity sphere, as long as the corresponding fiducial points are observable and distinguishable.…”

“…The refined position and its related unit vectors are stored into a list L of nodes to be visited, while the refined position is output to a 3D structure C, filled with fiducial point apparent position estimates. This structure contains at address (row, column, slice) the estimate of the apparent position of the corresponding node in images and will enable fiducial point position estimation and processing methods like the one we described in [1]. Note that the address (row, column, slice) of a refined node position estimate in C defines the adjacency relationship between nodes.…”

“…On the basis of the fact that the fields vary continuously in space, and that the signal loss usually happens at much larger radii than the reference radius, a recent paper by Aquaviva et al [1] describes how to obtain a set of SHC's valid outside of the homogeneity sphere, as long as the distortion is observable and no aliasing effects occur. The method is based on reliable estimates of the positions of a large number (e.g.…”