Meeting Banner
Abstract #3217

Practical Aspects of Correcting ADC Bias Due to Gradient Nonlinearity in Media of Arbitrary Anisotropy

Dariya I. Malyarenko1, Brian D. Ross1, Thomas L. Chenevert1

1Radiology - MRI, University of Michigan, Ann Arbor, MI, United States

Relatively large ADC bias error, well exceeding measurement noise, can be observed for anatomical regions imaged distant from magnet isocenter. Gradient nonlinearity is a main source of bias. Our previously-described approach allowed substantial reduction of spatial ADC bias for media of arbitrary anisotropy using DWI encoding along any three orthogonal directions. This work focuses on practical aspects of ADC correction implementation via system nonlinearity tensor for derivation of correctors applicable to DWI or b-map generation. Correction performance was evaluated for isotropic and anisotropic media.

Keywords

according account achieved anatomy anisotropic anisotropy application applied arbitrary arbor aspects axes bias blue brain bulk calculation caused characteristic clinical close coil confounds consistent contract correct corrected correcting correction corrector correctors degree dependence dependent derivation described deviation differentiated diffusion diffusivity dimensions effectively efficiency employ encoding error every expansion expected experimental extreme fields framework generate generated gradient gradients grant green grid harmonic header health highly histograms identical illustrated implementation institutes intensities intensity inter interpolation isotopic isotropic known laboratory location locations maps marker measured media median medium model modeled national near noise nonlinearity numerically offset offsets onto original orthogonal performance phantom pixel pixels platform practical predicted previously primarily produce produced projecting projection proposed quantitative radiology reduce reduced regular removed sampled scale scaled scanner scenarios slices spacing spatial spherical square studies sufficient superior superiorly support system tensor tissue trace true unbiased unidirectional vendor versus vicinity volume water yield