Fig 1.
Numerical details about the analytical solutions .
a) Region in the parameter space (cosθ1, cosθ2, E1,0) in which both solutions satisfy the physical constrain 0 < E1 < 1. b) Region in the parameter space (cosθ1, cosθ2, E1,0) in which solution
applies. c) Region in the parameter space (θ1/θ2, θ2, E1,0) in which solution
applies. d) Family of curves parametrized by E1,0 (ranging from 0 [black] to 1 [light gray] in step of 0.1) that pose an upper limit to θ1/θ2 for
to be acceptable. The red curve corresponds to E1,0 = e−1 and entirely lies in the half-plane θ1/θ2 > 0.47.
Fig 2.
Optimal Flip Angles (FAs) for R1 mapping.
FAs and
of single- and double-GRE sequences that maximize the accuracy of R1 estimate as a function of the expected TR,0 ⋅ R1.
Fig 3.
Ratio between the SNR of the estimated R1 map and the SNR of the S images in an average brain parenchyma (with T1,par = 1 s) as a function of the expected T1, for θ1 = 2° and θ2 = 20°.
Fig 4.
Axial RESUME maps of the brain.
Slices at the level of midbrain (a), basal ganglia (b) and centra semiovalia (c) of: 1) map ([0 ∼ 50] s−1); 2) R1 map ([0 ∼ 2] s−1); 3) PD map ([0 ∼ 1] arbitrary units); 4) QSM ([−300 ∼ 300] ⋅ 10−9).
Fig 5.
Multi-planar reconstructions of the RESUME maps.
Axial (a), coronal (b) and sagittal (c) brain slices of: 1) map ([0 ∼ 50] s−1); 2) R1 map ([0 ∼ 2] s−1); 3) PD map ([0 ∼ 1] arbitrary units); 4) QSM ([−300 ∼ 300] ⋅ 10−9) maps. The axial and coronal slices are centered at the level of the cerebellar dentate nuclei (remarkably conspicuous in the
and QSM maps due to their increased iron content); the sagittal slice is centered on the midline, where the deoxyhemoglobin-rich intracranial veins are particularly evident in the
and QSM maps.
Fig 6.
Reproducibility and accuracy of the RESUME maps.
From left to right: a) RESUME maps of the spatial reference (HC1); b) voxelwise average, in the pool of subjects, of the maps spatially normalized to HC1; c) qMRI reference maps of HC1 obtained following [16] (for
, R1 and PD maps) and from the 8-echo GRE (for the QSM); d) voxelwise average, in the pool of subjects, of the
maps spatially normalized to HC1. From top to bottom: 1)
map ([0 ∼ 50] s−1); 2) R1 map ([0 ∼ 2] s−1); 3) PD map ([0 ∼ 1] arbitrary units); 4) QSM ([−300 ∼ 300] ⋅ 10−9). Please note that the averages of
(b) and
(d) are shown with an enhancement factor of 20.
Table 1.
Mean and standard deviation of the qMRI values in different brain regions, measured by RESUME or by the reference method [16] (Ref) in the healthy controls or found in the literature.
PD values are normalized to the cerebrospinal fluid (CSF). Relaxometry values are compared with other 3D measures at 3 T.
Fig 7.
Extended qMRI processing of the RESUME protocol.
Axial brain slices of: 1) map ([0 ∼ 50] s−1); 2) R1 map ([0 ∼ 2] s−1); 3) PD map ([0 ∼ 1] arbitrary units); 4) QSM ([−300 ∼ 300] ⋅ 10−9); 5) MAVEN vein segmentation; 6) OEF map ([0 ∼ 0.4]); 7) MTVF map ([0 ∼ 0.5]). MAVEN segmentation is Maximum-Intensity-Projected on a slab thickness of 20 mm and fused on the minimum-Intensity-Projection of the corresponding SWI slab (computed from the second echo of the dual-echo GRE), which serves as anatomical reference.