Fig 1.
Schematic flowchart of the implemented algorithm.
From FC-FLASH phase data the χ-map is directly derived (see §2.6). The TrueFISP collection is used to extract the band-free S0 dataset (see §2.2). Then, FC-FLASH and S0 series are denoised by the SVN-MNLM scheme (see §3.3). The -map is computed from resulting FC-FLASH series (see §2.4). The introduction of an estimate of the B1-map (actually measured or simply guessed) allows for the extraction of the R1- (see §2.3), PD- (see §2.5) and R2- (see §2.7) maps. Depending on the reliability of the B1-map estimate, a B1-correction scheme may be iteratively applied (dashed lines—see §3.4).
Table 1.
Design guidelines for minimal acquisition protocols (second column) making it possible to reconstruct each parameter listed in the first column.
The corresponding total acquisition time and the list of other maps that can be obtained for free are reported in the third and fourth columns, respectively.
Fig 2.
Axial brain slices at the 4 different levels showing the position of ROIs drawn for measurement of mean relaxometry values in selected brain structures: cortex, white matter and CSF (a); head of caudate and thalamus (b); globus pallidus and putamen (c); red nucleus and substantia nigra (d).
Table 2.
Percentual FWHM of the R1 and R2 distribution obtained without (pre) and with (post) -inhomogeneity correction.
For each case, we report the values for the entire phantom and for its eroded version (defined by a spherical structuring element with radius of 1 cm) simulating the brain without the skull.
Fig 3.
Axial (left), coronal (middle) and sagittal (right) brain slices of R1 ([0 ∼ 2] s-1), R2 ([0 ∼ 50] s-1), ([0 ∼ 50] s-1), PD ([0 ∼ 1] arbitrary units) and χ ([−300 ∼ 300] ppb) maps (from top to bottom) in a 43-year-old female HC.
As coronal and sagittal images are derived by a multi-planar reconstruction of the original axial dataset, their in-plane resolution is 0.65 × 1.3 mm2, with a slice thickness of 0.65 mm. Small insets pointing out the position of the slices on a perpendicular plane are shown as anatomical reference in the upper row.
Fig 4.
R1 ([0 ∼ 2] s-1), R2 ([0 ∼ 50] s-1), ([0 ∼ 50] s-1), PD ([0 ∼ 1] arbitrary units) and χ ([−300 ∼ 300] ppb) (from top to bottom) maps (left) and corresponding confidence interval (right) in a 43-year-old female HC, displayed with a markedly different (scale factor of 20) grayscale to highlight tiny differences in measure uncertainties.
A small inset pointing out the position of the slices on a sagittal plane is shown as anatomical reference in the upper row.
Fig 5.
Axial brain slices at the level of midbrain (left), basal ganglia (middle) and fronto-parietal convexity (right) of R1 ([0 ∼ 2] s-1), R2 ([0 ∼ 50] s-1), ([0 ∼ 50] s-1), PD ([0 ∼ 1] arbitrary units) and χ ([−300 ∼ 300] ppb) maps (from top to bottom) in a 43-year-old female HC.
Small insets pointing out the position of the slices on a sagittal plane are shown as anatomical reference in the upper row.
Fig 6.
R1 (a), FLAIR (b), R2 (c), PD (d), (e) and χ (f) axial brain slices at the level of the lateral ventricles in a 45-year-old female MS patient (Expanded Disability Status Scale—EDSS: 3.5; disease duration: 7 years).
A small inset pointing out the position of the slices on a sagittal plane is shown as anatomical reference in the upper row.
Table 3.
Relaxometry and susceptibility properties measured in selected brain locations.
In each cell, ROI mean and standard deviation are reported.