Struct FieldMappingConfig 

pub struct FieldMappingConfig {
    pub unwrapping_algorithm: UnwrappingAlgorithm,
    pub phase_offset_removal: bool,
    pub phase_offset_sigma_mm: [f64; 3],
    pub bipolar_correction: bool,
    pub b0_estimation: B0EstimationMethod,
    pub laplacian_solver: LaplacianSolver,
    pub laplacian_unmasked: bool,
    pub romeo_params: RomeoParams,
    pub prelude_params: PreludeParams,
}

Field mapping configuration

Fields

unwrapping_algorithm: UnwrappingAlgorithm

phase_offset_removal: bool

phase_offset_sigma_mm: [f64; 3]

Gaussian smoothing sigma of the MCPC-3D-S phase offsets, in mm per axis (default [7, 7, 7]); divided by the voxel size to give the smoothing in voxels (Self::phase_offset_sigma_voxels). The default is the ROMEO command line’s (what users run, romeo -B): --phase-offset-smoothing-sigma-mm, “default: [7,7,7]”, divided by the NIfTI voxel size (ROMEO.jl 08776d4, ext/RomeoApp/argparse.jl l. 79-84 and 217-224, caller.jl l. 137-138). The same physical width at every resolution. MriResearchTools’ mcpc3ds function itself defaults to sigma = [10, 10, 5] in voxels (src/mcpc3ds.jl l. 39), which QSM.rs used before; that is 10/10/5 mm at 1 mm voxels but a different width at any other resolution.

Measured on 1 mm data (ROMEO field, default V-SHARP, RTS), where the old default is 10/10/5 mm: χ moves by at most 0.003 XSIM, on the qsm-forward phantom XSIM 0.254 → 0.256, NRMSE 85.9 → 85.7 %, and on an in-vivo 3T Philips scan, agreement with the UK Biobank pipeline’s χ, XSIM 0.714 → 0.714, NRMSE 69.7 → 69.4 %.

bipolar_correction: bool

b0_estimation: B0EstimationMethod

How the unwrapped echoes are combined, with that method’s own settings (default B0EstimationMethod::WeightedAvg with B0WeightType::PhaseSNR).

laplacian_solver: LaplacianSolver

Poisson solver used when unwrapping_algorithm is UnwrappingAlgorithm::Laplacian (default LaplacianSolver::Dct). With LaplacianSolver::Fft { pad: [64; 3] } and b0_estimation B0EstimationMethod::WeightedAvg with B0WeightType::AssumedDecay at 40 ms, a multi-echo run gives UK Biobank’s field map (see run_field_mapping).

laplacian_unmasked: bool

Laplacian unwrapping on the raw phase over the whole volume, instead of on the phase zeroed outside the brain mask (default false). Zeroing first keeps the noise outside the brain out of the global Poisson solve, and is what UK Biobank (MRPhaseUnwrap(mask .* phase)) does. QSM.jl’s unwrap_laplacian takes the raw phase: its :dct solver (this crate’s LaplacianSolver::Dct) ignores the mask, so it is laplacian_unmasked: true here, while its default :mgpcg (not ported) multiplies the Laplacian by the mask and solves with a Dirichlet boundary on it, which also removes the background (see crate::unwrap::laplacian). STI Suite 3.0’s QSM_GUI and QSMART unwrap the raw phase (rules sti.unwrap-unmasked, qsmart.unwrap-unmasked; see crate::unwrap::laplacian). With B0EstimationMethod::QsmartEchofit the unwrapped echoes are also kept over the whole volume, as QSMART fits them; the other combinations use the voxels inside the mask. Only with Laplacian unwrapping.

romeo_params: RomeoParams

ROMEO’s settings (default RomeoParams::default, ROMEO.jl’s API defaults). RomeoParams::romeo_cli gives the romeo command line’s (romeo3 weights with a magnitude, no temporal re-unwrapping), which with the rest of this default config is the CLI’s B0 pipeline. Which of the two QSMxT should default to is being measured (QSM.rs#319).

prelude_params: PreludeParams

PRELUDE’s settings, used when unwrapping_algorithm is UnwrappingAlgorithm::Prelude (default PreludeParams::default: FSL prelude’s defaults, 8 phase partitions with regions labelled slice by slice, and the median wrap count removed per echo as FSL’s output has it).

Implementations

impl FieldMappingConfig

pub fn phase_offset_sigma_voxels(&self, voxel_size: (f64, f64, f64)) -> [f64; 3]

phase_offset_sigma_mm in voxels of voxel_size (mm), as the ROMEO command line converts it.

Trait Implementations

impl Clone for FieldMappingConfig

fn clone(&self) -> FieldMappingConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for FieldMappingConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for FieldMappingConfig

fn default() -> Self

Returns the “default value” for a type.