Enum B0EstimationMethod 

pub enum B0EstimationMethod {
    WeightedAvg {
        weighting: B0WeightType,
    },
    LinearFit(LinearFitParams),
    RobustFit(RobustFitParams),
    ComplexFit(FitPpmComplexParams),
    QsmartEchofit {
        fit_threshold: QsmartFitThreshold,
    },
}

How run_field_mapping combines the unwrapped echoes into one B0 map. Each method carries exactly the settings it reads, so a weighting cannot be set on a fit that ignores it.

Only multi-echo data is combined. A single echo is φ/TE whatever is chosen here, except that Self::RobustFit refuses a series too short for it (see RobustFitParams::min_echoes) and Self::ComplexFit needs the echo’s magnitude (it is also φ/TE there, as in MEDI).

Variants

WeightedAvg

Weighted mean of the per-echo frequencies, B0 = Σ wₑ (φₑ/TEₑ) / Σ wₑ, with wₑ from weighting (default B0WeightType::PhaseSNR). The default method.

Fields

weighting: B0WeightType

Per-echo weight; B0WeightType::AssumedDecay needs a finite, positive T2*.

LinearFit(LinearFitParams)

Magnitude-weighted least-squares fit of phase against TE (multi_echo_linear_fit). The weight is always the magnitude; there is no B0WeightType here.

RobustFit(RobustFitParams)

Motion-robust echo combination (multi_echo_robust_fit): the same magnitude-weighted fit as Self::LinearFit, iteratively reweighted so that an echo which departs from a straight line in TE at a voxel stops pulling the slope there. The weight is magnitude × robust weight; there is no B0WeightType, because the robust scale is measured on magnitude-standardised residuals and another weighting would break that.

Needs at least RobustFitParams::min_echoes echoes (4 with estimate_offset, else 3); fewer is PipelineError::InvalidConfig rather than a silent plain fit. The per-echo report is returned in FieldMappingResult::robust_fit.

ComplexFit(FitPpmComplexParams)

MEDI’s field map: the complex fit of Fit_ppm_complex.m (fit_ppm_complex) to the wrapped echoes, then spatial unwrapping of the fitted phase per echo spacing with unwrapping_algorithm, then division by the echo spacing. MEDI uses its region-growing unwrapPhase (UnwrappingAlgorithm::MediRegionGrowing); the rule complex-fit.unwrap-region-growing offers it as the authors’ setting, never preselected, and ROMEO (given MEDI’s iMag, the root-sum-of-squares of the echo magnitudes) stays the default.

This replaces the per-echo path rather than combining unwrapped echoes: the fit has its own phase offset (p0), so phase_offset_removal does not apply and is not run; bipolar_correction is refused (MEDI’s Fit_ppm_complex_bipolar is not ported). Needs the magnitude of every echo and evenly spaced echoes. Also returns the field’s noise map (FieldMappingResult::field_noise_ppm, MEDI’s N_std).

QsmartEchofit

QSMART’s echo fit (echofit.m, qsmart_echofit): the magnitude-weighted fit of phase against TE through the origin, over every voxel of the unwrapped echoes, with no per-echo constant removed after Laplacian unwrapping (with ROMEO, whole turns between echoes are aligned, as for the weighted mean). Also returns QSMART’s reliability map R_0 (FieldMappingResult::reliability): voxels whose fit residual, box-filtered over ~1 mm, reaches fit_threshold (by default the 99th percentile of the blurred residual, as QSMART’s authors run it; see QsmartFitThreshold). The field is converted to ppm with QSMART’s γ = 2.675e8 rad/s/T.

The echoes come from the configured unwrapping, phase-offset removal and bipolar correction. QSMART unwraps each echo with Laplacian unwrapping on the raw phase (unwrap_phase.m: LaplacianSolver::Mirror, FieldMappingConfig::laplacian_unmasked) and does neither of the others; with those settings this is QSMART’s field map, bit for bit. The rules qsmart.unwrap-laplacian, qsmart.laplacian-solver and qsmart.unwrap-unmasked suggest them for a QSMART run.

Fields

fit_threshold: QsmartFitThreshold

Residual threshold for R_0

Implementations

impl B0EstimationMethod

pub fn id(&self) -> &'static str

Stable id of the method (its settings are not part of it).

Trait Implementations

impl Clone for B0EstimationMethod

fn clone(&self) -> B0EstimationMethod

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for B0EstimationMethod

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

Formats the value using the given formatter.

impl Default for B0EstimationMethod

fn default() -> Self

Returns the “default value” for a type.

impl PartialEq for B0EstimationMethod

fn eq(&self, other: &B0EstimationMethod) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for B0EstimationMethod