Enum InversionAlgorithm 

pub enum InversionAlgorithm {
    Tkd,
    Tsvd,
    Tikhonov,
    Tv,
    Rts,
    Nltv,
    Medi,
    Tfi,
    Ilsqr,
    Tgv,
    Qsmart,
    Ndi,
    Fansi,
    FansiTgv,
    L1qsm,
    Whqsm,
    Hdqsm,
    AmpPe,
    Xqsm,
    Qsmnet,
    QsmnetPlus,
    Autoqsm,
    Qsmgan,
    Ir2qsm,
    Lpcnn,
    ModlQsm,
    Nextqsm,
    Iqsm,
    IqsmPlus,
}

Dipole inversion algorithm

Variants

Tkd

Tsvd

Tikhonov

Tv

Rts

Nltv

Medi

Tfi

Preconditioned Total Field Inversion (single-step, total field).

Ilsqr

Tgv

Qsmart

Ndi

Nonlinear Dipole Inversion (FANSI ndi.m).

Fansi

FANSI nonlinear Total Variation (nlTV).

FansiTgv

FANSI nonlinear Total Generalized Variation (nlTGV).

L1qsm

L1 data-fidelity QSM (FANSI nlL1TV / PI-QSM).

Whqsm

Weak-Harmonic QSM (FANSI WH_nlTV).

Hdqsm

Hybrid two-stage L1→L2 QSM (HD-QSM).

AmpPe

Approximate Message Passing with built-in Parameter Estimation (AMP-PE).

Xqsm

xQSM deep-learning dipole inversion (requires the onnx feature and the xqsm model weights; see crate::models).

Qsmnet

QSMnet deep-learning dipole inversion (requires the onnx feature and the qsmnet model weights; see crate::models).

QsmnetPlus

QSMnet+ deep-learning dipole inversion (susceptibility-scaling augmented; requires the onnx feature and the qsmnet-plus weights).

Autoqsm

AutoQSM single-step reconstruction (requires the onnx feature and the autoqsm weights). NOTE: takes the total field — it does its own background removal, so the local_field_ppm argument should be the total field.

Qsmgan

QSMGAN deep-learning dipole inversion (local field → χ; onnx + qsmgan weights).

Ir2qsm

IR2QSM unrolled deep-learning dipole inversion (local field → χ; onnx + ir2qsm weights).

Lpcnn

LPCNN learned-proximal deep-learning dipole inversion (local field → χ; onnx + lpcnn weights).

ModlQsm

MoDL-QSM model-based deep-learning dipole inversion (local field → χ33/STI component; onnx + modl-qsm weights).

Nextqsm

NeXtQSM single-step reconstruction (requires onnx + the two nextqsm weight files). NOTE: takes the total field — it does its own background removal.

Iqsm

iQSM single-step reconstruction from wrapped phase (end-to-end: joint unwrap + background removal + inversion). Not a dipole-inversion-stage option — use super::run_iqsm. Rejected by run_dipole_inversion.

IqsmPlus

iQSM+ single-step reconstruction from wrapped phase with orientation-adaptive feature editing (uses the B0 direction). End-to-end; use super::run_iqsm_plus.

Implementations

impl InversionAlgorithm

pub fn orientation_support(&self) -> OrientationSupport

How this method behaves when B0 is not along voxel +z.

Every classical inversion builds its dipole kernel from the supplied direction, directly or through the shared ADMM/FANSI spectral setup, so all of them handle oblique data on the acquired grid. The deep-learning methods take no direction at all.

impl InversionAlgorithm

pub const VARIANTS: &'static [Self]

pub fn dl_model_id(self) -> Option<&'static str>

Model-registry id for deep-learning variants, else None (classical methods).

Trait Implementations

impl Clone for InversionAlgorithm

fn clone(&self) -> InversionAlgorithm

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for InversionAlgorithm

impl Debug for InversionAlgorithm

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

Formats the value using the given formatter.

impl PartialEq for InversionAlgorithm

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for InversionAlgorithm

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for InversionAlgorithm