batch

batch ¶

Shared batch fitting utilities.

This module provides ParameterMap creation and image-level fitting helpers used by BaseFitter.fit_image().

create_parameter_maps(model, fitted_params, r2_values, converged, voxel_indices, shape, r2_threshold=0.5, fitting_method='levenberg_marquardt')

Create ParameterMap objects from fitted results.

PARAMETER	DESCRIPTION
`model`	Model that was fit (provides parameter names, units, reference). TYPE: `FittableModel`
`fitted_params`	Fitted parameters, shape `(n_params, n_voxels)`. TYPE: `NDArray`
`r2_values`	R-squared values, shape `(n_voxels,)`. TYPE: `NDArray`
`converged`	Convergence flags, shape `(n_voxels,)`. TYPE: `NDArray`
`voxel_indices`	Original voxel indices, shape `(n_voxels, 3)`. TYPE: `NDArray`
`shape`	Output volume shape `(nx, ny, nz)`. TYPE: `tuple[int, int, int]`
`r2_threshold`	Deprecated, ignored. Kept for signature compatibility; quality mask no longer applies an R² cutoff. TYPE: `float` DEFAULT: `0.5`
`fitting_method`	Name of fitting method for metadata. TYPE: `str` DEFAULT: `'levenberg_marquardt'`

RETURNS	DESCRIPTION
`dict[str, ParameterMap]`	Parameter maps including `"r_squared"`. The attached `quality_mask` simply marks voxels that were fit — callers are expected to filter on r² or parameter values downstream.

create_empty_maps(model, shape)

Create empty parameter maps when no voxels to fit.