[BUG] Confusion on reference / reference to

[ClementCaporal]

Hello,

I hope this issue finds you well. Thank you for the great tool!

I am not sure this can be considered as a bug, feel free to update the issue label.

Describe the bug 1
The description for the command line and the napari plugin differ for the smoothing_sigma_reference, smoothing_sigma_floating, comparing the file

brainreg/brainreg/core/backend/niftyreg/parser.py

Lines 85 to 104 in 7b7b883

	niftyreg_opt_parser.add_argument(
	"--smoothing-sigma-reference",
	dest="smoothing_sigma_reference",
	type=float,
	default=-1.0,
	help="Adds a Gaussian smoothing to the reference (the one being "
	"registered to) image, with the sigma defined by the number. "
	"Positive values are interpreted as real values in mm, "
	"negative values are interpreted as distance in voxels.",
	)
	niftyreg_opt_parser.add_argument(
	"--smoothing-sigma-floating",
	dest="smoothing_sigma_floating",
	type=float,
	default=-1.0,
	help="Adds a Gaussian smoothing to the floating image (the one being "
	"registered), with the sigma defined by the number. Positive "
	"values are interpreted as real values in mm, negative values "
	"are interpreted as distance in voxels.",
	)

and

brainreg/brainreg/napari/register.py

Lines 334 to 343 in 7b7b883

	smoothing_sigma_reference: int
	Adds a Gaussian smoothing to the reference image (the one being
	registered), with the sigma defined by the number. Positive
	values are interpreted as real values in mm, negative values
	are interpreted as distance in voxels.
	smoothing_sigma_floating: float
	Adds a Gaussian smoothing to the floating image (the one being
	registered), with the sigma defined by the number. Positive
	values are interpreted as real values in mm, negative values
	are interpreted as distance in voxels.

"the one being registered" vs "the one being registered to"

---

Describe the bug 2

I think I am also confused about the correspondence between floating / reference vs atlas / image in the napari plugin

brainreg/brainreg/napari/register.py

Lines 206 to 223 in 7b7b883

	smoothing_sigma_reference=dict(
	value=DEFAULT_PARAMETERS["smoothing_sigma_reference"],
	label="Smoothing sigma (image)",
	min=-99.0,
	),
	smoothing_sigma_floating=dict(
	value=DEFAULT_PARAMETERS["smoothing_sigma_floating"],
	label="Smoothing sigma (atlas)",
	min=-99.0,
	),
	histogram_n_bins_floating=dict(
	value=DEFAULT_PARAMETERS["histogram_n_bins_floating"],
	label="Histogram bins (atlas)",
	),
	histogram_n_bins_reference=dict(
	value=DEFAULT_PARAMETERS["histogram_n_bins_reference"],
	label="Histogram bins (image)",
	),

Intuitively I would match:

• floating -> image
• reference -> atlas

Maybe I misunderstood something?

Thank you for your time,

Clément

[adamltyson]

Thanks for raising this, I agree we should make this clearer. It's complicated by the fact that that within brainreg, the registration occurs in both directions, i.e. atlas to sample and sample to atlas.

[alessandrofelder]

As @IgorTatarnikov suggested, maybe we can move registration nomenclature in the docs to always use "sample" and "template"