linumpy.geometry.interface#

Tissue interface detection and fitting.

Functions#

`find_tissue_depth`(vol[, zmin, zmax, agarose_intensity])	Detect the tissue interface depth in given volume.
`get_interface_depth_from_mask`(vol)	Compute the interface depths from a 3D tissue mask.
`find_tissue_interface`(vol[, s_xy, s_z, use_log, mask, ...])	Detect the tissue interface.
`find_cutting_plane`(vol, z0map, agarose_mean, agarose_std)	Find the cutting plane using agarose segmentation.
`remove_z0_outliers`(z0map)	Remove outlier interface depths from the z0 map using median absolute deviation.
`fit_interface`(…)	Fit a model on the given interface.
`quadratic_interface`(pos, a, b, c, d, e, f, g, h)	Evaluate a quadratic surface model for the tissue interface.
`get_quadratic_interface`(popt[, volshape])	Compute the tissue interface map from quadratic fit parameters.
`linear_homogeneous_profile`(z, z0, dz, I0, Ib, sigma)	Intensity profile based on a single homogeneous tissue Beer-Lambert model (covered by some amount of water).
`estimate_lh_profile_parameters`(vol[, s])	Estimates the linear-homogeneous intensity profile parameters.
`detect_interface_z`(vol[, sigma_xy, sigma_z, use_log])	Detect water/tissue interface along Z using gradient-based method.

Module Contents#

linumpy.geometry.interface.find_tissue_depth(vol, zmin=15, zmax=100, agarose_intensity=5000)[source]#

Detect the tissue interface depth in given volume.

This algorithm first segments the volume into tissue vs background(agarose) using the Li thresholding method and user-defined agarose intensity value. It then computes the XZ mean projection of the tissue mask, and detects the main edge position of the tissue/water interface using morphological operations and relative maximum detection. Other operations are done on the data to reduce the effect of intensity noise and artefacts on the water/tissue interface depth detection.

Parameters:

vol (ndarray) – Volume to analyze
zmin (int) – Minimum depth of interface in pixel
zmax (int) – Maximum depth of interface in pixel
agarose_intensity (int) – Agarose mean intensity value used to restrict analysis to tissue voxels

Returns:

Tissue interface depth

Return type:

int

Notes

The default depth is 40 px

linumpy.geometry.interface.get_interface_depth_from_mask(vol)[source]#

Compute the interface depths from a 3D tissue mask.

Parameters:: vol ((NxMxK) ndarray) – Tissue mask
Returns:: ndarray – Interface depth (in pixel)
Return type:: (NxM)

linumpy.geometry.interface.find_tissue_interface(vol, s_xy=15, s_z=2, use_log=True, mask=None, order=1, detect_cutting_errors=False)[source]#

Detect the tissue interface.

Parameters:

vol (ndarray) – Containing the volume to analyze
s_xy (int) – Uniform filter kernel size (xy)
s_z (int) – 1st order gaussian kernel size (z)
use_log (bool) – If True, apply log transform before filtering.
mask (ndarray, optional) – Optional mask restricting the analysis region.
order (int) – Gaussian filter order.
detect_cutting_errors (bool) – If True, detect and correct cutting artefacts.

Returns:

Tissue interface depth

Return type:

ndarray

linumpy.geometry.interface.find_cutting_plane(vol, z0map, agarose_mean, agarose_std)[source]#

Find the cutting plane using agarose segmentation.

Parameters:

vol (ndarray) – Input volume.
z0map (ndarray) – Interface depth map.
agarose_mean (float) – Mean intensity of agarose.
agarose_std (float) – Standard deviation of agarose intensity.

Returns:

popt
detected_interface (ndarray)
z0 (int)

Return type:

tuple[numpy.ndarray, numpy.ndarray, float]

linumpy.geometry.interface.remove_z0_outliers(z0map)[source]#

Remove outlier interface depths from the z0 map using median absolute deviation.

Parameters:: z0map (numpy.ndarray)
Return type:: numpy.ndarray

linumpy.geometry.interface.fit_interface(interface: numpy.ndarray, method: str = ..., return_center: Literal[False] = ...) → numpy.ndarray[source]#

linumpy.geometry.interface.fit_interface(interface: numpy.ndarray, method: str = ..., return_center: Literal[True] = ...) → tuple[numpy.ndarray, tuple[float, float]]

Fit a model on the given interface.

Parameters:

interface (ndarray) – Interface depth map to fit.
method (str) – Fitting method: ‘linear’, ‘quad’, ‘gauss’, or ‘sph’.
return_center (bool) – If True, also return the center of the fitted surface.

linumpy.geometry.interface.quadratic_interface(pos, a, b, c, d, e, f, g, h)[source]#

Evaluate a quadratic surface model for the tissue interface.

Parameters:

pos (numpy.ndarray)
a (float)
b (float)
c (float)
d (float)
e (float)
f (float)
g (float)
h (float)

Return type:

numpy.ndarray

linumpy.geometry.interface.get_quadratic_interface(popt, volshape=(512, 512, 120))[source]#

Compute the tissue interface map from quadratic fit parameters.

Parameters:

popt (numpy.ndarray)
volshape (tuple[int, int, int])

Return type:

numpy.ndarray

linumpy.geometry.interface.linear_homogeneous_profile(z, z0, dz, I0, Ib, sigma)[source]#

Intensity profile based on a single homogeneous tissue Beer-Lambert model (covered by some amount of water).

This will return the log(I).

Parameters:

z (ndarray) – Position where the intensity is evaluated
z0 (float) – Water-tissue interface depth
dz (float) – Interface Transition width
I0 (float) – Top tissue slice intensity (physics notation)
Ib (float) – Water intensity (physics notation)
sigma (float) – Tissue Attenuation coefficient

Returns:

Log(I) evaluated at each position z.

Return type:

ndarray

linumpy.geometry.interface.estimate_lh_profile_parameters(vol, s=25)[source]#

Estimates the linear-homogeneous intensity profile parameters.

Parameters:

vol (ndarray) – Volume for which the LHP parameters are evaluated
s (int) – Neighborhood used to average intensities at each depth

Returns:

float – z0 : Water-tissue interface depth
float – dz : Interface Transition width
float – I0 : Top tissue slice intensity
float – Ib : water intensity
float – sigma : Tissue Attenuation coefficient

Return type:

tuple[numpy.ndarray, numpy.ndarray, numpy.ndarray, numpy.ndarray, numpy.ndarray]

Note

This first version loops over all intensity profiles (x, y)

linumpy.geometry.interface.detect_interface_z(vol, sigma_xy=3.0, sigma_z=2.0, use_log=False)[source]#

Detect water/tissue interface along Z using gradient-based method.

Applies Gaussian smoothing then finds the peak of the first-order Z-derivative to locate the tissue surface.

Parameters:

vol (np.ndarray) – Volume with shape (X, Y, Z) – already transposed from OME-Zarr (Z, X, Y).
sigma_xy (float) – Gaussian smoothing sigma in XY before Z-gradient.
sigma_z (float) – Gaussian smoothing sigma for Z-gradient computation.
use_log (bool) – Apply log transform before gradient detection.

Returns:

Estimated interface depth in Z voxels.

Return type:

int