Fitting

This module provides functionality around fitting the Model to given data.

Compare Masks

area_diff_mask()

Computes a 2D array where the two masks differ.

penalty_area_diff()

Calculates the penalty based on difference in colors.

parents_diff_mask()

Computes a 2D penalty array which accounts if cells are related.

penalty_area_diff_account_parents()

Uses the parents_diff_mask() to calculate the associated penatly.
Work with Masks and Cell Positions

extract_positions()

Extracts a list of position from a given mask.

convert_pixel_to_position()

Converts positions in pixel units to units of length (typically µm).

convert_cell_pos_to_pixels()

Converts positions in length units (typically µm) to pixel units.
area_diff_mask(mask1, mask2) → ndarray

Calculates a 2D array with entries 1 whenever colors differ and 0 if not.

Parameters:

  • mask1 (np.ndarray) – Mask of segmented cells at one time-point

  • mask2 (np.ndarray) – Mask of segmented cells at other time-point

Returns:

A 2D array with entries of value 1 where a difference was calculated.

Return type:

np.ndarray

convert_cell_pos_to_pixels(cell_pos: ndarray, domain_size: ndarray | tuple[float, float] | float, image_resolution: tuple[int, int] | ndarray)

Converts the position of a cell (collection of vertices) from length units (typically µm) to pixels.

Warning

This function performs only an approximate inverse to the convert_pixel_to_position() function. When converting from floating-point values to pixels and back rounding errors will be introcued.

Parameters:

  • cell_pos (np.ndarray) – Array of shape (N,2) containing the position of the cells vertices.

  • domain_size (float) – The overall edge length of the domain (typically in µm).

  • image_resolution (tuple[int, int] | np.ndarray) – A tuple containing the resolution of the image. Typically, the values for width and height are identical.

Returns:

The converted position of the cell.

Return type:

np.ndarray

convert_pixel_to_position(pos_pixel: ndarray, domain_size: ndarray | tuple[float, float] | float, image_resolution: tuple[float, float])

Contains identical arguments as the convert_cell_pos_to_pixels() function and performs the approximate inverse operation.

Warning

This function performs only an approximate inverse to the convert_cell_pos_to_pixels() function. When converting from floating-point values to pixels and back rounding errors will be introcued.

Returns:

The converted position of the cell.

Return type:

np.ndarray

extract_positions(mask: ndarray, n_vertices: int = 8) → list[ndarray]

Extracts positions from a mask for each sub-mask associated to a single cell. To read more about the used methods, visit the Fitting-Methods page.

Parameters:

  • mask (np.ndarray) – Array of shape (D1, D2, 3), (D1, D2, 1) or (D1, D2) containing pixel values of a mask for multiple cells.

  • n_vertices (int) – Number of vertices which should be extracted from each given cell-mask.

Returns:

A list containing arrays of shape (n_vertices, 2) containing the

individual positions of the cells.

Return type:

list[np.ndarray]

penalty_area_diff(mask1, mask2) → float

Calculates the penalty between two masks based on differences in color values (See also: area_diff_mask()).

Parameters:

  • mask1 (np.ndarray) – Mask of segmented cells at one time-point

  • mask2 (np.ndarray) – Mask of segmented cells at other time-point

Returns:

The penalty

Return type:

float

penalty_area_diff_account_parents(mask1: ndarray, mask2: ndarray, cell_container: CellContainer, parent_penalty: float = 0.5) → float

Calculates the penalty between two masks while accounting for relations between parent and child cells.

Parameters:

  • mask1 (np.ndarray) – Mask of segmented cells at one time-point

  • mask2 (np.ndarray) – Mask of segmented cells at other time-point

  • cell_container (CellContainer) – See CellContainer

  • parent_penalty (float) – Penalty value when one cell is daughter of other. Should be between 0 and 1.

Returns:

A 2D array containing penalty values between 0 and 1.

Return type:

np.ndarray

points_along_polygon(polygon: list[ndarray] | ndarray, n_vertices: int = 8) → ndarray

Returns evenly-spaced points along the given polygon. The initial and final point are always included.

Parameters:

  • np.ndarray (polygon(list[np.ndarray] |) – Ordered points which make up the polygon.

  • n_vertices (int) – Number of vertices which should be extracted.

Returns:

Array containing all extracted points (along the 0th axis).

Return type:

np.ndarray

parents_diff_mask(mask1, mask2, cell_container, parent_penalty=0.5)

Calculates the difference between two masks and applies a lower value where one cell is the daughter of the other.

Parameters:

  • mask1 (np.ndarray) – Mask of segmented cells at one time-point

  • mask2 (np.ndarray) – Mask of segmented cells at other time-point

  • cell_container (CellContainer) – See CellContainer

  • parent_penalty (float) – Penalty value when one cell is daughter of other. Should be between 0 and 1.