spots_in_yeasts.spotsInYeasts module

class spots_in_yeasts.spotsInYeasts.YeastsPartitionGraph(o_graph, cell_to_nuclei, nucleus_to_cells, labeled_yeasts, labeled_nuclei)[source]

Bases: object

Class implementing the Hopcroft-Karp algorithm to create a maximal dual-matching.

bfs()[source]
dfs(node)[source]
find_partition(vertex, cell_to_nuclei, nucleus_to_cells, o_graph)[source]
launch_hopcroft_karp()[source]
make_cells_lut()[source]
make_new_labels(old_labels, old_nuclei, cell_to_nuclei)[source]
make_nuclei_lut(cell_to_nuclei, cells_lut)[source]

Makes a dictionary to transform the indices of nuclei. It is used to make the nuclei indices match their owner cell’s index.

Parameters

  • cell_to_nuclei – Structure giving for a cell the index of its nucleus.

  • cells_lut – LUT used to make mother and daughter cells match each other’s label.

Returns

A LUT to transform the nuclei indices.

remove_borders(labeled_yeasts, labeled_nuclei)[source]
spots_in_yeasts.spotsInYeasts.adjacency_graph(labeled_cells, check_undirected=False)[source]

Creates the adjacency graph of the segmented cells. Two cells are called “adjacent” if their labels are touching one another in 8-connectivity.

Parameters

  • labeled_cells – The labeled image representing segmented cells.

  • check_undirected – Activate the cheking of the graph essential properties (only for debug purposes).

Returns

A dictionary representing the cells as a graph, indexed by cells’ labels. For each label, gives: its location and its neighbors

spots_in_yeasts.spotsInYeasts.assign_nucleus(labeled_cells, labeled_nuclei, covering_threshold, graph=None)[source]

First step of the nuclei segmentation. It starts by finding all the cells having “their own nucleus” (== a cell overlaped by a nucleus)

Parameters

  • labeled_cells – The image containing the labeled cells.

  • labeled_nuclei – The image containing the labeled nuclei.

  • covering_threshold – Percentage of a cell that must be covered by nuclei for this cell to be considered dead.

  • graph – Adjacency graph of the labeled cells.

Returns

  • The labeled cells from which we removed invalid individuals.

  • The labeled nuclei

  • The adjacency graph from which we removed dead individuals.

  • A structure mapping cells indices to nuclei indices.

  • A structure mapping nuclei indices to cells indices.

spots_in_yeasts.spotsInYeasts.associate_spots_yeasts(labeled_cells, labeled_spots, fluo_spots, area_threshold_down, area_threshold_up, solidity_threshold, extent_threshold, classification=None)[source]

Associates each spot with the label it belongs to. A safety check is performed to make sure no spot falls in the background.

Parameters

  • labeled_cells – A single-channeled image with dtype=uint16 containing the segmented transmission image.

  • labeled_spots – A labelised image representing spots

  • fluo_spots – The original fluo image containing spots.

Returns

A dictionary in which keys are the labels of each cell. Each key points to a list of dictionary. Each element of the list corresponds to a spot. An image representing labels in the fluo channel (a label per spot) is also returned.

spots_in_yeasts.spotsInYeasts.create_random_lut()[source]

Creates a random LUT of 256 slots to display labeled images with colors far apart from each other. Black is reserved for the background.

Returns

A cmap object that can be used with the imshow() function. ex: imshow(image, cmap=create_random_lut())

Return type

LinearSegmentedColormap

spots_in_yeasts.spotsInYeasts.create_reference_to(labeled_cells, labeled_spots, spots_list, name, control_dir_path, source_path, projection_cells, projection_spots, indices, labeled_nuclei, nuclei_fluo, spots_colors)[source]

Creates a folder containing everything a user needs to see in order to check whether the process ended correctly and produced a correct segmentation.

spots_in_yeasts.spotsInYeasts.distance_spot_nuclei(labeled_cells, labeled_nuclei, labeled_spots)[source]

Assign a class to every spot depending on its location according to the nucleus of the cell it is in. It can be ‘nuclear’, ‘cytoplasmic’ or ‘peripheral’.

Parameters

  • labeled_cells – The image containing labeled cells.

  • labeled_nuclei – The image containing labeled nuclei.

  • labeled_spots – The image containing labeled spots.

Returns

A dictionary giving for each spot label (int), its category (str).

spots_in_yeasts.spotsInYeasts.fill_holes(image)[source]

A hole is a background area surrounded by a unique label and from which we can’t reach the image’s border. No in-place editing.

Parameters

image – A labeled image on black background

Returns

The same image as input but without holes in labels.

spots_in_yeasts.spotsInYeasts.find_focused_slice(stack, around=2)[source]

Determines which slice has the best focus and selects a range of slices around it. The process is based on the variance recorded for each slice. Displays a warning if the number of slices is insufficient. A safety check is performed to ensure that the returned indices are within the range of the stack’s size.

Parameters

  • stack – (image stack) The stack in which we search the focused area.

  • around – (int) Number of slices to select around the most in-focus one.

Returns

A tuple centered around the most in-focus slice. If we call ‘F’ the index of that slice, then the tuple is: (F-around, F+around).

Return type

(int, int)

spots_in_yeasts.spotsInYeasts.get_neighbors(l, c, height, width)[source]
spots_in_yeasts.spotsInYeasts.is_undirected(graph)[source]

Checks whether a graph is undirected or not.

spots_in_yeasts.spotsInYeasts.nuclei_from_fluo(stack_fluo_nuclei)[source]

Produces a segmentation of nuclei based on the nuclei marking (either a single slice or a stack).

Parameters

marking. (An image or a stack representing the nuclei) –

Results:

The MIP of the provided image and a labeled version of nuclei.

spots_in_yeasts.spotsInYeasts.place_markers(shp, m_list)[source]

Places pixels with an incremental intensity (from 1) at each position contained in the list.

Returns

A mask with black background and one pixel at each intensity from 1 to len(m_list).

Parameters

  • shp – A 2D tuple representing the shape of the mask to be created.

  • m_list – A list of tuples representing 2D coordinates.

spots_in_yeasts.spotsInYeasts.prepare_directory(path)[source]

Prepares a directory to receive the control images for the batch mode.

spots_in_yeasts.spotsInYeasts.remove_excessive_coverage(labeled_cells, labeled_nuclei, covering_threshold)[source]

Removing cells in which the nucleus occupies too much surface (dead cell).

Parameters

  • labeled_cells – The image containing the labeled cells.

  • labeled_nuclei – The image containing the labeled nuclei.

  • covering_threshold – Percentage of a cell that must be covered by nuclei for this cell to be considered dead.

Returns

Two sets containing the indices of discarded nuclei and the indices of discarded cells.

spots_in_yeasts.spotsInYeasts.remove_labels(image, labels)[source]

Macro-function removing some labels from an image. The modification is made in-place.

Parameters

  • image – A labeled image.

  • labels – A set of indices that we want to remove from image.

spots_in_yeasts.spotsInYeasts.segment_nuclei(labeled_yeasts, stack_fluo_nuclei, threshold_coverage)[source]

Launches the procedure to segment nuclei from the dedicated fluo channel, and merge mother cells with their daughter if the division process is still ongoing.

Parameters

  • labeled_yeasts – Image containing the labeled yeast cells.

  • stack_fluo_nuclei – Image containing the stained nuclei.

  • threshold_coverage – Percentage (in [0.0, 1.0]) of a cell that must be covered by a nucleus to be considered dead.

Returns

  • The maximal projection of the stained nuclei channel.

  • The labeled yeasts from which we removed the dead cells.

  • The image containing the labeled nuclei.

spots_in_yeasts.spotsInYeasts.segment_spots(stack, labeled_cells, death_threshold, sigma=3.0, peak_d=5, threshold_rel=0.7)[source]
Parameters

stack – A numpy array representing the fluo channel

Returns

A dictionary containing several pieces of information about spots.

  • original: The input image after maximal projection

  • contrasted: A version of the channel stretched on the whole histogram.

  • mask: A labeled image containing an index per detected spot.

  • locations: A list of 2D coordinates representing each spot.

spots_in_yeasts.spotsInYeasts.segment_transmission(stack, gpu=True, slices_around=2)[source]

Takes the path of an image that contains some yeasts in transmission.

Parameters

stack – A numpy array representing the transmission channel

Returns

A uint16 image containing labels. Each label corresponds to an instance of yeast cell.

spots_in_yeasts.spotsInYeasts.segment_yeasts_cells(transmission, gpu=True)[source]

Takes the transmission channel (brightfield) of yeast cells and segments it (instances segmentation).

Parameters

transmission (image) – Single channeled image, in brightfield, representing yeasts

Returns

(image) An image containing labels (one value == one individual).

spots_in_yeasts.spotsInYeasts.write_labels_image(image, font_scale)[source]

Creates an image on which the indices of labels are literally written in the center of each label.

Parameters

  • image – A labeled image.

  • fonst_scale – Size of the font to write digits.

Returns

A binary mask representing the literal index of each label.