Tracking centroids of segmented cells

%pip install -q --upgrade -r requirements.txt

Importing packages

laptrack.LapTrack is the core object for tracking.

We also import regionprops_table from skimage to calculate the centroids of the segmentation masks.

import napari
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
from skimage.measure import regionprops_table

from laptrack import LapTrack
from laptrack import datasets
from laptrack.data_conversion import convert_split_merge_df_to_napari_graph

Load images and showing in napari

First, we load images to the napari viewer.

viewer = napari.Viewer()
images, labels = datasets.cell_segmentation()

viewer.add_image(images)
labels_layer = viewer.add_labels(labels)

Calculating label centroids by regionprops_table

We then calculate the centroids for each labeled region in each frame.

regionprops = []
for frame, label in enumerate(labels):
    df = pd.DataFrame(regionprops_table(label, properties=["label", "centroid"]))
    df["frame"] = frame
    regionprops.append(df)
regionprops_df = pd.concat(regionprops)

Tracking centroids by LapTrack

This is where we use LapTrack. We track the centroids of the segmented regions.

lt = LapTrack(track_cost_cutoff=15**2, splitting_cost_cutoff=30**2)
track_df, split_df, merge_df = lt.predict_dataframe(
    regionprops_df.copy(),
    coordinate_cols=["centroid-0", "centroid-1"],
    only_coordinate_cols=False,
)

track_df = track_df.reset_index()
graph = convert_split_merge_df_to_napari_graph(split_df, merge_df)
viewer.add_tracks(
    track_df[["track_id", "frame", "centroid-0", "centroid-1"]],
    graph=graph,
)

<Tracks layer 'Tracks' at 0x7f8ec8bcbc10>

Showing clonal cells by the same colors

To check the result, we show the clonal cell regions by the same colors.

new_labels = np.zeros_like(labels)

for i, row in track_df.iterrows():
    frame = int(row["frame"])
    inds = labels[frame] == row["label"]
    new_labels[frame][inds] = int(row["tree_id"]) + 1
viewer.add_labels(new_labels)
labels_layer.visible = False

viewer.dims.current_step = (9, 0, 0)
plt.imshow(viewer.screenshot())
plt.xticks([])
plt.yticks([])

([], [])