Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

Cartoon illustration of a ridged microfluidic channel.

A system that can be used for sorting cells according to their biomechanical properties.

More »

Fig 1 Expand

Fig 2.

Example data.

(a) a short segment of video recording shown by overlapping multiple frames, (b) desired single-cell trajectories to be extracted.

More »

Fig 2 Expand

Fig 3.

Foreground identification.

We first estimate the background by the median of nearby frames, then perform linear regression of the frame of interest against the estimated background, threshold the regression residue to identify foreground pixels, and finally perform median filtering to refine the foreground.

More »

Fig 3 Expand

Fig 4.

Matching events in consecutive frames.

Two examples of sets of possible matchings between events in current frame (light gray) and events in the next frame (dark gray), along with scores of the matchings. Numbers are used to label the events.

More »

Fig 4 Expand

Table 1.

Algorithm for forward matching of consecutive frames.

More »

Table 1 Expand

Fig 5.

Multiple-to-one matching.

By shrinking the distances among the multiple events and rotating them together, templates are generated to represent possible configurations of the multiple in the next frame. Templates are evaluated by their number of pixels that overlap with the one event in the next frame. The maximal overlapping template is used to segment the one event in the next frame into multiple pieces, turning the multiple-to-one matching into one-to-one.

More »

Fig 5 Expand

Fig 6.

The forward and backward matching process.

Each column corresponds to one image frame. Left-to-right is the forward directions. Numbers are used to indicate the cell ID associated to each event. Each vertical arrow represents one step of the forward or backward matching between two consecutive frames. As the algorithm proceeds, multiple-to-one matchings either cause the multiple merge or the one to split. One-to-multiple matchings either cause the multiple to merge (not contained in this example), or cause the multiple to receive new cell IDs.

More »

Fig 6 Expand

Fig 7.

Summary of tracking results on recordings of cells under two perturbation conditions.

(a) Size of the data and tracking performance. (b) Comparing the size of events associated to correct and incorrect trajectories, the incorrectly tracked events are mostly small debris.

More »

Fig 7 Expand

Fig 8.

Visualization of tracking results.

Overlay of the correctly tracked trajectories on the background of the recordings.

More »

Fig 8 Expand

Fig 9.

Comparison of 0CD (stiff) and 1.5CD (soft) cells.

(a) Stiff cells drift up along the y-direction, whereas soft cells tend to have a slightly negative drift. (b) The speed on ridge of stiff cells is smaller than their speed in gap. Soft cells are less affected by the ridges. (c) Stiff cells tend to travel faster after passing each ridge, whereas (d) soft cells travel at a relatively constant speed irrespective of the ridges.

More »

Fig 9 Expand