Matlab s image processing toolbox

During test sequences, brightfield images were taken every 0.5 seconds. Image processing was completed using Matlab’s Image Processing Toolbox (Mathworks). Image sequences were cropped to include the area immediately surrounding the cell. Each cropped frame in the image sequence was then binarized to isolate the cell body as a single connected component. The threshold for binarization was determined by Otsu’s method using the “greythresh” function.⁵⁴ If needed, a multiplier was used to adjust the threshold for a sequence to prevent any debris in the vicinity from being counted as part of the cell. The two-dimensional cell centroid was determined for each frame using Matlab’s “regionprops” function. Cell centroid displacement was quantified by measuring the distance between the cell centroid and the lower left corner of the cropped image. The lower left corner of the cropped image was chosen because directed DEP pushing force was consistently applied towards the upper right corner of the image; thus, when the cell deformed, the distance between the cell centroid and the lower left corner of the image increased. All measurements were then normalized relative to the initial cell position at the beginning of the test sequence.

Ultrasonic vocalization recordings were analyzed using the WAAVES program (Reno et al., 2013 (link)). Briefly, this program reads audio files and produces a spectrogram. The spectrogram is then searched for sound objects using MATLAB’s Image Processing Toolbox (MathWorks, Inc. Natick, MA). A series of filters is then applied to separate background sounds from USVs. Finally, several measurements of interest are taken on each USV and stored for subsequent analysis.