Andreas Link, Irene Luna Pardo, Bernd Porr and Thomas Franke (2023) AI based image analysis of red blood cells in oscillating microchannels. RSC Advances 41.

• Paper: https://doi.org/10.1039/D3RA04644C
• Code: https://doi.org/10.5281/zenodo.7789864

Opens avi videos and extracts single frames, calculates the background and returns an image with its corresponding label.

Returns the frame positions where the cell is detected in narrow or wide section of the zig-zag channel.

Combines the framegenerator.py and celldetect.py to extract the frames where the cell is detected in the wide or narrow section.

Uses native and chem mod avi files to train and test a model where the user specifies the cell location in the channel. It evaluates the prediction accuracy and shows a plot for the best and worst detected images.

Usage: python train.py -n -w -h -f <maxframes>

-w: wide channel

-n: narrow channel

-f maximum of frames to use for debugging

-h prints this help text