Fig. 2

Image pre-processing and initial candidate detection. a Upon initiation of automated analysis, each imported raw image undergoes a series of pre-processing steps that aid in identifying prospective candidate cilia and excluding other non-relevant objects present in the image. After binarization using adaptive threshold and preliminary selection, only the most likely tens of candidate cilia will be shown. b The detection of true candidates can be further automatically optimized using the software’s four tuning parameters. Data pertaining to length and directional score depend only on the binary mask, while data pertaining to signal-to-noise ratio (SNR) and standard deviation (STD) depend on both the binary mask and raw cilia ROI