Segmentation.

<p>(<b>A</b>) A phase image region around the previously selected seed is opened. (<b>B</b>) The phase image is transformed into a series of binary images by applying thresholds such that any pixel that is lower than the threshold is assigned as ‘putative cell’ (red regions) or ‘non-cell’ (blue regions). In total 256 ( = 2⁸) thresholds are applied corresponsing to all possible values the phase image can take (example thresholds 2, 50 and 130 are shown here). (<b>C</b>) The watershed algorithm is applied to all thresholded images from (B). The result is a set of distinct regions. Each region is compared with the seed (inset) and the phase image and classified as ‘cell’ or ‘non-cell’. (<b>D</b>). All ‘cell’ areas, from all thresholds, are summed together (<b>E</b>). The composite image allows for more accurate segmentation than an optimized single threshold.