Figure 1. Process for obtaining nuclear morphometric data, showing a): a Feulgen-stain subimage, one of dozens obtained by breaking up whole slide images scanned at 400× on an Aperio ScanScope-CS®; b): a binary rendition of the previous image, produced by K-means clustering to identify pixels containing DNA; c): the same image with nuclei segmented using watershed algorithms available in Matlab®; d): three-dimensional maps showing optical density for DNA at each pixel in representative benign and PCa nuclei, respectively.
