A template matching approach for segmenting microscopy images

We describe a generic method for segmenting microscopy images based on supervised statistical modeling. The idea is to utilize example input segmentations to learn a statistical model of the shape and texture of the structures to be segmented. Segmentation of a test image then can be performed by maximizing the normalized cross correlation between the model and neighborhoods in the test image, followed by a final adjustment that utilizes nonrigid registration. We demonstrate the application of the method in segmenting several types of microscopy images of cells and nuclei.