from Microscopic Imagery using Ellipse Detection

Cell image segmentation is a necessary first step of many automated biomedical image processing procedures. There certainly has been much research in the area. We add to this a new method that automatically extracts cells from microscopic imagery, and does so in two phases. Phase 1 uses iterated thresholding to identify and mark foreground objects or "blobs" with an overall accuracy of >97%. Phase 2 of the method uses a novel Genetic Algorithms-based ellipse detection algorithm to identify cells, quickly and reliably. The mechanism as a whole, has an accuracy rate >96%, and takes less than 1 minute (given our specific hardware configuration) to operate on a microscopic image.