Rapid acquisition of high-volume microscopic images using predicted focal plane

For an automated microscopic imaging system, the image acquisition speed is one of the most critical performance features because many applications require to analyse high-volume images. This paper illustrates a novel approach for rapid acquisition of high-volume microscopic images used to count blood cells automatically. This approach firstly forms a panoramic image of the sample slide by stitching sequential images captured at a low magnification, selects a few basic points (x, y) indicating the target areas from the panoramic image, and then refocuses the slide at each of the basic points at the regular magnification to record the depth position (z). The focusing coordinates (x, y, z) at these basic points are used to calculate a predicted focal plane that defines the relationship between the focus position (z) and the stage position (x, y). Via the predicted focal plane, the system can directly focus the objective lens at any local view, and can tremendously save image-acquisition time by avoiding the autofocusing function. The experiments showed how to determine the optimal number of the basic points at a given imaging condition, and proved that there is no significant difference between the images captured using the autofocusing function or the predicted focal plane.