Depth determination in crystal lography using auto-focusing algorithms - biomed 2010

Many drugs used in modern medicine are created via sophisticated drug creation techniques that require crystallography. While crystallography systems are becoming more and more automated, there are still signi?cant engineering obstacles to achieving the desired goal of full automation. One of the biggest challenges in automating crystallography has been the harvesting step, which is still typically performed using a manually-controlled micro-manipulator with the operator looking through a microscope. We set out to further automate this step of the crystallography process by using machine vision to identify viable crystals and determine their three dimensional coordinates. The coordinates could then be sent to a computer controlled robotic manipulator for harvesting. The most di?cult challenge in determining the 3D coordinates of the crystal is accurately determining its depth as it sits in one of the wells of a micro titer plate. In this paper we describe a study in which we investigated the use of auto-focusing algorithms similar to those used in digital single-lens re?ex DSLR) cameras to determine crystal depth in the well. This method could be an alternative to relatively complex and expensive approaches such as optical coherence tomography (OCT) or confocal re?ectometry. We investigated seven di?erent auto-focusing algorithms, and report on the results.