Distribution error in quantitative digital image analysis systems and its correction algorithm

One of the major sources of the coefficient of variation in quantitative digital image analysis systems is the distribution error. It is caused by uneven optical density distribution within the measuring area. In digital image analysis systems, the measuring area is an individual pixel. The distribution error would be eliminated if we could measure the average optical density within the measuring area. However, most cameras used with light microscopes measure the average transmittance of a pixel. The image analysis system then converts the measured transmittance into an optical density. If the optical density distribution within a pixel is not uniform, the average optical density converted from the measured transmittance is different from the true optical density of the pixel. The authors call this conversional distribution error (CDE). They have analyzed and characterized this error, and developed an algorithm to minimize CDE by estimating the optical density distribution within a pixel.<>