Fundamental study of automatic cyto-screening for uterine cancer. III. New system of automated apparatus (CYBEST) utilizing the pattern recognition method.

An investigation was carried out, both from the point of view of combining features which are considered as being of the highest reliability based on pattern analysis of the nuclear structure and cytoplasm of cells, and the complexity of the apparatus. Several features such as "ambiguity" function were extracted by mathematical and statistical processing, and these were then made the parameters of a diagnostic logic. Also, a method in which a "differential histogram" and a "threshold value" are combined was developed for segmentation of cells. It became possible to operate an automatic scanning machine at a speed of about six minutes per slide using two-stage coarse and fine scanning of a flying spot scanner. The advantage of this system is flexibility of diagnostic program logic and speed of algorithms for a determined number of cellular features. An automatic focusing system was also developed and it functions satisfactorily at high speed on a fully extended regular object slide.