An automated apparatus for cancer prescreening: CYBEST

The shortage of trained cytotechnologists has become a serious problem as the need increases for mass population screening for early detection of cervical cancer. Since less than five out of one-thousand cervical smears in mass population screening usually have the possibility of cancer, one of the most practical ways to solve this problem is to develop an automated prescreening instrument [1-4] which can distinguish reliably between nolznal and suspicious smears. The system named CYBEST [5,6] (Cyto-Biological Electronic Screener by Toshiba) has been developed based on pattern recognition techniques, because the cell malignancy appears predominantly in its morphological patterns. The purpose of this paper is to describe briefly the feature evaluation, the system design and implementation, and some techniques of picture processing used in the CYBEST system.