Automated cytopathologic diagnosis of bronchial carcinoma. II. Preliminary results of classifying cytocentrifuged bronchial brushings.

ABSTRACT

An automated discrimination between healthy and neoplastic bronchial cells was performed on eight bronchial smears prepared by cytocentrifugation. An image analyzer was used to examine 415 cells in these smears. The nuclear surface of each cell was measured, as was the total integrated optical density for 25 programmed thresholds. The results show that it is possible to distinguish healthy from cancerous cells in a given subject using these two measured parameters and two new parameters deduced mathematically. It appears difficult, however, to demonstrate a typical healthy and typical cancerous bronchial cell that could be used as a reference for all subjects. It is thus the presence of cell heterogeneity in a given subject that enables him or her to be characterized as healthy or having cancer.