AUTOMATED 3-DIMENSIONAL IMAGE-ANALYSIS OF THICK AND OVERLAPPED CLUSTERS IN CYTOLOGIC PREPARATIONS - APPLICATION TO CYTOLOGIC SMEARS

Methods are presented for automated analysis of thick and heavily over lapped regions of cytologic preparations, such as cervical/vaginal sme ars. Current systems are unable to process these regions although they contain diagnostically valuable information. We argue that analysis o f such regions is inherently a three-dimensional (3-D) problem that ca nnot be solved reliably with conventional two-dimensional methods. Fur thermore, this issue cannot be sidestepped by special thin preparation methods. Even with 3-D imaging, analysis of these regions is complica ted by the high variability in the image gray level and textural featu res resulting from the uncontrollable cell overlaps and folds and larg e computational requirements. A novel approach based on 3-D imaging an d adaptive 3-D analysis algorithms based on the principles of localiza tion, adaptive data reduction and clustering theory is presented. It w as successful in detecting and separating deeply embedded and overlapp ing nuclei, cytoplasmic folds and creases in thick and overlapped regi ons of conventional smears and special thin preparations.