Automated analysis of cervical specimens using the T.A.S.

Abstract

The Leyden Television Analysis System (LEYTAS) is applied to the automated analysis of five positive specimens (2 cases of severe dysplasia, 2 cases of carcinoma in situ and 1 case of invasive carcinoma) and five negative specimens (4 cases of inflammation and 1 normal specimen). Two detection criteria based on size and contrast (so called 'top hat' transformation) are used to detect suspect nuclei. Thresholds and sizes have been used at two levels: 'high level' to detect very large and/or dark nuclei, and 'low level' to detect nuclei, of which absorption and/or size is only slightly increased compared to the nuclei of normal epithelial cells. In all positive specimens analysed with the 'high level' program suspect nuclei are automatically detected, whereas no nuclei are detected in any of the negative specimens. Using the 'low level' program many more suspect nuclei are detected in the positive specimens (up to a number of 294). This program results in some detected nuclei in the negative specimens, but the number remains relatively low (not more than 3 in the investigated cases).