Studies on the standardization of uterine cervical smears using a high-quality image analytical system

Abstract

We developed a rapid image sensor analytical system (RISAS) with a linear type charge-coupled device (CCD). We performed a discriminating standardization by Mahalanobis' generalized distance using a method of principal component analysis (multivariate analysis) on the basis of cells diagnosed by more than one cytopathologist as class III b (four cases, 54 cells) and class IV (three cases, 37 cells) in uterine cervical smears. We examined the system for quality control in cytodiagnosis by registering atypical cells (class III b, 14 cases, 81 cells; IV, 11 cases, 33 cells) diagnosed by cytoscreeners. Almost all cells diagnosed by the cytoscreeners were plotted into their respective class area made as a discriminated standard. It is suggested, therefore, that an image analytical system like ours is useful for quality control in cytodiagnosis.