Use of computerized cell image analysis to characterize cell nucleus populations from normal and neoplastic renal tissues.

We studied 55 renal tissue samples from 16 patients corresponding to normal (19 samples, group 1), low-grade (18 samples, group 2), and high-grade (18 samples, group 3) tumoral tissues. For this purpose, we used digital cell image analysis (the SAMBA 200 processor) to describe the morphonuclear patterns of Feulgen-stained nuclei from the 3 above-mentioned groups. Our results show that nuclear DNA ploidy is positively correlated with histopathological differentiation, which is also positively correlated with an increase in nuclear DNA heterogeneity. Morphometric and textural parameters computed on such Feulgen-stained nuclei make it possible to describe the typical morphonuclear patterns of normal, low-grade, and high-grade neoplastic renal tissues. Using multiparametric, i.e. principal-component and canonical analyses, we set up preliminary morphonuclear data banks that we used to assess the diagnosis of 6 ungraded samples. We expect that this kind of morphonuclear data banks might be helpful, on one hand, to select specific morphonuclear parameters related to patient survival, and on the other hand to establish the cytological diagnosis of deep fine-needle aspiration material, sonographically assisted, on suspicious kidneys. Such hypotheses are now under further study.