Objective: The cytological diagnosis of coelomic fluid is essential for examining malignant mesothelioma (MM). However, reactive mesothelium (RM), caused by various factors, is morphologically similar to MM and thus often complicates the differential diagnosis. Here, nuclear luminance and steric alterations were assessed for the discriminant analysis of MM and RM. Study Design: Thirteen epithelial MM and 11 RM cases were included. One hundred alterations in the numbers of nuclear pixels and focus layers and the coefficient of variation of nuclear luminance among layers were determined for each case to conduct discriminant analysis using the Mahalanobis distance. Results: A cutoff value of 0.072 allowed highly accurate discrimination of MM (89.5%) and RM (89.6%). Fifteen cells appeared in 6 agglomerates of indiscriminable MM cases. The 6 agglomerates were individually examined. Malignant cells were dominant in 3 agglomerates (50%), allowing the discrimination of malignant cases. Conclusion: Discrimination using nuclear luminance and steric alterations is useful for morphologically indiscriminable MM cases. Three-dimensional analysis of agglomerates will be further investigated to improve the diagnostic accuracy.

Objective: Since well-differentiated adenocarcinoma cells of the lung (G1 cancer cells) show mild atypia, their differentiation from benign columnar epithelial cells (benign cells) is often difficult based on morphology. We performed discriminant analysis to distinguish benign from malignant cells by measuring 3-dimensional (3D) changes in nuclear luminance. Study Design: Discriminant analysis of 231 atypical cells prepared by bronchial brushing cytology, which were difficult to morphologically classify as benign or malignant, was performed using 100 G1 cancer cells. One hundred benign cells of bronchial brushing cytology specimens served as controls. The number of pixels of the nucleus, the number of focus layers and the level of change in the coefficient of variation (CV) of nuclear luminance between layers (3D-CV) were used as analytic parameters, and benign cells were discriminated from malignant cells based on the Mahalanobis distance. Results: As a result of discriminant analysis using a cutoff value determined in the control group, about 90% of the atypical cells difficult to classify as benign or malignant could be classified. Conclusion: For cells difficult to morphologically classify as benign or malignant, discriminant analysis based on 3D changes in nuclear luminance may be useful. This method can provide objective parameters for cancer diagnosis.