In pathology, accurate cell segmentation is essential in determining valuable quantitative diagnostic information for pathologists. In this article, we present a generalized clustering approach for segmentation of microscopic cytological lung cell images. The cluster centroids or representative vectors are generalized and expanded from single vectors to sets of vectors to adaptively fit to the lung cell clustering shapes. Experimental results of the proposed approach for the cytological color lung cell images are provided and compared with those of classical K-means clustering approach. The algorithm is also applied to thyroid cell images and the segmentation results show that the approach is applicable without modification.