Objective: To find efficient morphometric nucleolar features for distinguis
hing different prostatic atypia groups in fine needle aspiration biopsies.
Study design: A computerized interactive morphometry program was used to ou
tline nucleoli of prostate cells. The samples were divided into three group
s: group 1 (definitely benign [n = 10] and atypical but benign [n = 13]), g
roup 2 (moderately atypical [n = 11] or highly suspicious [n = 13]) and gro
up 3 (definitely malignant [n = 17]).
Results: The analysis revealed a difference ein a number of nucleoli betwee
n definitely benign samples and other atypia groups but not between the lat
ter. Nucleolar size features were best in distinguishing between atypia gro
ups. The sample-associated mean size features were mode powerful than cell
group-associated size features.
Conclusion: The sample-associated mean area, defined from cells selected by
an experienced cytologist, if larger than 2 mum(2), was strongly associate
d with definitely malignant samples. The same was true for the largest nucl
eolar size features appeared efficient in distinguishing between definitely
malignant and other samples.