Reproducibility of stereologic analysis of histologic sections
Abstract
In this study, inter-observer and intra-observer errors were compared in 17 stereologic estimates from the counts of five observers on five histologic components of fresh frozen sections of seven muscle biopsies, four histologic components of five carcinomas of the breast and two histologic components of three senile myocardium. Inter-observer error, defined as the difference of the estimates of five observers taken in pairs, had a range of 4 to 56 and 25th, 50th and 75th percentiles of 7, 21 and 28 percent. In contrast, intra-observer error, defined as the difference of two estimates of four observers, had a range of 2 to 12 and 25th, 50th and 75th percentiles of 4, 5 and 8 percent. Our findings indicate that experimental error caused by wrong counts in stereologic analysis of histologic sections can be reduced by using a single observer.