A Cluster Classification for Histologic Diagnoses of CNS Tumors in an Epidemiologic Study

Abstract

This paper introduces a method which was utilized to sort multiple pathologic diagnoses pertaining to brain tumors into clusters of equivalent diagnostic terms for an epidemiological study of brain cancer. The method enabled the medical record abstractors to ascertain cases for the study. It enabled the reviewing pathologist to identify differences between original and reviewed diagnoses. It identified the large range of classification systems of brain tumors used by practicing pathologists as a factor contributing to confusion of data if slide reviews are not incorporated into epidemiological studies, and demonstrated how the use of a cluster classification alleviates this problem.