Image analysis systems for the detection of disseminated breast cancer cells on bone‐marrow cytospins

Abstract

We assessed the accuracy of automated cell imaging systems when compared to manual evaluation of cytospin slides in determining the presence of cytokeratin-positive, disseminated breast cancer cells in bone marrow aspirates. A total of 298 cytospin slides of bone marrow aspirates were first evaluated by individual screening by one expert immunocytologist. Subsequently, all 298 slides were evaluated by the Automated Cell Imaging System (ACIS) by ChromaVisiontrade mark. Two separate analysis cycles were performed using ACIS. The results of the two ACIS analysis cycles were almost identical: in 293 out of 298 samples (98.3%), identical numbers of disseminated breast cancer cells were detected. In the remaining five samples (1.7%), the result of the two ACIS analysis cycles differed by only one tumor cell. By using the manual technique, 120 cytospin samples were found to be positive. ACIS was able to detect additional tumor cells in 64 cases. Not once did ACIS miss tumor cells when compared to the manual technique. Using ACIS, we were able to determine the bone marrow status of patients with nonmetastatic breast cancer faster, with greater accuracy, and with greater reproducibility than with the manual technique.