A diffusion approach to labeling rows and columns in an irregular array.

A robust algorithm is presented for labeling rows and columns in an irregular array. The algorithm is based on hierarchical pattern matching to a local lattice, which is used as a template. Starting from the best local match, the pattern is expanded hierarchically to encompass the entire array. An application to labeling digitized images of an array of tissue sections mounted on a microscope slide is discussed.

Detection of circulating cytokeratin-positive cells in the blood of breast cancer patients using immunomagnetic enrichment and digital microscopy.

PURPOSE: To examine the feasibility for identifying and enumerating cytokeratin positive (CK+) cells in the peripheral blood of breast cancer patients. EXPERIMENTAL DESIGN: Blood specimens from 34 normal donors (negative controls), 15 samples to which carcinoma cells were added (positive controls), and 84 breast cancer patients [27 node-negative (N-), 29 node-positive (N+), and 28 metastatic] were studied. RBCs were lysed with ammonium chloride and the resulting cell suspension incubated with anti-EpCAM-conjugated immunomagnetic beads for carcinoma cell enrichment. Immunomagnetically selected cells were placed on slides; stained for CKs 8, 18, and 19; and evaluated with an automated digital microscopy system that rapidly scanned the slide and collected images of cells meeting predefined staining and cytomorphological criteria. A montage of the CK+ cells was reviewed to confirm tumor cell morphology. RESULTS: Eighteen specimens (9 normal, 2 N-, 4 N+, and 3 metastatic) were excluded because of poor cytomorphology or staining artifact. All 15 of the positive controls [95% confidence interval (CI), 78-100%] and none of the 25 negative controls (95% CI, 0-14%) demonstrated CK+ cells. Twenty-one of the 75 (28%; 95% CI, 18-40%) samples from breast cancer patients demonstrated CK+ cells including 76% of patients with metastatic disease (95% CI, 55-91%), 8% with N+ disease (95% CI, 1-26%), and none of those with N- disease (95% CI, 0-14). The mean number of CK+ cells detected in the 21 CK+ patients was 18.4 (range, 1-120). CONCLUSIONS: Breast carcinoma cells can be detected in the blood from a significant fraction of metastatic breast cancer patients using immunomagnetic cell enrichment and digital microscopy. The incidence of CK+ cells was low in those with resected N+ disease (at most 26%) and those with resected N- breast cancer (at most 14%). This technique could be used in large prospective studies of patients with breast cancer to learn whether the detection of rare carcinoma cells is a useful predictive or prognostic factor.