In-use evaluation of the Omnicon automated tumor colony counter.

The reproducibility and accuracy of the Omnicon (Bausch and Lomb Inc., Rochester, NY) automated tumor colony counter for counting tumor colonies growing in double layer soft agar is evaluated and the reproducibility is compared with manual tumor colony counting. Replicate within day run-to-run colony counts of the Omnicon show a median correlation coefficient (r) of greater than 0.985, and day-to-day median r of greater than 0.980. In contrast, for manual colony counting, the best intra-observer reproducibility achieved is a r of 0.943 and the best inter-observer reproducibility is a r of 0.831. Analysis of results from individual culture plates counted by the Omnicon on 5 separate days shows a median coefficient of variation of 10% with 77% of the culture dishes showing coefficients of variation of colony counts over 5 days of less than 20%. Counting of culture plates during incubation shows that the Omnicon is counting tumor colonies developing after plating of a single cell suspension.