Wednesday 7 December 2011
Automated fluorescent in situ hybridization scanners are more rapid and accurate than people and are becoming more popular.
However, the methods used by these machines to score and interpret fluorescent in situ hybridization signals in nuclei are still those used by human observers.
A new approach is presented to make the software classify the fluorescent patterns in additional relevant categories, thus avoiding the rejection of relevant information and consequent bias.
The statistical interpretation of the fluorescent pattern distributions is carried out with a maximum likelihood estimation of the most likely proportions of various cell lines in the sample, and thus determines the diagnosis and the level of mosaicism in a single step.
This approach is faster and more accurate than those used by human observers.
It requires the scanning of fewer nuclei, less efforts for validation, and it makes the interpretation of results much simpler.