Computer-aided quantification of RNA levels detected by in situ hybridization of tissue sections

A semi-automatic program, designed for non-computer scientists, was developed for quantification of RNA levels detected by in situ hybridization in heterogeneous tissues. A video camera was used to acquire microscopic images of autoradiographed tissue sections which are then digitized on a video monitor for semi-automated quantification of silver grains. We describe a data entry and analysis procedure for systematic quantification of RNA levels in which about 300 cells per tissue sample can be analysed within 10 min. When compared with visual counting, computer-aided quantification was found to be more objective and reliable, with the highest variation coefficient between individuals being 7.5% using computer-aided quantification, compared to 24% with manual counting of the same section areas. A comparative study of c-myc oncogene expression in 11 mammary adenocarcinomas from 3 independent experiments showed the good reproducibility of results using the computer-aided method, with an 18% maximum variation between experiments. The program, with its simple user-interface, reliability and rapidity, is convenient for measuring specific genetic expression levels in clinical studies requiring large numbers of specimens.