- J Vigo: Groupe de Microfluorimétrie Quantitative, U.A. CNRS 1289, Université de Perpignan, France.
Image analysis is being increasingly used in biology and medicine; however, in order to obtain truly quantitative data and thus avoid errors in interpretation, a certain number of precautions must be taken when the image is digitized, well before any attempt is made to analyse or interpret the data. This is particularly true for image microfluorometry. In this article we will examine an image analysis system for fluorescent images composed of a mercury lamp, a microscope, a high sensitivity video camera and an image analyser and evaluate the principal sources of random and non-random errors, various constraints, and their relative importance. A signal correction protocol is proposed to minimize non-random errors during digitalization. A few examples are given to illustrate its efficiency.