Color image analysis to grade shades of a color and its application to quantify stained tissues

This paper presents an approach to assign scores to the different shades of a particular color based on how close the shade is to the pure color. This can be used to quantify the stains which represent density of the substance under study in a histological tissue section. We developed the algorithm based on the YC_rC_b color model. C_r component provides a measure of 'redness' and C_b component provides a measure of 'blueness'. We extended the YC_rC_b color model to measure the color differences of magenta (Cm), yellow (Cy), cyan (Cc) and green (Cg). With the combination of these components with appropriate weightages, empirical formulae to quantify shades of each of these basic colors were developed. Using these formulae, each shade of a particular color is assigned a score which indicates how close it is to the particular color under consideration. This method was then applied to a study on the progress of pancreatic tumor in mice. The stains were quantified and the result verified.