Automatic counting of fungiform papillae by shape using cross-correlation

• An algorithm for the automatic detection and counting of fungiform papillae was proposed.
• The algorithm is based on shape of the fungiform papillae as measured by cross-correlation.
• The algorithm counts the fungiform papillae very similar to the manual procedure.
• This algorithm is valuable to the sensory evaluation and medical community.

The determination of the number of fungiform papillae (FP) on the human tongue is important for taste sensitivity studies. Most of the time, the counting of the FP is done manually. In this paper we propose a novel algorithm to count the FP using shape characteristics measured by cross-correlation. The accuracy of the algorithm is evaluated by counting the FP manually on the same images and then doing a statistical analysis. A Poisson regression model is fitted using maximum likelihood. The result is that the algorithm counts are very similar to the human experts. Another advantage of the algorithm is its facility of use, velocity and that it can work on a plain tongue image, without the need to stain the tongue as is usual in manual counting.