Electronic Nose and Tongue for Pet Food Classifi cation

Commercial canned dog and cat foods (four type of each) were classifi ed by electronic nose (EN) and tongue (ET) methods. Th e classifi cation was performed by canonical discriminant analysis (DA) followed by cross-validation, using the ET and EN sensory values separately (7 and 18 sensors) and also jointly. Th e number of entered variables corresponding to the total number of sensors (n=25) were decreased by using a stepwise procedure during DA. First the dog and cat samples were classifi ed than the discrimination were performed on the canned foods (eight type). Th ereaft er two groups were formed depending on the compositional characteristics of the foods (pure animal vs animal and plant origin), and fi nally these groups were divided into four subgroups according to the concerning species (dog vs cat). In general, the lowest discriminating results were achieved by the single application of ET method (58.381.7 %). Th e highest classifi cation power (85–98.3%, CV% 83.3–95.8) derived from the joint application of the two sensory methods. According to the results achieved, the common application of EN and ET technology seems to be a promising tool for the aroma classifi cation of pet foods.