A new multivariate approach to the problem of fish quality estimation

Use of the interaction of electromagnetic fields with moist materials has typically meant the use of dielectric properties for measurement of water content. Recent work has widened the scope of such dielectric measurements by using a combination of dielectric spectroscopy and multivariate analysis to determine, among other things, some loosely defined qualities of foods. This has led to the work described here on the determination of the quality of chilled Baltic cod (Gadus morhua). sh obtained had known history and were stored in ice for up to 24 days. At intervals of two days, the dielectric properties of individual fish were measured in the time domain in a range from .01 to 1.0 ns, using an open-ended coaxial sensor and time domain reflectometer. In addition an `electronic noseʹ sensor array was used to gather data on various spoilage volatiles. The data so obtained were subjected to principal component analysis and regression to produce a calibration equation for the prediction of number of days on ice and quality index method (QIM).