Non-destructive stochastic model-based detection of diet-induced alterations in fish texture

TitleNon-destructive stochastic model-based detection of diet-induced alterations in fish texture
Publication TypeJournal Article
Year of Publication2012
AuthorsGrigorakis, K, Dimogianopoulos D
JournalSensing and Instrumentation for Food Quality and Safety
Volume6
Issue1-4
Pages35 - 47
KeywordsDicentrarchus labrax, Diet, Sea bass, Statistical hypothesis test, Stochastic model-based diagnosis, Texture, Traceability
Abstract

A stochastic model-based scheme detecting diet-induced textural alterations in fish muscle in an innovative, non-destructive manner is presented. The scheme operates on proven fault diagnosis principles as used in mechanical systems. It combines a cost-effective instrument setup, along with a non-destructive, vibration-like testing of fish samples and an accurate (under typical uncertainties) stochastic modeling of their response. The identified models provide key indicators indirectly related to the tested fish's texture, itself affected by (and, thus indicative of) its dietary history. Statistical hypothesis tests perform comparisons of such key indicators from tests with fish samples of different dietary histories. The issued statistical decisions allow for reliable tracing of fish on the basis of the diet-induced textural changes, while quantifying the risk of the decision-making process. Validation tests involved samples from two distinct fish groups sharing similar dietary histories except for a supplementation of dietary taurine at 1 % level in one of them. The altered textural characteristics in taurine-supplemented fish (initially suggested by taste panel evaluation) were effectively detected via the proposed scheme. These promising results suggest the scheme's potential as a non-destructive, cost-effective and reliable solution for fish traceability. © 2012 Springer Science+Business Media New York.

URLhttp://www.scopus.com/inward/record.url?eid=2-s2.0-84871291037&partnerID=40&md5=a1a994ce453ff89eb22f4430427f81db

User login