Vitenskapelig artikkel

Rapid control of smoked Atlantic salmon (Salmo salar) quality by electronic nose: Correlation with classical evaluation methods

Haugen, John-Erik; Chanie, Eric; Westad, Frank Ove; Jónsdóttir, R.; Bazzo, Sandrine; Labreche, Said; Marcq, Pauline; Lundby, Frank; Olafsdottir, Gudrun


Tidsskrift: Sensors and actuators. B, Chemical, vol. 116, p. 72–77, 2006

Internasjonale standardnumre:
Trykt: 0925-4005
Elektronisk: 1873-3077

Open Access: none


This has been a European CRAFT project named “FishNose” (QLK1-CT-2002-71304), and its purpose has been to develop a dedicated gas-sensor based instrument for quality control of smoked salmon.

A prototype solid state based gas sensor array system including a gas sampling unit, the “FishNose”, for direct quality measurements of smoked salmon has been developed. Quality changes of smoked salmon during storage were monitored by the FishNose and compared with the results of traditional sensory, chemical, and microbial measurements.

Salmon samples were obtained from smokehouses in different countries (Norway, Iceland and Germany). Storage studies were carried out in laboratories in the different countries. Samples were taken from the same salmon in each case and analysed by all the methods. Sensory analysis based on Quantitative Descriptive Analysis (QDA) for smoked salmon was used and the assessors evaluated the samples each time by using 19 descriptors of odor flavor, appearance and texture. Chemical analyses of water, total fat, TVB-N, and salt content, were done according to AOCS official methods. The microbial analyses included total viable counts (TVC) (psychrotrophic counts) and Lactic acid bacteria (LAB) counts. A FOX 4000 (Alpha M.O.S, Toulouse, France) electronic nose equipped with 18 metal oxide semiconductors (MOS) sensors was used in the project.

Gas-sensor selection was optimised for the detection of changes in the very volatile compounds mainly representing microbial metabolism during spoilage. Sensor readings of repeated measurements of calibration samples showed a repeatability for the 6 sensors of the array of 5 %, and for repeated measurements of fish samples the repeatability was 4 %, without purge of the system between the measurements. Exploration of the data generated with “ FishNose” of fresh and stored smoked salmon were in agreement with the results from the reference methods (sensorial, chemical, microbial).

The system was further tested on-site in a smoked salmon production plant. Due to varying ambient air conditions at the production plant during the measurements, the sensor readings had to be corrected for by subtracting the fluctuating reference air sensor readings to obtain high classification rates of 93 to 100 % of respectively good and bad samples.

This work demonstrates that the FishNose is suitable for monitoring quality changes occurring during storage of smoked salmon and that the system is able to predict the quality related attributes like sweet/sour and off odor, and microbial counts. However, best prediction was obtained on samples from single producers.

Since it could be demonstrated that the Fishnose prototype measures the key bacterial metabolites related to spoilage, it could basically also be applied to unprocessed raw salmon. This will be looked at in a further future study.