Study Of Rapid Detection Of Spoilage Spawned By Alicyclobacillus Spp. In Apple Juice Using E-nose And E-tongue

The inherent acid-resistance and hot-tolerance of Alicyclobacillus spp.enables the survival of its spores in commercial pasteurized apple juice products,which will then germinate,grow and produce off-flavor compounds like guaiacol that can lead to the flavor degradation of apple juice and consequently the loss of manufacturers' interest as well as theconsumer confidence.Therefore,a timely and effective detection approach targeting this microbe is increasingly essential and particularly important.Based on the background above,this study aimed at the real-time detection of Alicyclobacillus spp.and optimized the working parameters of e-nose and e-tongue for spoilage detection of apple juice.Under the optimized conditions,appropriate qualitative models were built based on the response signals of e-nose and e-tongue for discrimination of apple juices spoiled by different microorganisms as well as diagnosis of the contamination spawned by Alicyclobacillus spp.Moreover,the characteristic metabolite of Alicyclobacillus spp.in apple juice were also fastly recognized and quantified according to the signals of e-nose and e-tongue.Finally,the quality variation of apple juice caused by A.acidoterrestris spawned contamination was investigated,and the early diagnosis and quick prediction models regarding A.acidoterrestris spawned spoilage were established based on the fusing signals of e-nose and e-tongue.The specific research contents and results are as follows:(1)The optimized working conditions of e-nose and e-tongue were determined by single factor test combined with one-way analysis of variance,Duncan multiple comparison and principle component analysis.The detailed working parameters for e-nose were as follows: volume of apple juice 10 mL,headspace gas generation time 60 min,flow rate of carrier gas 300 mL/min,and the optimized stirring rate of e-tongue was 2 rps.The application of e-nose and e-tongue in tandem with principle component analysis and Fisher linear discriminant analysis could recognize Alicyclobacillus spp.spawned spoilage in apple juice and apple juice beverage,and the Fisher linear discriminant analysis exerted a higher distinguishing ability.(2)The characteristic odor and taste finger-print maps of apple juice spoiled by different microbes were captured by e-nose and e-tongue,and appropriate eigenvalues were then extracted for analysis.The results indicated that apple juices contaminated by microbes from different families could be distinguished according to the e-nose signals combined with principle component analysis and Fisher linear discriminant analysis,and the Fisher discriminant analysis exhibited better distinguishing ability compared to principle component analysis.On the other hand,e-tongue in tandem with either Fisher or Bayes discriminant analysis could facilitate the diagnosis of contamination and differentiation among different types of spoilage in apple juice.Furthermore,the results of Bayes discriminant analysis was better than that of Fisher discriminant analysis,which could distinguish the apple juice spoilage spawned by microbes from different strains but belonging to the same family.Moreover,combined with Fisher discriminant analysis,the fusing signals of e-nose and e-tongue could classify the apple juice samples spoiled by different microbes correctly,distinguish Alicyclobacillus spp.spawned spoilage in apple juiceand particularly,differentiate the spoilage caused by different strains of Alicyclobacillus spp.,exhibiting better discriminating ability than exclusive use of e-nose or e-tongue.(3)E-nose and e-tongue were utilized for detection of guaiacol in apple juice,and the resultant signals were subjected to statistic analysis to investigate the feasibility of detecting this characteristic metabolite of Alicyclobacillus spp.The results showed that apple juices contaminated by guaiacol of different concentrations were discriminated by e-nose responses combined with Fisherdiscriminant analysis,while few apple juice samples were erroneously classified,obtaining a detection limit lower than 0.5 mg/L.On the other hand,all the apple juice samples could be quite correctly grouped based on the signals of e-tongue in tandem with Fisherdiscriminant analysis,with a detection limit lower than 0.25 mg/L.The loadings analysis suggested that sensors S2,S6,S7,S8 and S9 of e-nose and sensors CA,GA,HA,JB and JE of e-tongue had contributed enormously to the classification of apple juices contaminated by guaiacol of different concentrations.(4)The quality variation of apple juice at different stages of spoilage were studied through determination of physicochemical properties and count of A.acidoterrestris,as well as sensory evaluation.The results indicated that the contents of total sugar in apple juice decreased with increased contamination time,while the total acid remained constant during spoilage.The contamination of A.acidoterrestris experienced a lag phase in the initial 8 h,multiplied in the following hours and reached 107 cfu/mL after 24 h.The intrinsic fruity and sweet flavor of apple juice had gradually been covered up by herbal,smoky and medicinal off-flavor.All the panels could identify apple juice samples contaminated for 24 h,but could not recognize apple juice contaminated for 4 h,and only 1-2 of them could barely discern apple juice contaminated for 8-12 h.(5)E-nose and e-tongue were applied for detection of apple juice samples contaminated by A.acidoterrestris at different period,and the resultant odor and taste finger-print maps of apple juice were subjected to chemometrics for analysis.The results suggested that e-nose signals combined with Fisher discriminant analysis could identify apple juice contaminated for 4 h,which however,had not been recognized by e-tongue signals in conjunction either with principle component analysis or Fisher discriminant analysis.With the fusing signals of e-nose and e-tongue in tandem with Fisher discriminant analysis,not only could the apple juices at different spoilage stage be distinguished,but also the apple juice contamination could be discerned in 4 h,corresponding to the count of A.acidoterrestris at 2.2×102 cfu/mL,without production of off-flavor metabolite.The quantitative model for prediction of spoilage status of apple juice(i.e.the count of A.acidoterrestris)were established separately based on the signals of e-nose,e-tongue and the fusing signals of e-nose and e-tongue via partial least square regression,with determination coefficient R2 of 0.91,0.95 and 0.97 respectively,as well as root mean square error of 0.56,0.50 and 0.42 correspondingly.The results indicated that by synthesis of the odor and taste information,the fusing signals of e-nose and e-tongue had higher classification and prediction ability than exclusive use of e-nose or e-tongue.On this basis,the predictive ability of multiple linear regression,partial least square regression and back-propagation neural network regarding the count of A.acidoterrestrisin apple juice was compared,and the results showed that the best prediction performance was obtained by back-propagation neural network,with determination coefficient R2 of 0.99 and root mean square error of 0.32.In conclusion,the aforementioned study had provided theoretical foundation and technical guidance for early detection and prediction of A.acidoterrestris-spawned spoilagein apple juice during daily production and circulation.