Flavor Evaluation of UHT Reconstituted Milk Processed by Direct and Indirect Heat Treatment Using Sensors

Food Res. & Dev. Inst., Morinaga Milk Industry Co., Ltd., 5-1-83, Higashihara, Zama-City 228-8583, Japan 1Production Dept., Morinaga Milk Industry Co., Ltd., 5-33-1 Shiba, Minato-ku, Tokyo 108-8384, Japan 2Food Sci. & Tech. Inst., Morinaga Milk Industry Co., Ltd., 5-1-83, Higashihara, Zama-City 228-8583, Japan 3Dept. of Electronics, Grad. Sch. of Info. Sci. and Elect. Engin., Kyushu Univ., 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan


Introduction
Various types of milk products are available on the market, for example, milk, lowfat milk, nonfat milk, composition-controlled milk, processed milk, milk drinks, and so on.Drinks containing these milk products generally use raw milk and processedmilk materials made from raw milk, including skim milk powder, cream, butter, and (1) the pasteurization temperature, (2) the pasteurization method, (3) and the homogenization pressure. (4)However, of food.This approach, however, involves some problems, including the effect of subjectivity on data reproducibility, for instance, individual differences and the physical condition of the panel members, the limited number of samples used in consideration of the load placed on the panel, and so on.Recently, there have been an increasing number (5) For example, we have previously reported the application of (6,7)   reconstituted milk produced by combining various types of milk materials and different pasteurization methods.We conducted a comprehensive evaluation of an experimental model system of ultra high temperature (UHT) processed milk and four types of reconstituted milk prepared with a plate-type UHT pasteurizer with indirect heating and GC analysis.We report the results here.

Sample preparation
of UHT processed milk and four samples made with combinations of two types of reconstituted milk and two types of pasteurizers.The materials used were raw milk collected in our plant, and skim milk powder and butter manufactured from the same raw milk.The pasteurizers were a plate-type pasteurizer (Plate) and an infusion pasteurizer (Inf).The two types of reconstituted milk were reconstituted milk in which raw milk was not blended, and reconstituted milk in which raw milk was blended at a ratio of 50% to produce the same composition as that of raw milk (fat 4.0%, solids not fat (SNF) 8.5%).Pasteurization was conducted at 130°C for 2 s, followed by homogenization at a pressure of 14 MPa (3.5 MPa at a second stage).Samples were prepared once.They were used for the sensory evaluation, GC analysis, odor sensor analysis, and taste sensor analysis.

Sensory evaluation
Sensory evaluation was performed in accordance with the method used by Iwatsuki et al. (2) Researchers in their 20s and 30s belonging to the Research and Development 10 males and 10 females (in total 20 panel members) who have a keen sense of taste and habitually drink milk evaluated 13 attributes on a seven-point absolute evaluation scale, with the maximum score being +3 and the minimum score being -3.

GC analysis 2.3.1 GC/MS
Volatile compounds in each milk were extracted by a solid phase micro-extraction (SPME) method and analyzed by GC/MS.The extraction was performed by inserting manufactured by SUPELCO into the headspace of a vial containing 10 ml of a sample injected into the GC.GC was performed using an Agilent 6890 gas chromatograph, and MS was performed using an Agilent 5973 mass spectrometer.The separation column was INNOWax (cross-linked polyethylene glycol; 30 m long × 0.25 mm inner diameter, increased from 40 to 120°C at a rate of 4°C/min and from 120 to 250°C at a rate of 6°C/ data from the samples with those of the Wiley library.

GC
GC was performed using an Agilent 6890, in which a SIEVERS 355 sulfur chemiluminescense detector (SCD) was mounted.A DB-1 methyl siloxane separation was adopted as the carrier gas, and the oven temperature was raised from 35 to 260°C at a rate of 15°C/min.After warming a vial containing a 10 ml sample to 40°C and maintaining that temperature for 30 min, 5 ml of the headspace was injected into the GC.

Odor sensor analysis
A FOX3000 odor sensor (Alpha M.O.S., France) was used, on which 12 metal oxide sensors made of semiconducting materials were mounted.The odor sensors used are listed in Table 2.A special glass container holding a 4 ml sample was warmed to 37ºC and that temperature was maintained for 5 min.Then the headspace was fed to the measured.The response value was determined using the formula: response value=(maximum resistance -initial resistance) / initial resistance (1)   Measurements were conducted four times for each sample.Of these four measurements, three were analyzed.

Taste sensor analysis
The sensor's electric potential response for each sample was measured in accordance with the method of either Toko (8) or Yamada et al. (9) using a taste sensing system SA402 (Intelligent Sensor Technology, Ltd., Japan) equipped with eight electrodes made of lipid/polymer membranes.Table 3 shows the lipids used for the sensor membranes.The sensor's electric potential response for a sample was calculated as an electric potential relative to the sensor's electric potential response for a reference solution.Measurements analyzed.One lot of milk manufactured by our plant (sterilized at 140°C for 2 s) was used as the reference solution and a rinse solution.

Statistical analysis
Tukey HSD test, principal component analysis (PCA), standard deviation in the response values of the odor sensors and the taste sensors, and correlation analysis were performed using SPSS 13.0J software for Windows.Tukey HSD test was used to samples in taste sensor analysis.PCA was performed after standardization.Correlation analysis was performed to examine the relationships among the sensory evaluation, the amount of volatile compounds, and the sensor.

Sensory evaluation
1) and the four types of reconstituted milk (Samples 2, 3, 4, and 5), which were adjusted to have the same fat and SNF content as the milk (Sample 1), exhibited different features.
Comparison of samples using the same pasteurization method, in other words, comparison of the Plate-pasteurized Samples 2 and 3 (raw milk 0%, Plate), showed the 5 also showed the trend that Sample 4, containing raw milk, exhibited similarly stronger attributes.Thus, it was found that the use of raw milk can add richness.
Comparison of Samples 2 and 4 (raw milk 50%, Inf), with the same blending ratio, Tetradodecylammonium bromide, 1-tetradecanol trend of being similarly weaker.These results are the same as those in a previous study reporting that, with direct heating pasteurization, the effect of the heating was lower, As mentioned, sensory evaluation revealed that the use of different raw materials or

GC and GC/MS analysis
Figure 2 shows the results of the analysis of five compounds present in large Compared with the four types of reconstituted milk (Samples 2, 3, 4, and 5), the milk (Sample 1) showed a trend of being abundant in these compounds; in particular, it had a notable differences; for instance, the Inf-pasteurized Samples 4 and 5 had lower volatile compounds than the Plate-pasteurized Samples 2 and 3. Comparison of samples using the same pasteurization method, that is, comparison of Samples 2 and 3 and comparison of Samples 4 and 5, did not reveal any marked difference caused by the different blending ratios.Comparison of samples with the same blending ratios, that is, comparison of Samples 2 and 4 and comparison of Samples 3 and 5, revealed that the Inf-pasteurized samples had lower amounts of every compound.These results support the evaluation results of the sensory test described, namely, that the Inf-

Odor sensor analysis
The results of the odor sensor analysis are shown in Fig. 3 and Table 4. Figure 3 shows the results processed by PCA and the following varimax method of the response values of the odor sensors, narrowed down to the three sensors SY/LG, SY/GCTL, and P10/1 by excluding the sensors having high correlations.The proportion of the component (PC2) was 39%, totaling 82%.The results can be substantially accounted for by this two-dimensional diagram.The results were discriminated mainly into a Platepasteurized group and an Inf-pasteurized group.This indicates that the odor sensor analysis discriminated the difference due to the different pasteurization methods, instead of the difference due to the different blending ratios.
Table 4 shows the results of examining the correlation of these odor sensor analysis results with the sensory evaluation and volatile compounds.The results show high

Taste sensor analysis
The results of the taste sensor analysis are shown in Fig. 4, Table 5, Table 6, and Table 7. Table 5 shows the average electric potential and standard deviation in each channel of Ch.7 using Tukey HSD test.Table 6 shows the results of examining the correlation among taste sensor, sensory evaluation, and volatile compounds.Channel 5 shows high sensor response values.The proportion of the PC1 was 69%, and the proportion of the PC2 was 18%, totaling 87%.The results can be substantially accounted for by this tworeconstituted milk, were discriminated independently.milk and four types of reconstituted milk.We found that, in order to add richness, it is effective to mix raw milk and conduct Plate-type pasteurization, and in order to give a volatile compounds.Organizing the results of the odor sensor and the taste sensor analyses in relation to the results of the sensory evaluation and GC analysis, with the odor sensor we found a correlation between the amount of volatile compounds and "cooked or GC analysis, and that they can be indicated objectively.We therefore conclude that the sensor analysis used in this experimental system is system was simple.It used common milk materials.In products on the market, the combination of milk materials is more complex.Therefore, when examining such combinations, it is necessary that the examination be performed in an experimental system that contains many other milk materials.

Fig. 1 .
Fig. 1.Average score of the sensory attributes of milk and the four types of reconstituted milk by a panel of experts.* p < 0.05.

Fig. 3 .
Fig. 3. Graphical representation of principal component analysis applied to the responses for milk and four types of reconstituted milk by odor sensor.* : proportion.

Table 1
Milk and reconstituted milk samples.
The composition of each sample was 4.0% fat and 8.5% solids not fat (SNF).

Table 2
The list of odor sensors.

Table 3
Lipids used for the sensor membranes.

Table 4 the
Graphical representation of principal component analysis applied to the responses for milk and four types of reconstituted milk by taste sensor.

Table 5
Electric potential in each channel of the taste sensor.