Method and system for evaluating the antioxidant status of fish

Abstract

The application belongs to the technical field of feed analysis and data processing, and provides a fish antioxidant state evaluation method and system, specifically: first, the fish meat slice sample is collected by hyperspectral to obtain a spectral image, the spectral image is selected and divided into several segmentation zones, then the content characteristics of vitamin E in the segmentation zone are obtained, finally, the segmentation zones are graded according to the time sequence change of the content characteristics, and the antioxidant state of different samples is evaluated through the grading results. The antioxidant state grading method reflecting the antioxidant micro-mechanism is constructed to evaluate the stability, the problem that the fish antioxidant state evaluation results are difficult to be compared horizontally due to the influence of the differences in tissue structure and the inconsistency in thickness in the prior art is solved, not only the adaptability of the comparison between samples and the fairness of the antioxidant capacity evaluation are improved, but also a fast, stable and comparable evaluation tool is provided for the application scenarios such as feed regulation, transportation stress judgment and health state monitoring in fish culture.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of feed analysis and spectral data processing, and particularly relates to a fish antioxidant state evaluation method and system. BACKGROUND

[0002] In the process of fish farming, the addition of antioxidants in feed can effectively reduce oxidative stress problems and improve fish immunity and growth rate, but whether the addition amount of antioxidants meets the fish body demand range needs to be realized through fish antioxidant state evaluation. And in the field of fish transportation or disease prevention and control, it is often necessary to evaluate the antioxidant state in fish meat, which refers to the balance of the antioxidant system in the body, and specifically refers to the relative relationship between oxidative stress and antioxidant defense. When the active oxygen produced by metabolism in fish meat is excessive, it will cause health problems of oxidative damage to fish body, including lipid peroxidation, protein oxidation and DNA damage, etc. The traditional method of fish antioxidant state analysis is to evaluate by detecting R0S content and antioxidant enzyme activity, which needs to be measured in a rigorous and time-consuming laboratory environment, and the timeliness is very low. In the forefront of research rooms, high spectral equipment can be generally used to evaluate the antioxidant capacity of fish meat. Specifically, the data of oxidative stress products, antioxidant substances and tissue metabolites in fish meat slices are collected by high spectral imaging and the content is obtained by inversion, and the antioxidant performance of the evaluation sample in the whole is evaluated through the database. However, the transverse evaluation between samples has a serious misalignment characteristic. The reason for this misalignment is that even if the same parts are selected from different fish bodies for slicing, there are still significant differences in muscle fiber structure, water content and lipid content, which will directly affect the reflection and absorption of light; in addition, high spectral imaging is sensitive to tissue thickness, and the difference in sample thickness will change the absorption path length and mislead the actual content judgment. The misalignment characteristic leads to deviation in the content inversion of high spectral samples, so the direct transverse comparison of samples in the database is easy to obtain results that do not conform to the true state, and in the process of feed development, the antioxidant state of fish is often evaluated by using the average value of multiple slices to reduce the risk of data distortion, but this working form still cannot effectively establish the transverse comparison standard between different fish individual samples; or by selecting one or a group of samples with good antioxidant level as a reference, and comparing the distance or similarity of other samples with it in the high spectral feature space, this method still ignores the misalignment characteristic between samples. The method ingeniously classifies the misaligned areas in the spectral image by taking dehydroascorbic acid generated by easy oxidation of vitamin C in the air as a reference bridge, and inverses the antioxidant state substance content of the fish body through the classification results, to improve the comparison accuracy between samples. SUMMARY

[0003] The present application aims to provide a fish antioxidant state evaluation method and system to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0004] In order to achieve the above-mentioned purpose, according to an aspect of the present application, a fish antioxidant state evaluation method is provided, the method comprising the following steps:

[0005] S100, a fish slice sample is collected by hyperspectral imaging to obtain a spectral image;

[0006] S200, the spectral image is selected for white meat area and divided into several segmentation areas;

[0007] S300, the content characteristics of vitamin E in the segmentation area are obtained;

[0008] S400, each segmentation area is graded according to the time sequence change of the content characteristics;

[0009] S500, different samples are evaluated for antioxidant state through the grading results.

[0010] Further, in step S100, the method of collecting a fish slice sample by hyperspectral imaging to obtain a spectral image is: collecting a fish slice sample by hyperspectral imaging to obtain three-dimensional hyperspectral image data containing several wavebands, i.e. a spectral image, wherein the hyperspectral image data covers visible light to near-infrared wavebands to reflect the spectral characteristics related to the antioxidant state in the fish tissue.

[0011] Preferably, the fish slice sample is selected from the abdominal or dorsal muscle tissue of the fish body, spread on a non-fluorescent platform with a diffuse reflection background, and collected by line scanning or area array collection through a hyperspectral imaging device, wherein the thickness of the fish slice sample ranges between 1mm-3mm, and after the thickness is determined, the thickness of the remaining samples should be ensured to be the same to ensure the consistency of the spectral absorption path.

[0012] The imaging device used for spectral image collection has at least 100 continuous wavebands and scans in the wavelength range of 400nm-1000nm, i.e. visible light to near-infrared wavebands. Before spectral image collection, the fish slice needs to be pre-treated at a constant temperature to make it stand still in an environment of 20°C to 25°C for at least 20 minutes to reduce the interference of the external environment on the image collection results.

[0013] Further, in step S200, the method of selecting the white meat area of the spectral image and dividing it into several segmentation areas is: constructing a pseudo-color image from several key wavebands in the spectral image, and extracting the white meat area of the fish meat through the pseudo-color image;

[0014] Preferably, the contour detection operator is a findContours function in OpenCV;

[0015] The white meat region is divided into several segmentation regions based on the similarity of spectral features, so that each segmentation region corresponds to a local region in fish muscle tissue with similar spectral response characteristics.

[0016] Usually, the similarity calculation adopted by the SLIC superpixel segmentation algorithm uses the Euclidean distance. Here, it should be replaced by the SAM spectral angle distance. Because the Euclidean distance assumes that the spectral space is linearly distributed, it is not sensitive to the direction of high-dimensional spectral curves, is easily affected by light intensity or sample thickness, and the key difference in the hyperspectral image is often reflected in the spectral direction rather than the amplitude. Using a complete similarity vector rather than a neighborhood average angle distance is to avoid the problem of ignoring local microstructure by the mean method, as it may mask the contribution of key pixels to heterogeneous regions, leading to misclustering or weakening of principal component features. The similarity result ignores the principal component layer with less contribution, and retaining the complete similarity vector as a similarity measure feature can retain local diversity and microscopic feature differences, improve segmentation sharpness and spectral direction accuracy.

[0017] Further, in step S300, the method for obtaining the content feature of vitamin E in the segmentation region is: for any segmentation region, average processing or median processing is performed on the spectral curve of the pixels in the segmentation region to form the spectral curve of the segmentation region, and the content of vitamin E in the spectral curve is obtained through an inversion algorithm, which is denoted as the content feature.

[0018] The specific steps are as follows: extract the hyperspectral reflectance curve of all pixels in the segmentation region, and obtain a representative spectral curve representing the overall spectral feature of the segmentation region by performing average processing or median processing on the reflectance of each pixel at each waveband. The representative spectral curve is input into a pre-trained inversion model to obtain a predicted value of the vitamin E content of the segmentation region; the pre-trained inversion model is a common technique in hyperspectral content analysis and will not be described again.

[0019] The principle of using vitamin E as a time series observation material is as follows: firstly, vitamin E is a significant fat-soluble non-enzymatic antioxidant in fish, and is a common additive in fish feed. It can directly reflect the antioxidant defense level of fish, so it is often used as a hyperspectral indicator of fish antioxidant. Compared with other non-antioxidant indicator substances, it has the advantage of hyperspectral system integration without additional modeling.

[0020] Secondly, the content change process has significant time correlation and regional difference when vitamin E is exposed to air or oxygen concentration specific gas environment, which can be used as a bridge variable for further dynamic classification. Compared with other antioxidants, its uniformity is stable, and under this premise, it can show a controllable and slow downward trend, so it has the advantage of response variable for time series modeling.

[0021] Further, in step S400, the method of classifying each partition according to the time sequence change of the content feature is:

[0022] A time period is set as a monitoring period Dpet, Dpet∈[5, 10] hours, wherein the monitoring period ends when the content feature change tends to be static, and the content feature of each partition is obtained every 20 minutes within the monitoring period, and the time scale of obtaining the content feature is recorded as a measurement point;

[0023] In one embodiment, a salmon sample is selected, the white meat area of the fish body is selected for slicing, the slicing thickness is 2 mm, a constant temperature device is used to ensure that the sample is always in a temperature environment of 20±1°C, and the monitoring period is set to 6 hours, at which time the vitamin E content feature change tends to be insignificant;

[0024] Within the monitoring period, the content features of each partition constitute a feature sequence Ft.Ls; each Ft.Ls is subjected to Savitzky-Golay filter smoothing processing; the window size is set to 5 and the polynomial order is set to 2; the Savitzky-Golay filter is a digital filter based on local polynomial regression, and the main reason for using the Savitzky-Golay filter is that it can reduce noise while maintaining high-order moments of the signal, which means that the peak value, valley value and other characteristics of the signal can be well maintained.

[0025] At both ends of Ft.Ls, if the polynomial fitting cannot be performed because the window cannot completely cover 5 content features, the original data at both ends of Ft.Ls is retained as the fitting value.

[0026] The regression fitting value of Ft.Ls in the linear regression model and the piecewise linear model is calculated by the BIC Bayesian Information Criterion method: in the linear regression model, for any partition, assuming that the corresponding Ft.Ls conforms to the linear regression model, the least squares method is used to construct a fitting model and calculate the residual sum of squares, denoted as ARs;

[0027] The regression fitting value is obtained by inputting the residual sum of squares into the BIC method, denoted as a single estimate Fio.bic;

[0028] The calculation process of the segmented linear model fitting value is as follows: assuming that there is a turning point in the segmentation zone, two linear regression models before and after the turning point are fitted respectively by using the least square method, that is, the segmented linear model; the number of the segmented linear model is the number of the turning point plus 1;

[0029] In Ft.Ls, the measuring point at which the data before and after the measuring point follows different linear relationships is recorded as a turning point; each candidate turning point HTC can be a turning point TC; the candidate turning point is all the measuring points except the measuring points corresponding to the two ends of Ft.Ls, that is, the measuring points corresponding to the first end and the tail end cannot be candidate turning points;

[0030] Based on the Bayesian information criterion, the regression fitting values of each segmented linear model are calculated by using the BIC method, and are recorded as segmented estimates Sio.bic.

[0031] The Bayesian information criterion BIC aims to select the optimal model from the candidate models. By revealing whether the antioxidant capacity of fish tissue can maintain a stable physiological state at different rates of vitamin E consumption, the obtained oxidation breakdown rate and oxidation classification sequence can be used to reflect the dynamic changes or differences of the antioxidant defense system in different tissues of the same sample, and provide a standardized and quantitative analysis tool for the horizontal comparison between multiple samples. The use of the Bayesian information criterion to determine whether there is a significant turning point in the segmentation zone can not only enhance the comparability of the evaluation results between different tissue types, but also reduce the misalignment effect caused by natural differences between samples.

[0032] For any segmentation zone, if min(Sio.bic(HTC i ))<Fio.bic, the segmented linear model is accepted, otherwise, there is no turning point in the segmentation zone, and the linear regression model is accepted;

[0033] Wherein, min() is the minimum value function;

[0034] By calculating the regression fitting value corresponding to each candidate turning point and selecting the model corresponding to the minimum value as the optimal model, it is determined whether there is a statistically significant content change node in the segmentation zone. When the segmented linear model is better than the overall linear model, it indicates that there may be a stage change in the oxidation rate during the exposure process, such as accelerated oxidation response caused by local antioxidant depletion, tissue structure exposure or enhanced micro-zone ventilation, etc. The appearance of such turning points can reflect the turning structure of the antioxidant capacity in the tissue over time, thereby serving as a signal marker of antioxidant state imbalance. This method not only improves the fitting effect of the regional modeling, but also describes the complexity of the physiological oxidation process by using the segmented dynamic structure, which helps to identify potential oxidation-sensitive regions and supports the accurate division of different tissue states.

[0035] The absolute value of the slope of the corresponding model of each partition is calculated and recorded as the oxidation loss rate; wherein, for a piecewise linear model, the oxidation loss rate is the maximum of the absolute values of the slopes of the two segments;

[0036] The oxidation loss rates of the respective partitions are constituted into an oxidation grading sequence, and in the oxidation grading sequence, the oxidation loss rates are evenly divided into K levels according to the MinMax method;

[0037] K is the number of levels, and the oxidation loss rates are evenly divided into K equal-width intervals, and the interval width is (max.Ox-min.Ox) / K, wherein max.Ox is the maximum of the oxidation loss rates in the oxidation grading sequence, and min.Ox is the minimum of the oxidation loss rates in the oxidation grading sequence; then, the interval threshold is used to assign levels to each oxidation loss rate, and the interval is left-closed and right-open, that is, the left endpoint is included but the right endpoint is not included, but the right endpoint of the Kth interval includes the maximum of the oxidation loss rates, so as to cover the complete range;

[0038] The slopes of all partitions are sorted, and the partitions are divided into K levels according to the equal number principle, wherein the partitions with smaller absolute values of the slopes are assigned to higher levels, indicating that they have stronger antioxidant capacity, and the partitions with larger absolute values of the slopes are assigned to lower levels, indicating that they have weaker antioxidant capacity.

[0039] The absolute value of the slope of the final output of the model in each partition is defined as the oxidation loss rate, and a regional grading sequence is constructed accordingly, and in the grading process, an equal-width interval division method is used to divide the antioxidant capacity levels. The smaller the oxidation loss rate, the slower the consumption of vitamin E by the tissue per unit time, and the more likely the antioxidant system is in a steady state or a good self-regulation stage; the larger the oxidation loss rate, the more likely the tissue region is to quickly enter an oxidative stress state under exposure conditions, resulting in accelerated loss of antioxidant substances or limited defense mechanisms of the tissue. In particular, when the inflection point and the high-slope segment coexist, the tissue can be further considered to be in a critical antioxidant collapse state. Based on this rate grading method, the interpretability of the dynamic change trend is effectively enhanced, and a functional atlas with spatiotemporal resolution is constructed based on the regional nature of the spectral image, and a more biologically meaningful reference is provided for the evaluation of tissue activity.

[0040] However, in specific applications, the above-mentioned slope-based change rate grading method can quickly reflect the linear decay trend of the vitamin E content in a set time interval, and is suitable for partitions with regular antioxidant state changes and obvious trends. However, it ignores the total amount of accumulated loss of vitamin E during the entire exposure process, although it can capture the influence of linear fluctuations, it is often not suitable for tissue regions with local dramatic changes or complex antioxidant release processes. Therefore, a more preferred scheme is proposed as follows:

[0041] Preferably, in step S400, the method of grading each segmentation area according to the time sequence change of the content characteristics is: set a time period as the monitoring period TETH, TETH ∈ [5, 10] hours, and obtain the content characteristics every 10-20 minutes within the monitoring period; the content characteristics of any segmentation area within the monitoring period form a content time sequence, and the content time sequence is processed by Min-Max normalization to obtain a normalized time sequence, wherein the normalized time sequence x_norm after Min-Max normalization is obtained by the formula transformation, x is any value in the content time sequence, min(x) is the minimum value of the content time sequence, and max(x) is the maximum value of the content time sequence, and the purpose is to eliminate the asynchronous decay difference of the content time sequence of different fish segmentation areas due to different oxidation rates, and then compare and grade the segmentation areas of different fish bodies;

[0042] For any segmentation area, the absolute value of the difference between the content characteristics at each time in the normalized time sequence and the content characteristics at the first time in the reverse time direction is taken as the local subsidence Losua, the median of all local subsidence is recorded as the local subsidence median, and the local subsidence greater than the local subsidence median is accumulated to form the variable point subsidence cumulative value Dpcot of the segmentation area. This variable is used to measure the total loss level of the segmentation area in the content significant change stage;

[0043] The calculation principle of the variable point subsidence cumulative value is: the median is used to effectively exclude the interference of noise in the stationary period, so that more accurate reflection can be obtained in the rapid consumption period of the oxidation process; the calculation of the local subsidence is based on the characteristic of vitamin E in the natural law, which presents a stage decay in the oxidation process. The transition from the rapid consumption period to the stationary period is the core feature of the antioxidant kinetics. Through this way, the consumption of antioxidant substances can be accurately tracked;

[0044] For any segmentation area, the normalized time sequence is subjected to multi-scale decomposition by wavelet transform, wherein the wavelet transform uses the wavedec function of the PyWavelets library in Python; the high-frequency detail coefficients and the low-frequency approximation coefficients are separated, and the coefficients are further reconstructed to obtain the high-frequency energy and the low-frequency energy at each time, which are respectively recorded as the high-frequency content characteristics Hfrct and the low-frequency content characteristics Lfrct. Here, reconstruction refers to inverse transform processing: the high-frequency detail coefficients and the low-frequency approximation coefficients are respectively recombined into time sequences by inverse transform, so that the high-frequency energy corresponding to each time is obtained in the time sequence synthesized by the high-frequency detail coefficients, and the low-frequency energy corresponding to each time is obtained in the time sequence synthesized by the low-frequency approximation coefficients; the cumulative subsidence trend under the contribution of high and low frequency components is calculated according to the high-frequency content characteristics, the low-frequency content characteristics and the local subsidence, and is recorded as the denoising cumulative subsidence area Decus:

[0045] where t is the cumulative variable, T is the number of all monitoring time points in the monitoring period, the initial monitoring time point cannot calculate the local subsidence, so the denoising cumulative subsidence area starts to calculate from the second monitoring time point, Hfrct t and Lfrct t are the high-frequency content features and low-frequency content features corresponding to the tthtime point, respectively, Losua t is the local subsidence corresponding to the tthtime point, and max() is the maximum value function.

[0046] Hyperspectral noise and sample thickness difference are easy to introduce high-frequency interference, so the wavelet transform is used to perform multi-scale decomposition on the time series to separate the real antioxidant signal and noise, and the denoising cumulative subsidence area index is constructed to improve the grading robustness and reliability.

[0047] According to the variable point subsidence cumulative value and the denoising cumulative subsidence area, the comprehensive subsidence grading value Cprsu is calculated:

[0048] The lower quartile and the upper quartile of the variable point subsidence cumulative values of all segmentation zones are calculated. If the variable point subsidence cumulative value of a segmentation zone is greater than the upper quartile, the weight wta=0.7 is taken; if it is between the lower quartile and the upper quartile, the weight wta=0.5 is taken; if it is less than the lower quartile, the weight wta=0.3 is taken.

[0049] The quartile is not sensitive to extreme values and is suitable for the common non-normal distribution in antioxidant biological data. The quartile of the variable point subsidence cumulative value of the segmentation zone is used to automatically divide the antioxidant high-loss, medium-loss and low-loss segmentation zones, avoiding the subjective bias of manually setting the threshold. The variable point subsidence cumulative value greater than the upper quartile represents the antioxidant high-loss region and directly reflects the loss in the significant oxidation stage. At this time, the noise interference is less, the high-frequency energy in the denoising cumulative subsidence area has been suppressed, and the fish sample is in acute oxidative stress. The loss in the rapid consumption stage is the key to evaluation, so the variable point subsidence cumulative value should be trusted and given a higher weight. The variable point subsidence cumulative value between the lower quartile and the upper quartile reflects the normal fluctuation, so the complementarity of the variable point subsidence cumulative value and the denoising cumulative subsidence area should be considered, and the same weight should be given. The variable point subsidence cumulative value less than the lower quartile represents the slow consumption of the antioxidant of the healthy fish sample. The variable point subsidence cumulative value is disturbed by the noise in the stable period, while the denoising cumulative subsidence area is more reliable through multi-scale denoising, so the weight of the denoising cumulative subsidence area should be increased.

[0050] ​The segmentation area with the minimum comprehensive subsidence classification value is selected as the reference sample of high level. The ratio of the difference value between any segmentation area and the reference sample of high level to the comprehensive subsidence classification value of the segmentation area is recorded as the subsidence relative difference ratio. The density peak clustering is used to cluster the subsidence relative difference ratios of all segmentation areas to obtain K levels. The density peak clustering is realized by calling the dpc function of the pyclustering library in Python. K is a preset number, and the default value is 10.

[0051] The segmentation area cluster with a small subsidence relative difference ratio is assigned to a higher level, indicating that the difference with the reference sample of high level is small, and the antioxidant capacity is maintained well. The segmentation area cluster with a large subsidence relative difference ratio is assigned to a lower level, indicating that the difference with the reference sample of high level is large, and the antioxidant capacity decays significantly.

[0052] The segmentation area corresponding to the minimum comprehensive subsidence classification value represents the least antioxidant loss, reflecting the best state of the antioxidant system in this area. This area corresponds to healthy tissue that has not been subjected to oxidative stress, and is an ideal reference. The objective selection based on data avoids human error, and the ratio eliminates the influence of absolute value dimension, making the results of different segmentation areas comparable. Because the antioxidant capacity presents continuous or non-uniform classification, the density peak clustering adaptively determines the number of clusters, avoiding the limitation of pre-specifying K value in K-means clustering, and achieving high-robustness antioxidant state classification.

[0053] Beneficial effects: The slope method focuses on rate identification, reflecting the linear decay trend of vitamin E in a set time interval, and is suitable for classifying segmentation areas with regular antioxidant state changes and obvious trends, or classifying tissue areas with slow changes in oxidation reaction. The area method focuses on the total amount of cumulative loss, especially in research involving feed type antioxidant release. It can better handle local intense changes or complex antioxidant consumption processes by capturing the rapid consumption period of the oxidation reaction to reflect the actual loss effect of vitamin E, thereby classifying sample tissue areas. Therefore, the slope method is suitable for fish samples with small content differences obtained by spectral inversion algorithm between samples, while the area method is suitable for fish samples with obvious oxidation peak values or irregular consumption areas.

[0054] Further, in step S500, the method of evaluating the antioxidant state of different samples through the classification results is as follows: the number of segmentation areas corresponding to different levels is recorded as the total number of levels, and the top 5 representative levels are selected in descending order of the total number of levels; the intersection of the representative levels of the current sample and the standard sample is taken as the comparison level, and all segmentation areas corresponding to the comparison level are taken from the original spectra of the current sample and the standard sample as evaluation areas, and the fish antioxidant state of the two evaluation areas is evaluated by hyperspectral.

[0055] Wherein the current sample refers to the sample that needs to be used for evaluation, or the sample with the antioxidant state not labeled read from the database, and the standard sample is the benchmark sample with the antioxidant state labeled existing in the database.

[0056] The general method for evaluating the antioxidant status of fish in two evaluation regions by hyperspectral is to extract spectral data for each evaluation region and invert the antioxidant index. The core step is to input the spectral data into the previously trained inversion model to output the content prediction value of the antioxidant state related substance of interest, including MDA, GSH, CoQ10, etc., to form a feature vector set of the antioxidant state of the region. The various contents in the set are accumulated to form multiple antioxidant state indexes by PCA weighting method. The higher the index value, the better the antioxidant state.

[0057] The classification of different segmentation zones in the spectral image represents different stages of antioxidant loss, which is progressive in physiology. Therefore, hyperspectral inversion between zones of the same grade can effectively control the relative constancy of interference variables such as tissue type and oxidation stage, thereby ensuring the physiological consistency of the evaluation system and improving the scientificity and accuracy of further comparative evaluation.

[0058] Preferably, all undefined variables in the present application can be manually set thresholds if not specifically defined.

[0059] The present application also provides a fish antioxidant state evaluation system, which comprises a processor, a memory and a computer program stored in the memory and executable on the processor. The processor executes the computer program to realize the steps in the fish antioxidant state evaluation method. The fish antioxidant state evaluation system can run on desktop computers, notebook computers, palmtop computers and cloud data centers, and the executable system can include, but is not limited to, processors, memories, server clusters, and the processor executes the computer program to run in the following system units:

[0060] A spectral sample collection unit for collecting hyperspectral images of fish slice samples;

[0061] A region division unit for selecting white meat regions from the spectral image and dividing them into several segmentation zones;

[0062] A content feature inversion unit for obtaining the content feature of vitamin E in the segmentation zone;

[0063] A region classification unit for classifying each segmentation zone according to the time sequence change of the content feature;

[0064] The antioxidant assessment unit is used to assess the antioxidant status of different samples based on the grading results.

[0065] The beneficial effects of this invention are as follows: This invention provides a method and system for assessing the antioxidant status of fish. By introducing hyperspectral images to dynamically invert the vitamin E content in fish tissue, and combining this with the time-series changes in vitamin E content during oxidation, this invention constructs a method for grading antioxidant status that reflects both the microscopic mechanisms of antioxidant activity and possesses assessment stability. Taking advantage of vitamin E's role as a typical fat-soluble antioxidant in the early stages of oxidative stress in fish, and leveraging its slow decay process in tissues, which exhibits good spatial uniformity and spectral response consistency, effectively avoiding interference from factors such as thickness and tissue density differences in hyperspectral inversion, this invention proposes two strategies: the slope method and the area method. The former reflects the loss rhythm of antioxidants during stable decay by modeling the linear rate of change in content, suitable for identifying samples with continuous and stable tissue oxidation status. The latter reveals the total loss of antioxidants through integral analysis of local rapid decline phases, suitable for scenarios with significant fluctuations in oxidation reactions or complex feed release mechanisms. Both strategies can be flexibly applied according to sample characteristics, forming a multidimensional assessment framework oriented towards time-series antioxidant behavior, thereby improving the accuracy and physiological interpretability of antioxidant capacity comparisons between different samples. Attached Figure Description

[0066] The above and other features of the present invention will become more apparent from the detailed description of the embodiments shown in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals denote the same or similar elements. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort. In the drawings:

[0067] Figure 1 The diagram shows a flowchart of a method for assessing the antioxidant status of fish.

[0068] Figure 2 The diagram shows the structure of a fish antioxidant status assessment system. Detailed Implementation

[0069] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with the embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0070] like Figure 1 The diagram shown is a flowchart of a method for assessing the antioxidant status of fish. The following section will combine... Figure 1To elaborate a fish antioxidant status evaluation method according to the embodiment of the present application, the method comprises the following steps:

[0071] S100, the fish meat slice sample is collected by hyperspectral imaging to obtain a spectral image;

[0072] S200, the spectral image is selected for white meat area and divided into several segmentation areas;

[0073] S300, the content characteristics of vitamin E in the segmentation area are obtained;

[0074] S400, each segmentation area is graded according to the time sequence change of the content characteristics;

[0075] S500, different samples are evaluated for antioxidant status through the grading results.

[0076] Further, in step S100, the method for obtaining a spectral image by collecting a fish meat slice sample by hyperspectral imaging is: collecting a fish meat slice sample by hyperspectral imaging to obtain three-dimensional hyperspectral image data containing several wavebands, i.e. a spectral image, wherein the hyperspectral image data covers visible light to near-infrared wavebands to reflect the spectral characteristics related to the antioxidant status in the fish meat tissue.

[0077] Preferably, the fish meat slice sample is selected from the abdominal or dorsal muscle tissue of the fish body, spread on a non-fluorescent platform with a diffuse reflection background, and collected by line scanning or area array of a hyperspectral imaging device, wherein the thickness of the fish meat slice sample ranges between 1mm-3mm, and after the thickness is determined, the thickness of the remaining samples should be ensured to be the same to ensure the consistency of the spectral absorption path.

[0078] The imaging device used for spectral image collection has at least 100 continuous wavebands and scans in the wavelength range of 400nm-1000nm, i.e. visible light to near-infrared wavebands. Before spectral image collection, the fish meat slice needs to be pre-treated at a constant temperature to make it stand still in an environment of 20°C to 25°C for at least 20 minutes to reduce the interference of the external environment on the image collection results.

[0079] Further, in step S200, the method for selecting the white meat area of the spectral image and dividing it into several segmentation areas is: constructing a pseudo-color image from several key wavebands in the spectral image, and extracting the white meat area of the fish meat through the pseudo-color image;

[0080] The specific steps are: performing principal component analysis on the hyperspectral image to extract the first three principal component layers, the principal component layers correspond to the wavebands of 900 nm, 650 nm and 540 nm, which respectively represent the water absorption band, the fish meat absorption band and the myoglobin absorption band; constructing a pseudo-color image by selecting wavebands from each principal component layer; performing gray-scale processing on the pseudo-color image and extracting the foreground region by Otsu, identifying the connected regions in the foreground region by a contour detection operator, and excluding the connected regions that do not meet the characteristics of white meat, the excluded regions specifically including noise points, blood vessels and edge fat, and the obtained foreground region is used as the final determined white meat region.

[0081] Preferably, the contour detection operator is a findContours function in OpenCV.

[0082] The white meat region is divided into a plurality of segmentation regions based on the similarity of spectral characteristics, so that each segmentation region corresponds to a local region in the fish meat tissue having similar spectral response characteristics.

[0083] The similarity of spectral characteristics is calculated by using SAM spectral angle distance, and the specific process includes: obtaining a principal component score vector for each pixel by principal component analysis, taking the 8-neighborhood or 24-neighborhood of any pixel as its comparison pixel, calculating the SAM spectral angle distance between the pixel and each comparison pixel to form a sequence, denoted as a similarity vector, and using the similarity vector as a clustering distance item of a SLIC superpixel segmentation algorithm to perform regional segmentation on the white meat region to obtain each segmentation domain; wherein the expression of the clustering distance item of the superpixel segmentation algorithm is dis_spec(pi,pj) = 1-cos(vi,vj), pi and pj represent two pixels with serial numbers i1 and i2, and vi and vj represent the similarity vectors corresponding to the two pixels with serial numbers i1 and i2. i1 i2 i1 i2 i1 i2 i1 i2

[0084] Further, in step S300, the method for obtaining the content feature of vitamin E in the segmentation region is: for any segmentation region, performing average processing or median processing on the spectral curves of the pixels in the segmentation region to form a spectral curve of the segmentation region, obtaining the content of vitamin E in the spectral curve by an inversion algorithm, and denoted as the content feature.

[0085] ​​​​​​​​Specific steps are, extracting the hyperspectral reflectance curve of all pixels in the segmentation area, and obtaining the representative spectral curve representing the overall spectral characteristics of the segmentation area by averaging or median processing the reflectance of each pixel on each band, and inputting the representative spectral curve into the pre-trained inversion model to obtain the vitamin E content prediction value of the segmentation area; the pre-trained inversion model is a common technology in hyperspectral content analysis and will not be described again.

[0086] Further, in step S400, the method of classifying each segmentation area according to the time sequence change of the content characteristics is:

[0087] A time period is set as a monitoring period Dpet, Dpet∈[5, 10] hours, wherein the monitoring period takes the change trend of the content characteristics tending to be static as the termination condition of the monitoring period, and the content characteristics of the segmentation area are obtained every 20 minutes within the monitoring period, and the time scale of obtaining the content characteristics is recorded as a measurement point;

[0088] In one embodiment, salmon is selected to construct a sample, the white meat area of the fish body is selected for slicing, the slicing thickness is 2 mm, a constant temperature device is used to ensure that the sample is always in a temperature environment of 20±1°C, the monitoring period is set to 6 hours, and at this time the vitamin E content characteristics change tends to be insignificant;

[0089] Within the monitoring period, the content characteristics of each segmentation area constitute a feature sequence Ft.Ls; each Ft.Ls is subjected to Savitzky-Golay filtering and smoothing processing; wherein the window size is set to 5 and the polynomial order is set to 2;

[0090] At both ends of Ft.Ls, if the polynomial fitting cannot be performed due to the window being unable to completely cover 5 content characteristics, the original data at both ends of Ft.Ls is retained as the fitting value.

[0091] The regression fitting value of Ft.Ls in the linear regression model and the piecewise linear model is calculated by the BIC Bayesian information criterion method: in the calculation process of the regression fitting value corresponding to the linear regression model, for any segmentation area, it is assumed that the corresponding Ft.Ls conforms to the linear regression model, a fitting model is constructed using the least square method and the residual sum of squares is calculated, denoted as ARs;

[0092] The regression fitting value is obtained by inputting the residual sum of squares into the BIC method, denoted as a single estimate Fio.bic;

[0093] The formula of the single estimate is Fio.bic=Snp×ln(ARs / Snp)+2ln(Snp); wherein Snp is the number of elements in Ft.Ls, ln() is the logarithmic function with e as the base; the BIC Bayesian information criterion method is abbreviated as the BIC method.

[0094] Then, the calculation process of the segmented linear model fitting value is as follows: assuming that there is a turning point in the segmentation zone, two linear regression models before and after the turning point are fitted respectively by using the least square method, that is, the segmented linear model; the number of the segmented linear model is the number of the turning point plus 1;

[0095] In Ft.Ls, the measuring point at which the data before and after the measuring point follow different linear relationships is recorded as a turning point; each candidate turning point HTC can become a turning point TC; wherein, the candidate turning point is all the measuring points except the measuring points corresponding to the two ends of Ft.Ls, that is, the measuring points corresponding to the head and tail ends cannot become the candidate turning points;

[0096] Based on the Bayesian information criterion, the regression fitting value of each segmented linear model is calculated by using the BIC method, and is recorded as the segmented estimate Sio.bic;

[0097] The formula of the segmented estimate is Sio.bic(HTC i )=Snp×ln(ARs(HTC i ) / Snp)+4ln(Snp); wherein, i is the serial number of the candidate turning point in Ft.Ls, Snp is the number of elements in Ft.Ls, Sio.bic(HTC i ) and ARs(HTC i ) are the regression fitting value and the residual sum of squares of the segmented linear model when the i-th candidate turning point becomes a turning point, and ln() is the logarithmic function with e as the base;

[0098] For any segmentation zone, if min(Sio.bic(HTC i ))<Fio.bic, the segmented linear model is accepted, otherwise, there is no turning point in the segmentation zone, and the linear regression model is accepted;

[0099] Wherein, min() is the minimum value function;

[0100] The absolute value of the slope of the corresponding model of each segmentation zone is calculated and is recorded as the oxidation loss rate; wherein, for the segmented linear model, the oxidation loss rate is the maximum value of the absolute values of the slopes of the two segments;

[0101] The oxidation loss rates of each segmentation zone are used to form an oxidation grading sequence, and in the oxidation grading sequence, the oxidation loss rates are evenly divided into K levels according to the MinMax method;

[0102] wherein K is the number of grades, the oxidation loss rate is evenly divided into K equal intervals, the interval width is (max.Ox-min.Ox) / K, wherein max.Ox is the maximum value of the oxidation loss rate in the oxidation grading sequence, and min.Ox is the minimum value of the oxidation loss rate in the oxidation grading sequence; then the interval threshold is used to assign a grade to each oxidation loss rate, the interval is left-closed and right-open, that is, the left endpoint is included but the right endpoint is not included, but the right endpoint of the Kth interval includes the maximum value of the oxidation loss rate to cover the complete range;

[0103] The slope values of all the segmentation zones are sorted, and the segmentation zones are divided into K grades according to the equal number principle, wherein the segmentation zone with a smaller absolute value of the slope is assigned to a higher grade, indicating that the oxidation resistance of the segmentation zone is stronger, and the segmentation zone with a larger absolute value of the slope is assigned to a lower grade, indicating that the oxidation resistance of the segmentation zone is weaker.

[0104] Preferably, in step S400, the method for grading each segmentation zone according to the time sequence change of the content feature is as follows: a time period is set as a monitoring period TETH, TETH∈[5, 10] hours, and the content feature is obtained every 10-20 minutes in the monitoring period; the content feature of any segmentation zone in the monitoring period forms a content time sequence, and the content time sequence is subjected to Min-Max standardization processing to obtain a normalized time sequence, wherein the normalized time sequence x_norm after Min-Max standardization is obtained through the formula transformation, x is any value in the content time sequence, min(x) is the minimum value of the content time sequence, and max(x) is the maximum value of the content time sequence, and the purpose is to eliminate the asynchronous decay difference of the content time sequence of different fish segmentation zones caused by different oxidation rates, and then compare and grade the segmentation zones of different fish;

[0105] For any segmentation zone, the absolute value of the difference between the content feature at each time in the normalized time sequence and the content feature at the first time in the reverse time direction is taken as a local subsidence Losua, the median of all the local subsidence is taken as a local subsidence median, the local subsidence greater than the local subsidence median is accumulated to form a variable point subsidence cumulative value Dpcot of the segmentation zone, and the variable is used to measure the total loss level of the antioxidant in the content significant change stage of the segmentation zone;

[0106] For any partition, the multi-scale decomposition of the normalized time series is performed by wavelet transform, wherein the wavelet transform uses the wavedec function call in the PyWavelets library in Python; the high-frequency detail coefficients and the low-frequency approximation coefficients are separated, and the coefficients are further reconstructed to obtain the high-frequency energy and the low-frequency energy at each time, which are denoted as high-frequency content features Hfrct and low-frequency content features Lfrct respectively. Here, reconstruction refers to inverse transform processing: the high-frequency detail coefficients and the low-frequency approximation coefficients are respectively recombined into time series by inverse transform, so as to obtain the high-frequency energy corresponding to each time in the time series synthesized by the high-frequency detail coefficients, and the low-frequency energy corresponding to each time in the time series synthesized by the low-frequency approximation coefficients; the cumulative settlement trend under the contribution of high-frequency and low-frequency components is calculated according to the high-frequency content features, the low-frequency content features and the local subsidence, and is denoted as the denoising cumulative subsidence area Decus:

[0107] Wherein t is the cumulative variable, T is the number of all monitoring times in the monitoring period, the initial monitoring time cannot calculate the local subsidence, therefore the denoising cumulative subsidence area starts to calculate from the second monitoring time, Hfrct t and Lfrct t are the high-frequency content features and the low-frequency content features corresponding to the tth time respectively, Losua t is the local subsidence corresponding to the tth time, and max() is the maximum value function.

[0108] The comprehensive subsidence classification value Cprsu is calculated according to the variable point subsidence cumulative value and the denoising cumulative subsidence area: ;

[0109] The lower quartile and the upper quartile of the subsidence cumulative values of all variable points in the partition are calculated, if the subsidence cumulative value of a variable point in a partition is greater than the upper quartile, the weight wta=0.7 is taken; if it is between the lower quartile and the upper quartile, the weight wta=0.5 is taken; if the subsidence cumulative value of a variable point is less than the lower quartile, the weight wta=0.3 is taken;

[0110] The partition with the minimum comprehensive subsidence classification value is selected as the high-level reference sample, and the ratio of the difference value between any partition and the high-level reference sample comprehensive subsidence classification value to the comprehensive subsidence classification value of the partition is denoted as the subsidence relative difference ratio; the subsidence relative difference ratios of all partitions are clustered into K levels by using density peak clustering; wherein the density peak clustering is realized by calling the dpc function of the pyclustering library in Python; K is the preset number, and the default value is 10;

[0111] The clusters of the segmentation regions with relatively small sinking relative differences are assigned to a higher level, indicating that the clusters have small differences with the benchmark samples of the higher level and the antioxidant capacity is maintained well; and the clusters of the segmentation regions with relatively large sinking relative differences are assigned to a lower level, indicating that the clusters have large differences with the benchmark samples of the higher level and the antioxidant capacity is significantly attenuated.

[0112] Further, in step S500, the method for evaluating the antioxidant status of different samples by the grading result is as follows: the number of segmentation regions corresponding to different grades is recorded as a total amount of grading, and the top 5 representative grades are selected in descending order of each total amount of grading; the intersection of the representative grades of the current sample and the standard sample is taken as a comparison grade, and all segmentation regions corresponding to the comparison grade are taken from the original spectral images of the current sample and the standard sample as evaluation regions, and the fish antioxidant status of the two evaluation regions is evaluated by hyperspectral.

[0113] The general method for evaluating the fish antioxidant status of the two evaluation regions by hyperspectral is to extract spectral data from the evaluation regions and invert the antioxidant indicators, and the core step is to input the spectral data into the previously trained inversion model to output the content prediction value of the antioxidant status related substance of interest, including MDA, GSH, CoQ10 and the like, to form a feature vector set of the antioxidant status of the region, and various contents in the set are accumulated to form a plurality of antioxidant status indicators by the PCA weighting method, and the higher the index value is, the better the antioxidant status is.

[0114] The grade of the different segmentation regions in the spectral image represents different stages of antioxidant loss, which is progressive in physiology, so the hyperspectral inversion between the regions of the same grade can effectively control the relative constancy of interference variables such as tissue type and oxidation stage, thereby ensuring the physiological consistency of the evaluation system and improving the scientificity and accuracy of the further comparative evaluation.

[0115] The embodiment of the present application provides a fish antioxidant status evaluation system, as shown in Figure 2 The system structure diagram of the fish antioxidant status evaluation system of the embodiment of the present application is shown in the figure, and the fish antioxidant status evaluation system of the embodiment of the present application comprises a processor, a memory and a computer program stored in the memory and executable on the processor, and the processor implements the steps in the fish antioxidant status evaluation method embodiment when executing the computer program.

[0116] The system comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program running in the following system units:

[0117] The data acquisition unit is configured to acquire the spectral image by hyperspectral acquisition of the fish slice sample.

[0118] a dynamic merging unit configured to select a white muscle region of the spectral image and divide the white muscle region into a plurality of sub-regions;

[0119] a buffer pool construction unit configured to obtain a vitamin E content feature of the sub-regions;

[0120] a data transmission unit configured to grade the sub-regions according to a time sequence change of the content features.

[0121] The fish antioxidant status evaluation system can run on a desktop computer, a notebook computer, a palm computer, and a cloud server, etc. The fish antioxidant status evaluation system can run on a system which can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the example is only an example of the fish antioxidant status evaluation system, and does not constitute a limitation on the fish antioxidant status evaluation system, and can include more or less components, or combine certain components, or different components, for example, the fish antioxidant status evaluation system can also include an input / output device, a network access device, a bus, etc.

[0122] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor is a control center of the fish antioxidant status evaluation system running system, and connects each part of the fish antioxidant status evaluation system running system through various interfaces and lines.

[0123] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the fish antioxidant state evaluation system by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0124] Although the description of the present application has been quite detailed and particularly described with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present application. In addition, the present application is described above in embodiments that the inventors can foresee, and the purpose is to provide a useful description, and non-essential modifications to the present application that have not yet been foreseen can still represent equivalent modifications of the present application.

Claims

1. A method for assessing the antioxidant status of fish, characterized in that, The method includes the following steps: S100: Hyperspectral acquisition is performed on fish slice samples to obtain spectral images; S200: Select the white flesh region from the spectral image and divide it into several segmentation regions; S300, to obtain the content characteristics of vitamin E in the segmented region; S400 classifies each segmentation zone according to the temporal changes in content characteristics; S500 assesses the antioxidant status of different samples based on grading results; The time scale for acquiring content characteristics during the monitoring period is recorded as the measurement point; the content characteristics of each segment constitute the feature sequence Ft.Ls; each Ft.Ls is smoothed by Savitzky-Golay filtering; the regression fit values ​​of Ft.Ls in the linear regression model and the piecewise linear model are calculated by the BIC Bayesian information criterion method; The calculation process of the regression fitting value corresponding to the linear regression model is as follows. Assume that the corresponding Ft.Ls conforms to the linear regression model. The least squares method is used to construct a fitting model and calculate the sum of squared residuals, denoted as ARs. The regression fitting value is obtained by inputting the sum of squared residuals into the BIC method, denoted as the single estimate Fio.bic. Next, the calculation process of the regression fitting value corresponding to the piecewise linear model is as follows. Continuously assume that there is an inflection point in the segmentation area. The least squares method is used to fit two linear regression models before and after the inflection point respectively, that is, the piecewise linear model. Based on the Bayesian information criterion, the BIC method is used to calculate the regression fitting values of each piecewise linear model, denoted as the piecewise estimate Sio.bic. If the segmentation area satisfies min(Sio.bic(HTC i )) < Fio.bic, then the piecewise linear model is accepted; otherwise, the linear regression model is accepted, where Sio.bic(HTC i ) represents the regression fitting value of the piecewise linear model corresponding to the i-th candidate inflection point. Calculate the absolute value of the slope of the corresponding model for each segment and record it as the oxidation loss rate; construct an oxidation grading sequence from the oxidation loss rates of each segment, and divide the oxidation loss rates of the oxidation grading sequence into several grades on average according to the MinMax method.

2. The method for assessing the antioxidant status of fish according to claim 1, characterized in that, In step S100, the method for obtaining a spectral image by hyperspectral acquisition of the fish slice sample is as follows: hyperspectral image acquisition is performed on the fish slice sample to obtain three-dimensional hyperspectral image data containing several bands, which is the spectral image. The hyperspectral image data covers the visible to near-infrared bands to reflect the spectral characteristics related to the antioxidant state in the fish tissue.

3. The method for assessing the antioxidant status of fish according to claim 1, characterized in that, In step S200, the method for selecting the white meat region of the spectral image and dividing it into several segmentation regions is as follows: construct a pseudo-color image from several key bands in the spectral image, and extract the white meat region of the fish through the pseudo-color image; The white meat region was divided into several segments based on the similarity of spectral features.

4. The method for assessing the antioxidant status of fish according to claim 1, characterized in that, In step S300, the method for obtaining the vitamin E content characteristics in the segmented region is as follows: for any segmented region, the spectral curves of the pixels in the segmented region are averaged or medianized to form the spectral curve of the segmented region, and the vitamin E content of the spectral curve is obtained through an inversion algorithm, which is recorded as the content characteristics.

5. The method for assessing the antioxidant status of fish according to claim 1, characterized in that, In step S400, the method of classifying each segmented area according to the temporal changes of content characteristics is replaced by: the content characteristics of any segmented area during the monitoring period constitute a content time series, and the content time series is subjected to Min-Max standardization to obtain a normalized time series; the absolute value of the difference between the content characteristics at each moment in the normalized time series and the content characteristics at the first moment in the reverse time direction is taken as the local sinking amount, the median of all local sinking amounts is recorded as the local sinking median, and the local sinking amounts greater than the local sinking median are accumulated to form the variable point sinking cumulative value of the segmented area; For any segmented region, the normalized time series is decomposed into high-frequency detail coefficients and low-frequency approximation coefficients using wavelet transform at multiple scales. The high-frequency energy and low-frequency energy corresponding to each time step are obtained through reconstruction and are denoted as high-frequency content features and low-frequency content features, respectively. The distribution of the cumulative subsidence value of the strain point in the segmented region in the numerical set is used to set a weight wta. The cumulative subsidence trend under the contribution of high-frequency and low-frequency components is calculated based on the weight wta, high-frequency content features, low-frequency content features, and local subsidence and is denoted as the subsidence classification value. The segment with the smallest comprehensive subsidence grade value is selected as the benchmark sample of the higher grade. The ratio of the difference between the comprehensive subsidence grade value of any segment and the benchmark sample of the higher grade to the comprehensive subsidence grade value of the segment is recorded as the subsidence relative difference ratio. Density peak clustering is used to cluster the subsidence relative difference ratios of all segments to obtain several grades.

6. The method for assessing the antioxidant status of fish according to claim 1, characterized in that, In step S500, the method for evaluating the antioxidant status of different samples based on the grading results is as follows: the number of segmented regions corresponding to different gradings is recorded as the total grading amount, and the top 5 of each total grading amount are selected in descending order as representative gradings; the intersection of the representative gradings of the current sample and the standard sample is used as the comparison grading, and all segmented regions corresponding to the comparison gradings are extracted from the original spectra of the current sample and the standard sample as evaluation areas. The antioxidant status of fish in the two evaluation areas is evaluated by hyperspectral analysis.

7. A system for assessing the antioxidant status of fish, characterized in that, The fish antioxidant status assessment system includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the fish antioxidant status assessment method according to any one of claims 1-6. The fish antioxidant status assessment system runs on a desktop computer, a laptop computer, a handheld computer, or a cloud data center computing device.

Images