Serum lipid detection method based on Raman spectrum technology

Abstract

The invention belongs to the technical field of serum lipid detection, and provides a Raman spectrum technology-based serum lipid detection method, which comprises the following steps: serum sample collection, Raman spectrum collection, and detection identification. The construction process of the serum lipid detection model comprises the steps of sample set setting, VAP processing, spectral data acquisition, interval screening, characteristic peak identification, interference factor elimination and supervised learning training. By directly capturing the characteristic vibration spectrum of lipid molecules, high-sensitivity and non-destructive detection is realized, rapid identification and relative quantification of various lipid components can be completed, and the molecular structure and dynamic change information of the lipid can be synchronously analyzed, so that the information dimension and the detection efficiency are remarkably improved; through interval screening, characteristic peak identification and interference factor elimination, the operation is simplified, the interference is reduced, and the information dimension is improved.

Description

Technical Field

[0001] This invention relates to the field of serum lipid detection technology, and in particular to a serum lipid detection method based on Raman spectroscopy. Background Technology

[0002] Raman spectroscopy, or RS for short, is a vibrational spectroscopy analysis technique that provides fingerprint-like molecular structural information about substances, and its application potential in biomedical testing is increasingly evident. Compared with traditional biochemical methods that require complex pretreatment, Raman spectroscopy has significant advantages such as being label-free, virtually non-destructive, and insensitive to water interference, allowing for direct in-situ analysis of liquid biological samples. Therefore, new methods for serum lipid detection based on Raman spectroscopy not only hold promise as a rapid and efficient clinical screening tool but also provide a powerful new analytical approach for in-depth research into the molecular mechanisms of lipid metabolism-related diseases. Risk assessment for atherosclerotic cardiovascular disease increasingly demands a shift beyond traditional lipid indicators to the precise analysis of atherogenic lipoprotein subclasses. Abnormal lipid metabolism is a significant risk factor for cardiovascular disease, and accurate and efficient laboratory monitoring of it is crucial for disease prevention and management.

[0003] Currently, commonly used clinical serum lipid detection technologies, such as enzymatic methods, typically rely on complex sample pretreatment and labeling processes. These technologies have limitations such as long detection cycles, cumbersome operation, and susceptibility to interference. Furthermore, they are difficult to obtain information on the fine structure and dynamic changes of lipid molecules, resulting in gray areas in the test results and limiting their application in accurate diagnosis and early screening. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a serum lipid detection method based on Raman spectroscopy, which solves the problems of cumbersome operation, significant interference, and insufficient information dimension in existing technologies.

[0005] To achieve the above objectives, the present invention provides the following solution: A method for detecting serum lipids based on Raman spectroscopy, comprising: Collect serum samples from the target patients for analysis; Raman spectra of the serum sample to be analyzed are extracted to obtain the spectral data to be analyzed; The spectral data to be analyzed is input into a pre-constructed serum lipid detection model for identification to obtain serum lipid detection results; the construction process of the serum lipid detection model includes: Serum samples were collected from the research subjects to obtain a serum sample set; the serum sample set included: samples from the healthy control group and samples from the dyslipidemia group. VAP technology was used to preprocess, detect components, and separate serum subcomponents from pre-collected independent patient serum samples to obtain serum subcomponent separated samples. Set the instrument parameters, and perform Raman spectroscopy on the serum sample set and the serum subcomponent separated samples according to the instrument parameters to obtain raw Raman spectral data; Controlled experiments were conducted based on the quality control system and Raman spectroscopy techniques to obtain the characteristic Raman shift range. Based on the characteristic Raman shift range, the original Raman spectral data is used to identify characteristic peaks to obtain key characteristic Raman shifts; An interference factor elimination experiment was conducted on the key feature Raman shift to obtain the selected feature Raman shift; The partial least squares discriminant analysis model is trained in a supervised manner based on the selected feature Raman shift and the original Raman spectral data to obtain the serum lipid detection model.

[0006] Preferably, all blood lipid indicators of the subjects corresponding to the healthy control group samples are within the preset normal reference range; at least two blood lipid indicators of the subjects corresponding to the abnormal blood lipid group samples exceed the normal reference range; and the age range of the study subjects is between 25 and 70 years old.

[0007] Preferably, VAP technology is used to preprocess, detect components, and separate serum subfractions from pre-collected independent patient serum samples to obtain serum subfraction separated samples, including: 50 μL of the individual patient serum sample was mixed with 1950 μL of sample diluent to obtain a preliminary diluted solution; 1425 μL of the pre-diluted solution was mixed with 3480 μL of gradient solution of a preset density to obtain the original test solution; The original test solution is divided into several sub-test solutions by dividing it into smaller cups. The serum lipid subcomponents of the test solution were detected and extracted using VAP technology to obtain the serum subcomponent separated samples.

[0008] Preferably, the instrument parameters include: laser power of 130 nW, grating constant of 2-G600 / B500, and spectral acquisition range of 500 cm⁻¹. -1 Up to 3100cm -1 The integration time is 15 seconds, and the cumulative number of scans is 30.

[0009] Preferably, the Raman shift of the screening feature includes: 811 cm. -1 838cm -1 878cm -1 954cm -11002cm -1 1153cm -1 1279cm -1 1448cm -1 1504cm -1 1653cm -1 and 1739cm -1 .

[0010] Preferably, a controlled experiment is conducted based on the quality control system and Raman spectroscopy to obtain the characteristic Raman shift range, including: Two types of serum quality control products were selected, and the serum quality control products were serially diluted to obtain a series of multi-concentration samples; wherein, the first type of serum quality control products included: LDL-C, HDL-C, ApoA, and ApoB; the second type of serum quality control products included: TG and TC; The multi-concentration series of samples were detected using a microconfocal Raman spectroscopy system to obtain Raman spectral data; Based on the Raman spectral data, the region in serum associated with the target lipid component is determined, and the characteristic Raman shift region is obtained.

[0011] Preferably, interference factor elimination experiments are performed on the key feature Raman shifts to obtain the screened feature Raman shifts, including: Experimental groups were set up, including: a blank control group, a reagent background group, a serum matrix group, and a quality control spiked group. The blank control group consisted of ultrapure water. The reagent background group consisted of a mixture of diluent and density solution prepared according to the centrifugation protocol. The serum matrix group consisted of a mixed sample of mixed serum from target healthy individuals and a mixture of diluent and density solution. The quality control spiked group consisted of a mixed sample of lipoprotein control material and a mixture of diluent and density solution. Raman spectra are acquired for the experimental groups to obtain exclusion experimental data, and Raman shifts of the screening characteristics that do not meet the preset requirements are removed from the exclusion experimental data to obtain the screening characteristic Raman shifts.

[0012] The present invention discloses the following technical effects: This invention provides a serum lipid detection method based on Raman spectroscopy. By directly capturing the characteristic vibrational spectra of lipid molecules, it solves the problem that existing technologies usually rely on complex sample pretreatment and labeling processes, and achieves highly sensitive and non-destructive detection. Through interval screening, characteristic peak identification, and interference factor elimination, it solves the problems of cumbersome operation, obvious interference, and insufficient information dimension of existing technologies, and realizes early screening, dynamic monitoring and accurate assessment of dyslipidemia and related metabolic diseases. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of a serum lipid detection process based on Raman spectroscopy provided in an embodiment of the present invention; Figure 2 Raman overlap spectrum of lipoprotein control sample provided in the embodiments of the present invention; Figure 3 Raman γ-packing image of lipoprotein control samples provided in embodiments of the present invention; Figure 4 Raman overlap spectra of TG and TC quality control samples provided in embodiments of the present invention; Figure 5 Raman spectral density (GD) packing images of TG and TC quality control samples provided in this embodiment of the invention; Figure 6 A comparison of the average Raman spectra of the healthy population (con) and the dyslipidemia group (hlp) provided in this embodiment of the invention; Figure 7 The results of the difference analysis between the healthy group and the dyslipidemia group at the 2940 characteristic peak provided in the embodiments of the present invention; Figure 8 The overlapping Raman spectra of lipoprotein control samples in serum matrix provided in this embodiment of the invention; Figure 9 The above describes the superimposed spectra of original serum from 7 samples obtained using the VAP method provided in this embodiment of the invention. Figure 10 This is a Y-axis stacked image of 7 samples using the VAP method provided in this embodiment of the invention; Figure 11 The VAP-sample 6 provided in this embodiment of the invention contains 20 consecutive Raman spectra. Figure 12 The VAP-sample 5 provided in this embodiment of the invention contains 20 consecutive Raman spectra. Figure 13 The VAP-sample 2 provided in this embodiment of the invention contains 20 consecutive Raman spectra. Figure 14 The variation trend of the special peak value within 20 cups provided in the embodiments of the present invention; Figure 15 The overlapping spectra of various factors are provided in the embodiments of the present invention; Figure 16 The results of multivariate statistical analysis based on PLS-DA provided in the embodiments of the present invention; Figure 17 This is a PLS-DA bipolar diagram provided in an embodiment of the present invention; Figure 18 The results of 2000 permutation tests provided in this embodiment of the invention; Figure 19 The optimal component quantity verification results provided for the embodiments of the present invention; Figure 20 OPLS-DA score map provided for embodiments of the present invention; Figure 21 The core quality assessment diagram provided for embodiments of the present invention; Figure 22 The results of 2000 permutation tests provided in the embodiments of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] The purpose of this invention is to provide a serum lipid detection method based on Raman spectroscopy, which solves the problems of cumbersome operation, significant interference, and insufficient information dimension in existing technologies.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] Figure 1 This is a schematic diagram of a serum lipid detection process based on Raman spectroscopy provided in an embodiment of the present invention, as shown below. Figure 1 As shown, this invention provides a method for detecting serum lipids based on Raman spectroscopy, comprising: Step 100: Collect serum samples from the target patient for analysis; Step 200: Extract the Raman spectrum of the serum sample to be analyzed to obtain the spectral data to be analyzed; Step 300: Input the spectral data to be analyzed into a pre-constructed serum lipid detection model for identification to obtain serum lipid detection results; the construction process of the serum lipid detection model includes: Step 301: Collect serum samples from the research subjects to obtain a serum sample set; the serum sample set includes: samples from the healthy control group and samples from the dyslipidemia group; Step 302: Use VAP technology to preprocess, detect components, and separate serum subcomponents from pre-collected independent patient serum samples to obtain serum subcomponent separated samples; Step 303: Set the instrument parameters, and perform Raman spectroscopy on the serum sample set and the serum subcomponent separated samples according to the instrument parameters to obtain raw Raman spectral data; Step 304: Conduct controlled experiments based on the quality control system and Raman spectroscopy to obtain the characteristic Raman shift range; Step 305: Identify the characteristic peaks of the original Raman spectrum data according to the characteristic Raman shift range to obtain the key characteristic Raman shifts; Step 306: Conduct an interference factor elimination experiment on the key feature Raman shift to obtain the screened feature Raman shift; Step 307: Supervised learning training is performed on the partial least squares discriminant analysis model based on the selected feature Raman shift and the original Raman spectral data to obtain the serum lipid detection model.

[0019] Preferably, all blood lipid indicators of the subjects corresponding to the healthy control group samples are within the preset normal reference range; at least two blood lipid indicators of the subjects corresponding to the abnormal blood lipid group samples exceed the normal reference range; and the age range of the study subjects is between 25 and 70 years old.

[0020] Furthermore, VAP technology was used to preprocess, detect components, and separate serum subfractions from pre-collected independent patient serum samples, resulting in serum subfraction separated samples, including: 50 μL of the individual patient serum sample was mixed with 1950 μL of sample diluent to obtain a preliminary diluted solution; 1425 μL of the pre-diluted solution was mixed with 3480 μL of gradient solution of a preset density to obtain the original test solution; The original test solution is divided into several sub-test solutions by dividing it into smaller cups. The serum lipid subcomponents of the test solution were detected and extracted using VAP technology to obtain the serum subcomponent separated samples.

[0021] Preferably, the instrument parameters include: laser power of 130 nW, grating constant of 2-G600 / B500, and spectral acquisition range of 500 cm⁻¹. -1 Up to 3100cm -1 The integration time is 15 seconds, and the cumulative number of scans is 30.

[0022] Specifically, the Raman shift of the screening feature includes: 811 cm. -1 838cm -1878cm -1 954cm -1 1002cm -1 1153cm -1 1279cm -1 1448cm -1 1504cm -1 1653cm -1 and 1739cm -1 .

[0023] Preferably, a controlled experiment is conducted based on the quality control system and Raman spectroscopy to obtain the characteristic Raman shift range, including: Two types of serum quality control products were selected, and the serum quality control products were serially diluted to obtain a series of multi-concentration samples; wherein, the first type of serum quality control products included: LDL-C, HDL-C, ApoA, and ApoB; the second type of serum quality control products included: TG and TC; The multi-concentration series of samples were detected using a microconfocal Raman spectroscopy system to obtain Raman spectral data; Based on the Raman spectral data, the region in serum associated with the target lipid component is determined, and the characteristic Raman shift region is obtained.

[0024] Furthermore, interference factor elimination experiments were conducted on the key feature Raman shifts to obtain the screened feature Raman shifts, including: Experimental groups were set up, including: a blank control group, a reagent background group, a serum matrix group, and a quality control spiked group. The blank control group consisted of ultrapure water. The reagent background group consisted of a mixture of diluent and density solution prepared according to the centrifugation protocol. The serum matrix group consisted of a mixed sample of mixed serum from target healthy individuals and a mixture of diluent and density solution. The quality control spiked group consisted of a mixed sample of lipoprotein control material and a mixture of diluent and density solution. Raman spectra are acquired for the experimental groups to obtain exclusion experimental data, and Raman shifts of the screening characteristics that do not meet the preset requirements are removed from the exclusion experimental data to obtain the screening characteristic Raman shifts.

[0025] Specifically, regarding the study subjects and sample collection, this embodiment included 121 serum samples. Based on clinical lipid testing results, the samples were divided into two groups: ① Healthy control group: 60 cases, whose lipid indicators were all within the normal reference range; ② Dyslipidemia group: 61 cases, defined as having at least two routine lipid indicators (such as total cholesterol, triglycerides, and low-density lipoprotein cholesterol) exceeding the normal range. All study subjects were between 25 and 70 years old. In addition, for specific analysis, this embodiment also collected several subgroups of samples with abnormalities in a single lipid component, with each subgroup having more than 20 samples.

[0026] Preferably, VAP serum sample pretreatment and preparation. In this embodiment, serum samples from 7 independent patients were collected, and VAP technology was used to detect serum lipid subcomponents, and VAP-separated serum subcomponents were collected. Before VAP detection of blood lipids, serum samples need to be pretreated. The specific procedure is as follows: First, 50 μL of serum was mixed with 1950 μL of a dedicated sample diluent for initial dilution. Then, 1425 μL of the diluted serum was thoroughly mixed with 3480 μL of a gradient solution of a specific density, finally obtaining a total volume of 4905 μL of test solution. This process is equivalent to diluting the original serum approximately 137 times. The 4905 μL diluted serum and density solution mixture before centrifugation was divided into 20 portions, with the first 19 portions each containing 250 μL and the 20th portion containing 155 μL.

[0027] Optional, Raman spectroscopy acquisition and instrument parameters. Raman spectral data acquisition for all serum samples was performed on a microconfocal Raman spectrometer equipped with a 532 nm wavelength laser (Thermo Fisher Scientific analysis software). Key instrument parameter settings were as follows: laser power set to 130 nW to avoid sample damage; a grating with a grating constant of 2-G600 / B500 was used; the spectral acquisition range covered 500 cm⁻¹. -1 Up to 3100cm -1 The Raman shift range was determined. The data acquisition integration time for each spectral point was 15 seconds, and 30 scans were performed cumulatively to improve the signal-to-noise ratio. During measurement, an appropriate amount of serum sample was dropped onto a high-purity quartz glass slide to form droplets for analysis.

[0028] Preferably, data processing and statistical analysis are performed. Raw Raman spectral data undergoes preliminary preprocessing using the instrument's accompanying software (Thermo Fisher Analytical Software). Subsequent spectral data analysis (such as baseline correction, smoothing, and normalization) is completed in Origin 2022 software. Statistical analysis (such as inter-group comparisons and correlation analysis) is performed using GraphPad Prism 10 software. Furthermore, to explore the potential relationship between spectral features and lipid phenotypes, this embodiment utilizes the MetaboAnalyst online platform and its Python programming environment to construct multivariate statistical analysis models (such as principal component analysis and partial least squares discriminant analysis) for pattern recognition and classification prediction.

[0029] Specifically, Raman spectroscopy methodology validation based on serum lipid quality control samples. To verify the feasibility of Raman spectroscopy for the quantitative detection of serum lipid components, this embodiment first uses a quality control system for methodological investigation. Two types of serum quality control samples were selected: the first type is lipoprotein control samples, including low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), apolipoprotein A (ApoA), and apolipoprotein B (ApoB); the second type is triglyceride (TG) and total cholesterol (TC) quality control samples. Each quality control sample was rigorously serially diluted to prepare a series of samples with five different concentrations, as shown in Table 1. The samples at each concentration were detected using a microconfocal Raman spectroscopy system to obtain their Raman spectra within the characteristic wavenumber range. The results show that as the concentration of the quality control samples changes, the Raman absorption peak intensities corresponding to the characteristic vibrational modes of lipid and protein molecular structures exhibit a regular change. For example, the intensity of the Raman absorption peak at 700 cm⁻¹... -1 Up to 1800cm -1 2800cm -1 up to 3000cm -1 The characteristic peak intensities at various locations all show a good correlation with the dilution gradient, as referenced. Figures 2 to 5 Further analysis showed that within the selected concentration range, there was a clear dose-response relationship between the intensity of the characteristic peaks and the concentration of the corresponding components, and the trend was as expected. These results confirm that, under controlled conditions, Raman spectral signals have a sensitive response to changes in the concentration of major lipids and related protein components in serum, suggesting that this method has the potential for quantitative analysis. Through systematic analysis of a series of spectra, characteristic Raman shift ranges related to major lipid components in serum were preliminarily determined, such as lipoprotein components mainly around 700 cm⁻¹. -1 Up to 1800cm -1 The range of variation is the largest, while the characteristic absorption peaks of triglycerides (Total Glyceride TG) and total cholesterol (Total Cholesterol TC) are located at 2800 cm⁻¹. -1 up to 3000cm -1 This work provides crucial spectral interpretation for subsequent serum sample analysis. It lays the methodological foundation for applying Raman spectroscopy to quantify lipid components in actual serum samples and demonstrates that this technique can be used for rapid and non-destructive serum lipid profiling.

[0030] Table 1

[0031] Further, Raman spectral characteristics and analysis of clinical serum samples. Based on the completion of methodological feasibility verification, this embodiment further analyzes clinical samples to evaluate the ability of Raman spectroscopy to distinguish dyslipidemia in practical applications. A total of 121 clinical serum samples were collected and divided into a normal lipid control group (n=60con group) and a dyslipidemia group (n=61hlp group) according to clinical diagnostic criteria. The dyslipidemia group included common phenotypes such as hypercholesterolemia and hypertriglyceridemia. All samples underwent microscopic Raman spectroscopy under the same experimental conditions to obtain the raw spectral data for each individual. First, the spectra of the two groups were preprocessed (including baseline correction, normalization, etc.), and their average Raman spectra were calculated. Visual comparison revealed differences in morphology and intensity between the two average spectra in several specific wavenumber ranges, initially suggesting overall differences in their chemical composition. Figure 6 To conduct a quantitative comparison, this embodiment selected a location at 2940 cm. -1 The characteristic absorption peak at this location was analyzed, and it was mainly attributed to lipid molecules, showing a close correlation with serum lipid levels. Statistical analysis of the peak area of ​​this characteristic peak was performed between groups. The Mann-Whitney nonparametric test was used, with reference to... Figure 7 The results showed that the signal intensity at this characteristic peak was significantly higher in the dyslipidemia group than in the control group, with a statistically significant difference (P<0.05). This result indicates that serum molecular fingerprint information obtained based on Raman spectroscopy can effectively capture and quantify the differences in biochemical composition between individuals with dyslipidemia and healthy individuals. This embodiment is the first to demonstrate, through a clinical sample cohort, that Raman spectroscopy can be used to distinguish dyslipidemia states, providing direct experimental evidence for its use as a rapid, non-complex pretreatment tool for lipid screening or auxiliary diagnosis. Future work could further incorporate multivariate statistical analysis to construct discrimination models based on the full spectrum or multiple characteristic peaks to improve the accuracy and robustness of classification.

[0032] Preferably, the quantitative relationship of Raman spectroscopy in serum matrix is ​​verified. To explore the ability of Raman spectroscopy to quantitatively detect specific components in complex biological matrices and to assess the matrix effect that serum background may produce, this embodiment further designs an experiment to reflect the dose-effect relationship of serum matrix. Serum samples from healthy individuals are diluted by one-fold with an equal volume of ultrapure water to prepare a serum matrix solution, simulating the complex matrix environment of clinical samples and appropriately reducing background interference. Known concentrations of lipoprotein control products (containing LDL-C, HDL-C, ApoA, and ApoB) are accurately added to this matrix solution to prepare a series of spiked samples with eight gradient concentrations covering physiological and pathological ranges, as shown in Table 2. The samples are detected using a micro-Raman spectroscopy system with completely identical experimental conditions as described above. The characteristic absorption peaks showing inter-group differences in clinical samples (such as 700 cm⁻¹) are analyzed in detail. -1Up to 1800cm -1 (etc.). Reference Figure 8 The results showed that even in complex serum matrices, the intensity of the corresponding characteristic Raman peak exhibited a regular and monotonic change with increasing concentration gradient of the target component of the quality control material. PLS regression analysis of the characteristic peak intensity (I) and the added concentration (C) showed a good linear relationship (R²) within the selected concentration range. 2 =0.85, MSE=0.094, P<0.05). Furthermore, for the PLS regression design recovery experiment, the recoveries of 3 out of 7 components were within the normal range, and the average recoveries of the other 2 fell within the acceptable range of 80% to 120%, indicating that Raman spectroscopy can achieve relatively accurate quantitative analysis of these components in complex serum matrices. However, the recoveries of the remaining 2 components deviated from this range, suggesting that their Raman spectral signals may be significantly interfered with by the serum matrix, or that their characteristic peaks overlapped significantly with the spectra of other components in the matrix, affecting quantitative specificity. These results confirm that Raman spectroscopy combined with chemometrics (PLS) can effectively resolve the concentration information of multiple components in the serum matrix and achieve reliable quantitative assessment of some of them. Although the recoveries of some components are not ideal, the overall dose-response trend and the components with consistent recoveries confirm the basic feasibility of this method for quantitative analysis in serum matrices. This spiked experiment confirmed that, in a matrix environment simulating real serum, the Raman characteristic signals of the target lipid components still maintained their concentration-dependent response, indicating a clear dose-response relationship. This result eliminates the possibility that the complex background of serum could cause severe nonlinear interference in the quantitative detection of the target analyte, strongly demonstrating the feasibility and reliability of Raman spectroscopy for the quantitative analysis of specific lipid components in serum.

[0033] Table 2

[0034] Specifically, VAP (Vacuum-Assisted Propagation) was used to separate lipid subcomponents from sample cups using Raman spectroscopy. To further elucidate the direct correspondence between characteristic signals in Raman spectra and specific serum lipid components, and to verify its specificity in complex mixtures, this embodiment employs VAP technology for the physical separation of lipid subcomponents from serum samples. This method separates lipids based on differences in lipoprotein density and continuously collects the centrifuged liquid in elution time order, thereby obtaining a series of subsamples enriched with different lipid components. (Reference) Figure 9 and Figure 10 Raman spectroscopy was performed on seven independent serum samples, with each sample's centrifuged liquid divided into 20 aliquots. Subsequently, all components (a total of 140 aliquots) were analyzed using micro-Raman spectroscopy. (Reference) Figure 11The results showed a clear correlation between the Raman spectral signal and the subcomponents. In samples 9 through 12 of samples 1, 4, 6, and 7, characteristic absorption peaks with significantly increased intensity (such as the one at 987 cm⁻¹) were observed. -1 1137cm -1 and 1504cm -1 The peaks at these locations indicate that the main component of these samples is the enriched region of low-density lipoprotein (LDL), and the spectral characteristics are consistent with the lipid core components of LDL. Correspondingly, in samples 1 to 5 of samples 3 and 5 (expected to be enriched regions of high-density lipoprotein (HDL)) (refer to... Figure 12 If the peak pattern is detected, another distinctive characteristic peak mode will be detected. These findings demonstrate that Raman spectroscopy can effectively track and identify target lipoprotein components that have been physically separated and enriched in specific components.

[0035] Further, refer to Figure 13 None of the fractions from Sample 2 showed obvious characteristic absorption peaks. Based on the experimental procedure analysis, it is speculated that this may be because the sample was diluted too much before gradient centrifugation, resulting in the absolute concentration of the target lipoprotein being below the signal-to-noise ratio detection limit of the Raman spectrometer, thus masking the characteristic signal with background noise. To further quantify this relationship, this embodiment selected the identified characteristic peaks and plotted a trend graph of their peak intensity changes with the numbering of 20 fractions. The results show that the characteristic peak intensity exhibits a single-peak distribution within the expected elution window of the target lipoprotein, with the peak position highly consistent with the theoretical elution time and a clear peak shape. This trend strongly proves that the observed characteristic absorption peaks are indeed generated by specific lipid components (such as LDL or HDL), and the changes in their signal intensity can intuitively reflect the distribution of this component in the elution process (see reference). Figure 14 This result demonstrates for the first time at the experimental level that Raman spectroscopy can effectively identify specific lipoprotein subclasses after physical separation and obtain their corresponding molecular vibrational fingerprints. This lays a crucial experimental foundation and spectral analysis basis for subsequent direct spectral analysis of unseparated whole serum and inversely inferring the relative abundance or structural state of major lipoprotein subclasses such as LDL and HDL.

[0036] Specifically, the elimination and verification of spectral interference factors. To clarify whether the characteristic Raman absorption peaks observed in the samples separated by vertical automatic density gradient centrifugation truly reflect the component information of the target lipoprotein in the original serum, and to systematically evaluate whether the special diluent and density solution used in the centrifugation process introduce interference signals to the Raman spectral detection, this embodiment designed and carried out a series of control experiments. The experimental setup was as follows: (1) Blank control group: ultrapure water; (2) Reagent background group: a mixture of diluent and density solution prepared according to the centrifugation protocol; (3) Serum matrix group: a mixed sample of mixed serum from healthy individuals and the above-mentioned diluent-density solution mixture (simulating the serum matrix environment after centrifugation); (4) Quality control spiked group: a mixed sample of lipoprotein control sample and the above-mentioned diluent-density solution mixture. Raman spectroscopy was performed on all groups of samples under conditions completely consistent with the aforementioned experiments. Reference Figure 15 Spectral analysis results show that: firstly, the Raman spectrum of the mixture of diluent and density liquid is within the studied characteristic wavenumber range (e.g., 700 cm⁻¹). -1 Up to 1800cm -1 and 2800cm -1 up to 3100cm -1 No significant non-specific absorption peaks were generated within the sample, and its spectral profile was similar to that of the ultrapure water group, proving that neither reagent itself generates Raman signals that interfere with the target analysis. Secondly, after mixing serum or quality control samples with the reagents, all characteristic peaks appearing in the spectrum could be found in the original serum or quality control sample spectra with completely corresponding peak positions and shapes; no new peaks or significant shifts of existing characteristic peaks due to reagent addition were observed. The spectra of specific components (such as samples 9 to 12 of sample 1) collected after density gradient centrifugation and identified as having LDL or HDL characteristic peaks were directly compared with the corresponding original serum spectra without centrifugation. The results showed that both exhibited similar absorption at 987 cm⁻¹. -1 1137cm -1 1504cm -1 The key characteristic peaks showed a high degree of agreement, with peak position deviations less than the instrument resolution, and consistent relative intensity relationships. In summary, this series of control experiments confirmed the following two points: First, the diluent and density solution used in the vertical automated density gradient centrifugation method do not interfere with the Raman characteristic spectra of serum lipids, ensuring the specificity of subsequent spectroscopic analysis. Second, the characteristic absorption peaks detected from the separated fractions had spectral origins completely consistent with the target lipoprotein components in the original serum, thus directly verifying that the components enriched by this method can be reliably assigned and studied using Raman spectroscopy. This lays the methodological foundation for the direct analysis of specific lipoprotein components in complex serum based on Raman spectroscopy and eliminates key interfering factors that may be introduced by pretreatment steps.

[0037] Furthermore, multivariate statistical analysis was conducted based on characteristic spectra. This was based on the previously identified key Raman shift (811 cm⁻¹) associated with lipoprotein subclasses. -1 838cm -1 878cm -1 954cm -1 1002cm -1 1153cm -1 1279cm -1 1448cm -1 1504cm -1 1653cm -1 and 1739cm -1 This embodiment further employs multivariate statistical analysis to evaluate the overall discriminative ability of Raman spectroscopy for dyslipidemia. To construct a more discriminative predictive model, this embodiment then utilizes supervised learning. The Partial Least Squares-Discriminant Analysis (PLS-DA) model successfully and clearly distinguishes the two groups of samples in the latent variable space (see reference). Figure 16 and Figure 17 To ensure the robustness and effectiveness of the model, this embodiment underwent rigorous verification: the permutation test (n=2000 times) results showed ( Figure 18 The classification performance of the original model (e.g., Q) 2 The calculated value significantly outperformed the model constructed after random grouping (p<0.001), effectively ruling out the possibility of overfitting. Furthermore, the optimal component validation model obtained through 5-fold cross-validation maintained high levels of accuracy, sensitivity, and specificity, further confirming the model's predictive reliability (see reference). Figure 19 ).

[0038] Preferably, to maximize the extraction of classification-related information and reduce model complexity, this embodiment also establishes an Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA) model. This model systematically decomposes the variation in spectral data into group-related predictive components and group-independent orthogonal components, achieving clearer between-group separation and more intuitive interpretability (see reference). Figure 20 The evaluation of the core quality parameters of this model shows (see reference). Figure 21 ), its explanatory power (R) 2 Y) and predictive ability (Q) 2The Q value is at a moderate level, indicating that the model has a clear discriminative ability, but its predictive performance still has room for improvement and has not yet reached an extremely powerful level. 2 >0.9). Nevertheless, the permutation test (n=2000 times) results (p<0.001) indicate that the model has high statistical significance and no overfitting, making it a reliable analytical model. Figure 22 In summary, PLS-DA analysis demonstrates greater stability and advantages. A series of multivariate statistical analyses consistently show that Raman spectroscopy data based on selected lipid characteristic peaks can stably and effectively distinguish between healthy individuals and those with dyslipidemia. This provides a solid data foundation for developing Raman spectroscopy into an objective auxiliary tool for dyslipidemia diagnosis. Through the above implementation scheme, this embodiment provides a serum lipid detection method based on Raman spectroscopy, significantly improving detection sensitivity and accuracy, and providing direct experimental evidence for clinical lipid screening and auxiliary diagnosis.

[0039] The beneficial effects of this invention are as follows: This invention achieves highly sensitive and non-destructive detection by directly capturing the characteristic vibrational spectra of lipid molecules. It can not only rapidly identify and relatively quantify various lipid components, but also simultaneously analyze the molecular structure and dynamic changes of lipids, thus significantly improving the information dimension and detection efficiency. Through interval screening, characteristic peak identification, and interference factor elimination, the operation is simplified, interference is reduced, and the information dimension is improved.

[0040] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0041] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for detecting serum lipids based on Raman spectroscopy, characterized in that, include: Collect serum samples from the target patients for analysis; Raman spectra of the serum sample to be analyzed are extracted to obtain the spectral data to be analyzed; The spectral data to be analyzed is input into a pre-constructed serum lipid detection model for identification, and the serum lipid detection results are obtained. The process of constructing the serum lipid detection model includes: Serum samples were collected from the research subjects to obtain a serum sample set; the serum sample set included: samples from the healthy control group and samples from the dyslipidemia group. VAP technology was used to preprocess, detect components, and separate serum subcomponents from pre-collected independent patient serum samples to obtain serum subcomponent separated samples. Set the instrument parameters, and perform Raman spectroscopy on the serum sample set and the serum subcomponent separated samples according to the instrument parameters to obtain raw Raman spectral data; Controlled experiments were conducted based on the quality control system and Raman spectroscopy techniques to obtain the characteristic Raman shift range. Based on the characteristic Raman shift range, the original Raman spectral data is used to identify characteristic peaks to obtain key characteristic Raman shifts; An interference factor elimination experiment was conducted on the key feature Raman shift to obtain the selected feature Raman shift; The partial least squares discriminant analysis model is trained in a supervised manner based on the selected feature Raman shift and the original Raman spectral data to obtain the serum lipid detection model.

2. The method for detecting serum lipids based on Raman spectroscopy according to claim 1, characterized in that, All blood lipid indicators of the subjects in the healthy control group were within the preset normal reference range; at least two blood lipid indicators of the subjects in the dyslipidemia group were outside the normal reference range; the age range of the subjects was between 25 and 70 years old.

3. The serum lipid detection method based on Raman spectroscopy according to claim 1, characterized in that, VAP technology was used to preprocess, analyze, and separate serum subfractions from pre-collected independent patient serum samples, resulting in serum subfraction separated samples, including: 50 μL of the individual patient serum sample was mixed with 1950 μL of sample diluent to obtain a preliminary diluted solution; 1425 μL of the pre-diluted solution was mixed with 3480 μL of gradient solution of a preset density to obtain the original test solution; The original test solution is divided into several sub-test solutions by dividing it into smaller cups. The serum lipid subcomponents of the test solution were detected and extracted using VAP technology to obtain the serum subcomponent separated samples.

4. The serum lipid detection method based on Raman spectroscopy according to claim 1, characterized in that, The instrument parameters include: laser power of 130nW, grating constant of 2-G600 / B500, and spectral acquisition range of 500cm. -1 Up to 3100cm -1 The integration time is 15 seconds, and the cumulative number of scans is 30.

5. The method for detecting serum lipids based on Raman spectroscopy according to claim 1, characterized in that, The Raman shift of the screening feature includes: 811 cm. -1 838cm -1 878cm -1 954cm -1 1002cm -1 1153cm -1 1279cm -1 1448cm -1 1504cm -1 1653cm -1 and 1739cm -1 .

6. The serum lipid detection method based on Raman spectroscopy according to claim 1, characterized in that, Controlled experiments were conducted using a quality control system and Raman spectroscopy techniques to obtain characteristic Raman shift ranges, including: Two types of serum quality control products were selected, and the serum quality control products were serially diluted to obtain a series of multi-concentration samples; wherein, the first type of serum quality control products included: LDL-C, HDL-C, ApoA, and ApoB; the second type of serum quality control products included: TG and TC; The multi-concentration series of samples were detected using a microconfocal Raman spectroscopy system to obtain Raman spectral data; Based on the Raman spectral data, the region in serum associated with the target lipid component is determined, and the characteristic Raman shift region is obtained.

7. The method for detecting serum lipids based on Raman spectroscopy according to claim 1, characterized in that, Interference factor elimination experiments were conducted on the key feature Raman shifts to obtain the screened feature Raman shifts, including: Experimental groups were set up, including: a blank control group, a reagent background group, a serum matrix group, and a quality control spiked group. The blank control group consisted of ultrapure water. The reagent background group consisted of a mixture of diluent and density solution prepared according to the centrifugation protocol. The serum matrix group consisted of a mixed sample of mixed serum from target healthy individuals and a mixture of diluent and density solution. The quality control spiked group consisted of a mixed sample of lipoprotein control material and a mixture of diluent and density solution. Raman spectra are acquired for the experimental groups to obtain exclusion experimental data, and Raman shifts of the screening characteristics that do not meet the preset requirements are removed from the exclusion experimental data to obtain the screening characteristic Raman shifts.

Images