Complex sample detection method based on surface enhanced Raman spectroscopy and application

Abstract

The invention discloses a complex sample detection method based on surface enhanced Raman spectroscopy and application, and relates to the field of biomolecular detection.The method comprises the following steps that silver nanoparticles are prepared and serve as a surface enhanced Raman substrate; preparing a concentrated silver particle solution by using the silver nanoparticles; detecting surface enhanced Raman spectrum signals of the complex sample under excitation of different wavelengths; removing a background signal and noise from the obtained surface enhanced Raman spectrum signal; and averaging the surface enhanced Raman spectra obtained by processing each wavelength, and carrying out maximum and minimum normalization processing on the obtained average spectrum intensity. According to the method, richer and more comprehensive chemical information and component information of the complex samples can be obtained, more accurate recognition and classification of the complex samples can be achieved, the recognition accuracy is greatly improved, the method is convenient and rapid to operate, low in cost and high in generalization ability and practicability, and large-scale commercial application can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of biomolecule detection, and in particular to a complex sample detection method and application based on surface enhanced Raman spectroscopy. Background Art

[0002] Liquid biopsy refers to the screening or diagnosis of diseases by testing body fluids. Compared with traditional tissue biopsy, liquid biopsy has the advantages of being non- / minimally invasive, fast, and easy to monitor in real time, so it has developed rapidly in recent years. Body fluids contain a large amount of metabolic molecular information and have a high value for disease diagnosis. Unlike the genome and transcriptome, metabolic small molecules downstream of life activities most directly and rapidly reflect the immediate state of the body. At present, the detection of metabolites mainly uses mass spectrometry and nuclear magnetic resonance spectroscopy. The former is invasive, the instrument is complex and expensive, and the sample preparation and analysis are time-consuming, while the latter has low sensitivity. Therefore, it is necessary to develop a cheap, fast, non-invasive, and highly sensitive technology for the detection of biological samples.

[0003] Surface-Enhanced Raman Spectroscopy (SERS) has attracted extensive interest from researchers in the detection and structural characterization of biomolecules due to its high sensitivity, non-invasiveness and specific fingerprint. In addition to detecting pure substances, SERS is also widely used to detect complex biological samples for drug monitoring, sample identification, disease diagnosis, etc. There are two strategies for SERS detection of complex samples. One is the labeling method, which indirectly reflects the situation of the target molecule (whether it exists, what is the concentration) by detecting the Raman probe signal connected to the target molecule (antibody, aptamer, etc.); the other is the label-free method, which directly detects the Raman signal of the biological sample itself. In many cases, we do not know the information of the important target molecules in the sample. At this time, the label-free method has unique advantages in the identification of samples, and it does not require the design of the targeting molecule system. In addition, the SERS spectrum of biological samples measured by the label-free method contains rich composition information of the sample, which is more conducive to the identification of samples and the diagnosis of diseases, especially when a single or a few biomarkers cannot be well distinguished.

[0004] In some scenarios, the composition of complex biological samples varies slightly, and it is difficult to distinguish them well using a single SERS spectrum. Therefore, researchers have developed multidimensional SERS technology, which can obtain a series of SERS spectra with different spectral shapes from the same complex sample by adjusting the relevant detection variables in the SERS test. For example, the reported artificial nose-like SERS detection technology, the key lies in the design of a series of SERS substrates with different surface modifications. These substrates have high affinity with different molecules in complex samples, so a series of different SERS spectra can be obtained, reflecting complementary component molecular information. Combining these different SERS spectra, there is a higher probability of achieving more accurate sample identification or classification. However, artificial nose technology requires the design of specific SERS substrate arrays for specific samples, which greatly increases the complexity and cost of practical applications. In fact, Raman excitation wavelengths also have the potential to achieve selective molecular detection. Studies have found that when complex samples are excited with different wavelengths, the SERS spectra obtained often have different spectral shapes.

[0005] Therefore, technicians in this field are committed to developing a complex sample detection method and application based on surface enhanced Raman spectroscopy. Summary of the invention

[0006] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is that the surface enhanced Raman spectroscopy with single wavelength excitation has limited ability to identify and distinguish complex samples and cannot meet the demands of the field.

[0007] To achieve the above object, the present invention provides a complex sample detection method based on surface enhanced Raman spectroscopy, the detection method comprising the following steps:

[0008] S101: preparing silver nanoparticles, wherein the silver nanoparticles are used as surface enhanced Raman substrates;

[0009] S103: preparing a concentrated silver particle solution using the silver nanoparticles;

[0010] S105: Detect surface enhanced Raman spectroscopy signals of complex samples under excitation at different wavelengths;

[0011] S107: removing background signals and noise from the obtained surface enhanced Raman spectroscopy signals;

[0012] S109: averaging the surface enhanced Raman spectra obtained by processing each wavelength, and performing maximum and minimum normalization processing on the obtained average spectrum intensity.

[0013] Furthermore, the step S101 includes the following sub-steps:

[0014] S1011: Dissolve 12.3 mg of silver nitrate in 100 mL of ultrapure water, stirring constantly and heating to boiling;

[0015] S1012: Add 2 mL of 1% sodium citrate solution at one time and continue boiling for 1 hour;

[0016] S1013: Stir and cool to room temperature, and store at 4° C. to obtain the synthesized silver nanoparticles.

[0017] Furthermore, the step S103 includes the following sub-steps:

[0018] S1031: Obtain 1 mL of the silver nanoparticles for centrifugal concentration, wherein the speed during centrifugal concentration is set to 5000 rpm and the time is set to 10 min;

[0019] S1032: Aspirate 900 μL of supernatant;

[0020] S1033: Perform ultrasonic dispersion to obtain the concentrated silver particle solution with a concentration of 2 nM.

[0021] Furthermore, the step S105 includes the following sub-steps:

[0022] S1051: Take 10 μL of the concentrated silver particle solution, add 10 μL of the complex sample and mix by ultrasonication;

[0023] S1052: Take 10 μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the surface enhanced Raman spectroscopy signal of the sample.

[0024] Furthermore, in the step S1052, the sample area is set to 50×100 μm 2 The area is divided into 5×10 pixels, and the spectrometer 10X lens is used with a scanning step of 10μm and a single pixel acquisition time of 1 second. Lasers of different wavelengths are used for excitation to obtain the surface enhanced Raman spectrum of each pixel.

[0025] Furthermore, the wavelengths of the laser include 532nm, 638nm, 785nm and 1064nm, and the corresponding laser powers are 37.3mW, 12.67mW, 21mW and 20mW, respectively.

[0026] Furthermore, in the step S107, an adaptive iterative reweighted penalty least squares method is used to fit and remove background signals, and an SG filter is used to filter out noise, so as to make the spectral data smoother.

[0027] On the other hand, the present invention also provides an application of a complex sample detection method based on surface enhanced Raman spectroscopy, and the complex sample detection method based on surface enhanced Raman spectroscopy provided by the present invention is applied to identify and / or classify complex samples.

[0028] Furthermore, the application comprises the following steps:

[0029] S201: Detect surface enhanced Raman spectroscopy signals of complex samples of different categories under different wavelength excitation;

[0030] S203: removing background signals and noise from the obtained surface enhanced Raman spectroscopy signal, and performing normalization processing;

[0031] S205: construct training set and test set data;

[0032] S207: Use a machine learning algorithm to identify and / or classify the complex samples.

[0033] Furthermore, in the step S205, the spectral data in the training set and the test set are processed differently according to different excitation modes of surface enhanced Raman spectroscopy:

[0034] When single-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are sliding average spectra obtained under single-wavelength excitation;

[0035] When dual-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under two wavelength excitations;

[0036] When three-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under three wavelength excitation;

[0037] When four-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under four wavelength excitation.

[0038] In a preferred embodiment of the present invention, compared with the prior art, the present invention has the following beneficial technical effects:

[0039] 1. Compared with single-wavelength excitation SERS detection, multi-wavelength excitation SERS strategy can obtain richer and more comprehensive chemical and composition information of complex samples, and thus can more accurately identify and classify complex samples.

[0040] 2. Compared with the artificial nose SERS detection technology, the multi-wavelength excitation SERS detection technology does not require a series of surface enhanced Raman substrates to be specially designed for specific complex samples. It only needs to use different excitation wavelengths to obtain a series of SERS spectra with different spectral shapes, so it is easy to operate.

[0041] 3. Multi-wavelength excitation SERS detection technology is applicable to all complex samples. The same set of wavelengths can be used to excite different samples to obtain SERS signals with different spectral shapes. Unlike artificial nose SERS detection, which requires the design of different detection arrays for different samples, multi-wavelength SERS detection technology is beneficial to future clinical commercial applications.

[0042] 4. Multi-wavelength excitation SERS detection requires a SERS substrate and lasers of different wavelengths. The former can use silver nanoparticles stabilized by sodium citrate, the synthesis scheme is mature, the required raw materials are few and cheap and easy to obtain, and the latter can be used repeatedly for a long time after installation. Therefore, the multi-wavelength excitation SERS method has low cost and strong economic benefits.

[0043] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of a complex sample detection method based on surface enhanced Raman spectroscopy according to an embodiment of the present invention;

[0045] Figure 2 is a schematic diagram of Raman test results of the cell culture medium DMEM under different wavelength excitations of an embodiment of the present invention;

[0046] Figure 3 The biliverdin molecular structure of the embodiment of the present invention, the Raman test result diagram of the biliverdin molecule under different wavelength excitation, and the extinction spectrum diagram of the biliverdin aqueous solution;

[0047] Figure 4 The uracil molecular structural formula of an embodiment of the present invention, a Raman test result diagram of the uracil molecule under excitation of different wavelengths, and a schematic diagram of the extinction spectrum of a uracil aqueous solution;

[0048] Figure 5 The structural formulas of adenine and biliverdin molecules according to the embodiments of the present invention, Raman test results of the two molecules under different wavelength excitations, and extinction spectra of aqueous solutions of the two molecules are shown;

[0049] Figure 6 Schematic diagram of Raman test results of adenine-biliverdin mixed solution under different wavelength excitations in an embodiment of the present invention;

[0050] Figure 7 is a schematic diagram of Raman test results of urine supernatant samples under different wavelength excitations according to an embodiment of the present invention;

[0051] Figure 8 Schematic diagram of surface enhanced Raman spectroscopy sliding average and spectrum splicing according to an embodiment of the present invention;

[0052] Fig. 9 It is a schematic diagram of the process of applying the complex sample detection method based on surface enhanced Raman spectroscopy according to an embodiment of the present invention;

[0053] Fig.10 This is a graph of prediction accuracy results of classifying different urine supernatant samples using a random forest algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following describes several preferred embodiments of the present invention with reference to the drawings in the specification, so that the technical content is clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0055] In the drawings, components with the same structure are indicated by the same numerical reference numerals, and components with similar structures or functions are indicated by similar numerical reference numerals. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present invention does not limit the size and thickness of each component. In order to make the illustration clearer, the thickness of the components is appropriately exaggerated in some places in the drawings.

[0056] like Figure 1 As shown, the single-wavelength excitation surface enhanced Raman spectroscopy in the prior art has limited recognition and differentiation capabilities for complex samples (including but not limited to blood, urine, saliva, sweat, cerebrospinal fluid, lymph fluid, effusion, cell lysate, cell culture fluid, tissue leaching fluid, and tissue extract). The embodiment of the present invention provides a complex sample detection method based on multi-wavelength excitation surface enhanced Raman spectroscopy, which can achieve more accurate recognition and classification of complex samples compared to single-wavelength excitation. The detection method includes the following steps:

[0057] Step 1: Prepare silver nanoparticles, which serve as surface enhanced Raman substrates.

[0058] To realize multi-wavelength excitation surface enhanced Raman spectroscopy (SERS) detection, it is necessary to prepare a SERS substrate. Silver nanoparticles stabilized by sodium citrate can be used. The synthesis scheme is mature, the required raw materials are few, and they are cheap and easy to obtain.

[0059] When preparing silver nanoparticles, the following sub-steps are included:

[0060] 1. Dissolve 12.3 mg of silver nitrate in 100 mL of ultrapure water, stir constantly and heat to boiling;

[0061] 2. Add 2 mL of 1% sodium citrate solution at one time and continue boiling for 1 hour;

[0062] 3. Stir and cool to room temperature, and store at 4° C. to obtain the synthesized silver nanoparticles.

[0063] Step 2: using the silver nanoparticles to prepare a concentrated silver particle solution.

[0064] The silver nanoparticles synthesized in step 1 are nearly spherical, with a diameter of about 60 nm, uniformly dispersed, and a theoretical concentration of 0.2 nM. The concentration needs to be further concentrated. Take 1 mL of the silver nanoparticle solution and centrifuge and concentrate it. The speed during centrifugation is 5000 rpm and the centrifugation time is 10 min. After 900 μL of the supernatant is removed, ultrasonic dispersion is performed to obtain a concentrated silver particle solution with a final concentration of 2 nM.

[0065] Step 3: Detect the surface enhanced Raman spectroscopy signal of the complex sample under excitation at different wavelengths.

[0066] Take 10 μL of the concentrated silver particle solution prepared in step 2, add 10 μL of the complex sample and mix by ultrasonic. Take 10 μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the surface enhanced Raman spectroscopy (SERS) signal of the sample. During the test, the sample area is selected to be 50×100 μm 2 The area is divided into 5×10 pixels, and the spectrometer 10X lens is used. The scanning step is 10μm, the single pixel acquisition time is 1 second, and lasers of different wavelengths are used for excitation to obtain the surface enhanced Raman spectrum of each pixel. The above parameters in the test process can also be selected for different samples.

[0067] Lasers of different wavelengths can be selected according to actual needs. The wavelengths of the lasers can include 532nm, 638nm, 785nm and 1064nm, and the corresponding laser powers are 37.3mW, 12.67mW, 21mW and 20mW, respectively. In some embodiments, the wavelength and power of the laser can also select other parameters according to needs.

[0068] Step 4: Remove background signals and noise from the obtained surface enhanced Raman spectroscopy signals.

[0069] The obtained surface enhanced Raman spectroscopy signals were preprocessed, including removing background signals and noise. Specifically, the adaptive iterative reweighted penalized least squares (airPLS) fitting method was used to remove the background signal; the SG filter was used to filter out part of the noise to make the spectral data smoother.

[0070] Step 5: Average the surface enhanced Raman spectra obtained at each wavelength and perform maximum and minimum normalization on the average spectrum intensity.

[0071] The SERS spectra obtained at each wavelength were averaged, and then the average spectrum intensity was normalized to the maximum and minimum.

[0072] Compared with the prior art, the complex sample detection method based on surface enhanced Raman spectroscopy provided by the present invention uses multi-wavelength lasers to excite the surface enhanced Raman spectra of complex samples. Considering that the same complex sample has different spectral shapes under the excitation of lasers of different wavelengths, the spectral shapes of certain molecules contained in the complex sample will change with the wavelength, and the change ratio of Raman intensity between certain molecules in the complex sample with the wavelength change is different. This shows that the molecular information contained in the surface enhanced Raman spectra obtained by different excitation wavelengths is different. The method of using multi-wavelength excitation surface enhanced Raman spectroscopy can obtain more detailed and richer composition and chemical information of complex samples. Therefore, it is beneficial to more accurately identify and classify complex samples in combination with machine learning classification algorithms. Therefore, the present invention can achieve more accurate identification and classification of complex samples, and greatly improve the recognition accuracy. The method is convenient and fast to operate, low cost, and has strong generalization ability and practicality.

[0073] In addition, the present invention also provides an application of a complex sample detection method based on surface enhanced Raman spectroscopy, and the detection method provided by the present invention is applied to identify and / or classify complex samples, such as Fig. 9 and Fig.10 Compared with single-wavelength excitation SERS detection, multi-wavelength excitation SERS strategy can obtain richer and more comprehensive chemical and composition information of complex samples, and thus can more accurately identify and classify complex samples.

[0074] When the complex sample detection method based on surface enhanced Raman spectroscopy provided by the embodiment of the present invention is applied to more accurately identify and classify complex samples, the following steps are included: Fig. 9 As shown:

[0075] Step 1: Detect the surface enhanced Raman spectroscopy signals of complex samples of different categories under excitation at different wavelengths.

[0076] 532nm, 638nm and 785nm excitation can be used to collect SERS spectra of multiple complex samples.

[0077] Step 2: removing background signals and noise from the obtained surface enhanced Raman spectroscopy signals and performing normalization processing.

[0078] The SERS spectrum obtained in step 1 is normalized after removing background signals and noise. Similar to the processing method in the detection method described above, an SG filter can be used to filter out part of the noise to make the spectral data smoother, and an adaptive iterative reweighted penalized least squares (airPLS) fitting method can be used to remove the background. After completing the background signal and noise processing, the spectral intensity is normalized to the maximum and minimum. The pre-processed spectral data obtained at each wavelength of each sample is averaged to remove the random noise of the spectrum. The sliding window can be set to 20, and 1-20 SERS spectra are averaged, 2-21 are averaged, 3-22 are averaged, and so on, to obtain multiple average spectra. The average spectra of different wavelengths of the same sample are horizontally spliced ​​end to end to obtain multiple spliced ​​spectra. The schematic diagram of the sliding average and splicing of the SERS spectrum is shown in Figure 8 shown.

[0079] Step 3: Construct training set and test set data.

[0080] In order to use machine learning algorithms to achieve recognition and classification, it is necessary to construct training set and test set data. Complex samples can be randomly divided into training set and test set. The spectral data in the training set and test set are processed differently according to the different excitation modes of surface enhanced Raman spectroscopy:

[0081] 1) When single-wavelength excitation surface-enhanced Raman spectroscopy is used, the spectra in the training set and the test set are sliding average spectra obtained under single-wavelength excitation;

[0082] 2) When dual-wavelength excitation surface-enhanced Raman spectroscopy is used, the spectra in the training set and the test set are the spliced ​​spectra of the sliding average spectra obtained under the two wavelength excitations;

[0083] 3) When three-wavelength excitation surface-enhanced Raman spectroscopy is used, the spectra in the training set and the test set are the spliced ​​spectra of the sliding average spectra obtained under the three wavelength excitation;

[0084] 4) When four-wavelength excitation surface-enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing the sliding average spectra obtained under the four wavelength excitation.

[0085] Step 4: Use a machine learning algorithm to identify and / or classify the complex samples.

[0086] In this embodiment, a multi-wavelength SERS excitation method is used to classify clinical urine samples, and a random forest algorithm is used to classify the samples. The prediction accuracy of the prediction set under different wavelengths is as follows: Fig.10 As shown. Fig.10 It can be seen that as the number of wavelengths used increases, the prediction accuracy gradually improves.

[0087] When only one wavelength was used, the prediction accuracy was all below 60%;

[0088] When two wavelengths are spliced, the accuracy is improved, up to 76.0% (532nm+638nm);

[0089] When three wavelengths are used for splicing, the accuracy is further improved to 82.7%.

[0090] Therefore, compared with single-wavelength excitation, the classification accuracy of urine supernatant samples by triple-wavelength SERS excitation increased by more than 20%, indicating that the multi-wavelength SERS excitation strategy can achieve more accurate identification and classification of complex samples.

[0091] The present invention is described in detail below in conjunction with a number of preferred embodiments of the present invention.

[0092] Example 1: Detecting SERS signals of cell culture medium DMEM under excitation of different wavelengths.

[0093] DMEM is a cell culture medium containing amino acids, glucose, inorganic salts, vitamins, antibiotics and other substances. It is used to demonstrate that the SERS signals obtained from complex samples under excitation at different wavelengths have different spectral shapes.

[0094] Step 1: Synthesis of silver nanoparticles as surface-enhanced Raman substrate.

[0095] Citric acid-stabilized silver nanoparticles were prepared using a modified Lee-Meisel method. 12.3 mg of silver nitrate was dissolved in 100 mL of ultrapure water, then stirred and heated to boiling, 2 mL of 1% sodium citrate solution was added at once, and the mixture was boiled for 1 hour, then stirred and cooled to room temperature and stored at 4°C. The synthesized silver nanoparticles were nearly spherical, with a diameter of about 60 nm, uniformly dispersed, and a theoretical concentration of 0.2 nM.

[0096] 1 mL of the silver nanoparticle solution was concentrated by centrifugation at a centrifugal speed of 5000 rpm for 10 min, 900 μL of the supernatant was removed and then ultrasonically dispersed to obtain a silver nanoparticle solution with a final concentration of 2 nM.

[0097] Step 2: Detect the SERS signal of DMEM under different wavelength excitation.

[0098] Take 10 μL of the concentrated silver particle solution in step 1, add 10 μL of DMEM and mix by ultrasonic. Take 10 μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the SERS signal of the sample. During the test, select a sample area of ​​50×100 μm 2The area was divided into 5×10 pixels, and the spectrometer 10X lens was used with a scanning step of 10μm and a single pixel acquisition time of 1 second. The lasers of 532nm (laser power 37.3mW) and 638nm (laser power 12.67mW) were used for excitation to obtain the SERS spectrum of each pixel.

[0099] Step 3: Remove background signals and noise from the obtained SERS spectra.

[0100] The SG filter was used to filter out some noise to make the spectral data smoother, and the adaptive iterative reweighted penalized least squares (airPLS) fitting was used to remove the background. The 50 SERS spectra obtained at each wavelength were averaged, and then the average spectrum intensity was normalized to the maximum and minimum, as shown in Figure 2. Figure 2 As shown. Figure 2 It can be seen that the SERS signals obtained by DMEM under 532nm and 638nm excitation have different spectral shapes, and its SERS spectrum is wavelength-dependent.

[0101] Example 2: Wavelength dependence test of the SERS spectrum of a single molecule (biliverdin molecule).

[0102] The reason why the SERS spectra of the same complex sample have different spectral shapes under the excitation of lasers of different wavelengths includes: the spectral shapes of certain molecules contained in the complex sample will change with the wavelength, and these changes will be reflected in the overall SERS spectrum of the complex sample. In this embodiment, the biliverdin molecule is selected for demonstration. This molecule is an important metabolite of the human body and exists in body fluids such as blood and urine. Its structural formula is as follows Figure 3 As shown in a.

[0103] Step 1: Detect the SERS signal of biliverdin under different excitation wavelengths.

[0104] Prepare a 1mM biliverdin aqueous solution and take 10μL and add it to the concentrated silver particle solution (10μL) obtained in Example 1, and mix it by ultrasonic. Take 10μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the SERS signal of the sample. During the test, select a sample area of ​​50×100μm 2The area is divided into 5×10 pixels, and a 10X lens of the spectrometer is used. The scanning step length is 10μm, the single pixel acquisition time is 1 second, and 532nm (laser power 37.3mW) and 638nm (laser power 12.67mW) lasers are used for excitation to obtain the SERS spectrum of each pixel. The background signal and noise are removed from the obtained SERS spectrum. Specifically, the SG filter is used to filter out part of the noise to make the spectral data smoother, and the adaptive iterative reweighted penalized least squares (airPLS) fitting is used to remove the background. The 50 SERS spectra obtained by each wavelength processing are averaged, and then the average spectrum intensity obtained is normalized to the maximum and minimum, as shown in the figure. Figure 3 As shown in Figure b, the SERS signals of biliverdin molecules under 532nm and 638nm excitation have different spectral shapes. -1 The peak at 1610 cm -1 The relative intensity of the peak is higher under 638 nm excitation.

[0105] Step 2: Study the reasons why the SERS spectrum of biliverdin is wavelength-dependent.

[0106] A 500 μM aqueous solution of biliverdin was prepared and its extinction spectrum was tested using a UV-visible spectrophotometer. The extinction spectrum test range was 200-1100 nm. The maximum and minimum values ​​of the obtained extinction spectrum were normalized. The results are as follows: Figure 3 As shown in Figure c, biliverdin has an absorption peak at 666nm, which is close to the excitation wavelength of 638nm. Therefore, biliverdin will undergo molecular resonance Raman scattering under 638nm excitation, and resonance Raman scattering will cause the shape of the molecular Raman spectrum to change. In addition, biliverdin has an additional enhanced Raman peak of 1258cm under 638nm excitation. -1 These are vibration modes associated with the biliverdin chromophore, and resonance Raman scattering tends to enhance the vibration peaks associated with those chromophores, further proving that the change in the SERS spectrum of the biliverdin molecule under 532nm and 638nm excitation is likely to be caused by the molecular resonance Raman scattering effect.

[0107] This example illustrates that the SERS spectra of complex samples obtained under different wavelength excitation may contain different molecular vibration information, so the use of a multi-wavelength excitation strategy can obtain richer chemical information of the sample.

[0108] Example 3: Wavelength dependence test of the SERS spectrum of a single molecule (uracil molecule).

[0109] In this example, another molecule, uracil, is selected to demonstrate the wavelength dependence of its SERS spectrum. This molecule is an important metabolite of the human body and exists in body fluids such as blood and urine. Its structural formula is as follows: Figure 4As shown in a.

[0110] Step 1: Detect the SERS signal of uracil under different wavelength excitation.

[0111] Prepare a 10mM uracil aqueous solution and take 10μL of it and add it to the concentrated silver particle solution (10μL) obtained in Example 1, and mix it by ultrasonic. Take 10μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the SERS signal of the sample. During the test, a sample area of ​​50×100μm 2 The area was divided into 5×10 pixels, and the spectrometer 10X lens was used with a scanning step of 10μm and a single pixel acquisition time of 1 second. The lasers of 532nm (laser power 37.3mW) and 638nm (laser power 12.67mW) were used for excitation to obtain the SERS spectrum of each pixel.

[0112] The background signal and noise of the obtained SERS spectrum were removed, and the SG filter was used to filter out some noise to make the spectral data smoother. The adaptive iterative reweighted penalized least squares (airPLS) fitting was used to remove the background. The 50 SERS spectra obtained by each wavelength were averaged, and then the average spectrum intensity was normalized to the maximum and minimum, as shown in Figure 2. Figure 4 As shown in Figure b, the SERS signals of uracil molecules obtained under 532nm and 638nm excitation have different spectral shapes. -1 The peak and 1628cm -1 The peak intensity ratio of the peak is different at the two wavelengths. The peak of 1395cm under 532nm excitation is -1 Weaker than 1628cm -1 , while the opposite is true under 638nm excitation.

[0113] Step 2: Study the reasons why the SERS spectrum of uracil is wavelength-dependent.

[0114] Prepare a 2mM uracil aqueous solution and use a UV-visible spectrophotometer to test its extinction spectrum. The extinction spectrum test range is 200-1100nm. The maximum and minimum extinction spectra obtained are normalized. The results are as follows Figure 4 c. As shown in the figure, uracil has no absorption in the visible light band, so no molecular resonance Raman scattering occurs under 532nm and 638nm excitation. Therefore, the change in the SERS spectrum of uracil molecules at two wavelengths may be caused by the charge transfer effect. For systems with charge transfer between the surface enhanced Raman substrate and the adsorbed molecules, the SERS spectrum of the molecule is wavelength-dependent.

[0115] This example illustrates that the SERS spectra of complex samples obtained under different wavelength excitation may contain different molecular vibration information, so the use of a multi-wavelength excitation strategy can obtain richer chemical information of the sample.

[0116] Example 4: Wavelength dependence experiment of SERS spectrum of two-component system (adenine molecule plus biliverdin molecule).

[0117] The reason why the SERS spectra of the same complex sample have different spectral shapes when excited by lasers of different wavelengths includes: the Raman intensity changes between certain molecules in the complex sample with wavelength changes are different, which will cause the Raman peak intensity ratio between different molecules to change with wavelength. This is reflected in the SERS signal of the entire complex sample as its spectral shape shows wavelength dependence. In this embodiment, the simplest complex system - a two-component system is constructed to illustrate this. Adenine and biliverdin are selected to construct a two-component system. Both are important metabolites of the human body and exist in body fluids such as blood and urine. Their structural formulas are as follows: Figure 5 As shown in a.

[0118] Step 1: Detect the SERS signals of adenine molecules and biliverdin molecules under excitation at different wavelengths.

[0119] Prepare 1mM adenine aqueous solution and biliverdin aqueous solution respectively, and take 10μL of adenine / bililiverdin solution and add it to the concentrated silver particle solution (10μL) obtained in Example 1, and mix it by ultrasonic. Take 10μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the SERS signal of the sample. During the test, select a sample area of ​​50×100μm 2 The area was divided into 5×10 pixels, and the spectrometer 10X lens was used with a scanning step of 10μm and a single pixel acquisition time of 1 second. The lasers of 532nm (laser power 37.3mW) and 638nm (laser power 12.67mW) were used for excitation to obtain the SERS spectrum of each pixel.

[0120] The background signal and noise of the obtained SERS spectrum were removed. The SG filter was used to filter out some noise to make the spectrum data smoother. The adaptive iterative reweighted penalized least squares (airPLS) fitting was used to remove the background. The 50 SERS spectra obtained at each wavelength were averaged, and then the intensity of the average spectrum obtained at the two wavelengths was divided by the intensity of the strongest Raman peak in the 532nm average spectrum. The results are as follows: Figure 5 As shown in Figure b, the SERS intensity of the two molecules under 532nm excitation is stronger than that under 638nm, mainly due to two reasons: First, the power density of 532nm laser is greater than that of 638nm (532nm: 1.01×107mW / cm 2; 638nm: 2.40×106mW / cm 2 ), and secondly, the plasmon resonance generated by silver nanoparticles under 532nm excitation is stronger than that at 638nm (based on the extinction spectrum measured below), so the electromagnetic field enhancement factor of the particles under 532nm excitation is greater than that at 638nm. However, the decrease in the Raman intensity of adenine from 532nm to 638nm is significantly greater than that of biliverdin molecules. In order to more quantitatively demonstrate the difference between the two, the Raman intensity of adenine at 733cm -1 and biliverdin 1610cm -1 The Raman peaks of adenine and biliverdin were calculated by 532nm and 638nm respectively, which further shows that the decrease ratio of adenine Raman intensity from 532nm to 638nm is significantly higher than that of biliverdin.

[0121] Step 2: Test the extinction spectrum.

[0122] The reason for the different changes in the Raman intensity of adenine and biliverdin was studied. A 2mM adenine aqueous solution and a 500μM biliverdin aqueous solution were prepared and their extinction spectra were tested respectively. The extinction spectrum test range was 200-1100nm. The maximum and minimum intensities of the measured extinction spectra were normalized. The results are as follows: Figure 5 As shown in Figure c. As can be seen from the figure, adenine has no absorption in the visible light band, while biliverdin has an absorption peak at 666nm, which is close to the excitation wavelength of 638nm. Biliverdin will undergo molecular resonance Raman scattering under the excitation of 638nm, thus bringing additional Raman enhancement to biliverdin. Therefore, the difference in Raman intensity of biliverdin at the two wavelengths is not as large as that of adenine.

[0123] Step 3: Construct an adenine-bililiverdin two-component system and detect its SERS signal under excitation at different wavelengths.

[0124] Prepare 5mM adenine aqueous solution and 500μM biliverdin aqueous solution respectively, mix them in a volume ratio of 1:1 and mix them ultrasonically, then take 10μL and add it to the concentrated silver particle solution (10μL) obtained in Example 1, and mix them ultrasonically. Take 10μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the SERS signal of the sample. During the test, a sample area of ​​50×100μm 2 The area was divided into 5×10 pixels, and the spectrometer 10X lens was used with a scanning step of 10μm and a single pixel acquisition time of 1 second. The lasers of 532nm (laser power 37.3mW) and 638nm (laser power 12.67mW) were used for excitation to obtain the SERS spectrum of each pixel.

[0125] The background signal and noise of the obtained SERS spectrum were removed. The SG filter was used to filter out some noise to make the spectral data smoother, and the adaptive iterative reweighted penalized least squares (airPLS) fitting was used to remove the background. The 50 SERS spectra obtained by each wavelength were averaged, and then the average spectrum intensity was normalized to the maximum and minimum, as shown in Figure 2. Figure 6 As shown in the figure, the SERS spectra of the two-component system under the two wavelengths of excitation are obviously different. Specifically, the 733cm -1 The Raman peaks of biliverdin and 1610 cm -1 The intensity ratio of the Raman peaks is different at the two wavelengths. The former is stronger than the latter under 532nm excitation, while the opposite is true under 638nm excitation.

[0126] This example illustrates that the SERS spectra of complex samples obtained under different wavelength excitation may contain different molecular composition information, so the use of a multi-wavelength excitation strategy can obtain richer composition information of the sample.

[0127] Example 5: Detection of SERS signals of clinical urine samples under excitation of different wavelengths.

[0128] Step 1: Prepare urine supernatant sample.

[0129] The urine supernatant contains a large number of metabolic molecules, which can reflect the relevant metabolic information of the human body. The clinical urine sample obtained from the hospital came from a patient with T1 bladder cancer. T1 bladder cancer means that the tumor is limited to the mucosal layer and no muscle layer infiltration occurs. The urine is morning urine and is stored at -80°C after collection. Before testing, the frozen sample is thawed at room temperature, and after ultrasonic mixing, 1 mL is taken for two-step centrifugation, that is, first centrifuged at 500g for 10 minutes, and then immediately centrifuged at 4000g for 10 minutes. Carefully aspirate the supernatant after centrifugation to obtain the urine supernatant sample.

[0130] Step 2: Detect the SERS signal of urine supernatant under different wavelength excitation.

[0131] 10 μL of urine supernatant was added to the concentrated silver particle solution (10 μL) obtained in Example 1 and mixed by ultrasound. 10 μL of the mixture solution was placed in a quartz capillary and the SERS signal of the sample was collected using a confocal Raman spectrometer. During the test, a sample area of ​​60×500 μm was selected. 2The area is divided into 6×50 pixels, and a 10X lens of the spectrometer is used. The scanning step is 10μm, the single pixel acquisition time is 1 second, and 532nm (laser power 37.3mW), 638nm (laser power 12.67mW) and 785nm (laser power 21mW) lasers are used for excitation to obtain the SERS spectrum of each pixel. The background signal and noise are removed from the obtained SERS spectrum. Specifically, the SG filter is used to filter out part of the noise to make the spectral data smoother, and the adaptive iterative reweighted penalized least squares (airPLS) fitting is used to remove the background. The 300 SERS spectra obtained by each wavelength processing are averaged, and then the average spectrum intensity obtained is normalized to the maximum and minimum, as shown in FIG. Figure 7 As shown in the figure, the SERS spectra of urine supernatant samples obtained under three wavelengths of excitation have different shapes, indicating that the three wavelengths detect complementary chemical information of the samples.

[0132] Example 6: Classification of clinical urine samples using multi-wavelength SERS excitation method.

[0133] Step 1: Detect the SERS signals of urine supernatant samples of different categories under excitation at different wavelengths.

[0134] A total of 58 clinical urine samples were obtained from the hospital, from 18 healthy people, 25 patients with T1 bladder cancer, and 15 patients with T2 bladder cancer. T1 bladder cancer refers to tumors confined to the mucosal layer without muscle layer infiltration; T2 bladder cancer refers to tumors that have infiltrated the bladder muscle layer. For each urine sample, the method of step 1 in Example 5 was used to prepare urine supernatant. Then, the method of step 2 in Example 5 was used to excite each urine supernatant sample at 532nm, 638nm, and 785nm, respectively, and 300 SERS spectra were collected for each.

[0135] Step 2: Data preprocessing.

[0136] The SERS spectrum obtained in step 1 is normalized after removing background signals and noise. The SG filter is used to filter out part of the noise to make the spectral data smoother. The adaptive iterative reweighted penalized least squares (airPLS) fitting is used to remove the background, and then the spectral intensity is normalized to the maximum and minimum. The 300 pre-processed spectra obtained at each wavelength for each sample are averaged to remove random noise in the spectrum, and the sliding window is 20. Average 1-20 SERS spectra, average 2-21, average 3-22, and so on, to obtain a total of 281 average spectra. The average spectra of different wavelengths of the same sample are horizontally spliced ​​end to end to obtain a total of 281 spliced ​​spectra. The schematic diagram of the sliding average and splicing of SERS spectra is shown in Figure 8 shown.

[0137] Step 3: Use machine learning algorithms to classify samples.

[0138] The 58 urine samples were randomly divided into training and test sets. The training set included 14 healthy subjects (14×281=3934 SERS spectra), 20 bladder cancer T1 cases (20×281=5620 SERS spectra) and 12 bladder cancer T2 cases (12×281=3372 SERS spectra). The test set included the remaining 4 healthy subjects (4×281=1124 SERS spectra), 5 bladder cancer T1 cases (5×281=1405 SERS spectra) and 3 bladder cancer T2 cases (3×281=843 SERS spectra). When using single-wavelength excitation SERS data for classification, the training / testing spectrum is the sliding average spectrum obtained under single wavelength excitation, with a total of 3 cases, including 532nm, 638nm and 785nm; when using dual-wavelength SERS data for classification, the training / testing spectrum is the spliced ​​spectrum of the sliding average spectrum obtained under two wavelength excitation, with a total of 3 cases, including 532nm+638nm, 532nm+785nm and 638nm+785nm; when using three-wavelength SERS data for classification, the training / testing spectrum is the spliced ​​spectrum of the sliding average spectrum obtained under three wavelength excitation, with a total of 1 case, namely 532nm+638nm+785nm. Therefore, a total of 7 wavelength usage cases are included.

[0139] The random forest algorithm is used to classify the samples. The prediction accuracy of the prediction set under different wavelength usage is as follows: Fig.10 As shown. As can be seen from the figure, as the number of wavelengths used increases, the prediction accuracy gradually increases. When only one wavelength is used, the prediction accuracy is less than 60%; when two wavelengths are spliced, the accuracy is improved, with the highest being 76.0% (532nm+638nm); when three wavelengths are spliced, the accuracy is further improved to 82.7%. Therefore, compared with single-wavelength excitation, three-wavelength SERS excitation improves the classification accuracy of urine supernatant samples by more than 20%, indicating that the multi-wavelength SERS excitation strategy can achieve more accurate identification and classification of complex samples.

[0140] Compared with single-wavelength excitation SERS detection, the multi-wavelength excitation SERS strategy provided by the present invention can obtain richer and more comprehensive chemical information and composition information of complex samples, and thus can more accurately identify and classify complex samples.

[0141] Compared with the artificial nose SERS detection technology, the multi-wavelength excitation SERS detection technology does not require a series of surface enhanced Raman substrates to be specially designed for specific complex samples. It only needs to use different excitation wavelengths to obtain a series of SERS spectra with different spectral shapes, so it is easy to operate.

[0142] To realize multi-wavelength excitation SERS detection, it is necessary to prepare a SERS substrate and install lasers of different wavelengths. The former can use silver nanoparticles stabilized by sodium citrate. The synthesis scheme is mature, the required raw materials are few, cheap and easy to obtain. The latter only needs to be installed and can be used repeatedly for a long time. Therefore, the multi-wavelength excitation SERS method has low cost and strong economic benefits.

[0143] Multi-wavelength excitation SERS detection technology is applicable to all complex samples. The same set of wavelengths can be used to excite different samples to obtain SERS signals with different spectral shapes. Unlike artificial nose SERS detection, which requires the design of different detection arrays for different samples, multi-wavelength SERS detection technology is beneficial to future clinical commercial applications.

[0144] In summary, the complex sample detection method based on multi-wavelength excitation surface enhanced Raman spectroscopy provided by the present invention is expected to achieve large-scale commercial applications in the future.

[0145] The preferred specific embodiments of the present invention are described in detail above. It should be understood that ordinary technicians in the field can make many modifications and changes based on the concept of the present invention without creative work. Therefore, all technical solutions that can be obtained by technicians in the technical field based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A complex sample detection method based on surface enhanced Raman spectroscopy, characterized in that: The detection method comprises the following steps: S101: preparing silver nanoparticles, wherein the silver nanoparticles are used as surface enhanced Raman substrates; S103: preparing a concentrated silver particle solution using the silver nanoparticles; S105: Detect surface enhanced Raman spectroscopy signals of complex samples under excitation at different wavelengths; S107: removing background signals and noise from the obtained surface enhanced Raman spectroscopy signals; S109: averaging the surface enhanced Raman spectra obtained by processing each wavelength, and performing maximum and minimum normalization processing on the obtained average spectrum intensity.

2. The detection method according to claim 1, characterized in that The step S101 includes the following sub-steps: S1011: Dissolve 12.3 mg of silver nitrate in 100 mL of ultrapure water, stirring constantly and heating to boiling; S1012: Add 2 mL of 1% sodium citrate solution at one time and continue boiling for 1 hour; S1013: Stir and cool to room temperature, and store at 4° C. to obtain the synthesized silver nanoparticles.

3. The detection method according to claim 2, characterized in that The step S103 includes the following sub-steps: S1031: Obtain 1 mL of the silver nanoparticles for centrifugal concentration, wherein the speed during centrifugal concentration is set to 5000 rpm and the time is set to 10 min; S1032: remove 900 μL of supernatant; S1033: Perform ultrasonic dispersion to obtain the concentrated silver particle solution with a concentration of 2 nM.

4. The detection method according to claim 3, characterized in that The step S105 includes the following sub-steps: S1051: Take 10 μL of the concentrated silver particle solution, add 10 μL of the complex sample and mix by ultrasonication; S1052: Take 10 μL of the mixture solution and place it in a quartz capillary, and use a confocal Raman spectrometer to collect the surface enhanced Raman spectroscopy signal of the sample.

5. The detection method according to claim 4, characterized in that: In the step S1052, the sample area is set to 50×100 μm 2 The area is divided into 5×10 pixels, and the spectrometer 10X lens is used with a scanning step of 10μm and a single pixel acquisition time of 1 second. Lasers of different wavelengths are used for excitation to obtain the surface enhanced Raman spectrum of each pixel.

6. The detection method according to claim 5, characterized in that The wavelengths of the laser include 532nm, 638nm, 785nm and 1064nm, and the corresponding laser powers are 37.3mW, 12.67mW, 21mW and 20mW respectively.

7. The detection method according to claim 6, characterized in that In the step S107, an adaptive iterative reweighted penalty least squares method is used to fit and remove background signals, and an SG filter is used to filter out noise to make the spectral data smoother.

8. An application of a complex sample detection method based on surface enhanced Raman spectroscopy, characterized in that: The detection method as described in any one of claims 1 to 7 is applied to identify and / or classify complex samples.

9. The use according to claim 8, characterized in that The application comprises the following steps: S201: Detect surface enhanced Raman spectroscopy signals of complex samples of different categories under different wavelength excitation; S203: removing background signals and noise from the obtained surface enhanced Raman spectroscopy signal, and performing normalization processing; S205: construct training set and test set data; S207: Use a machine learning algorithm to identify and / or classify the complex samples.

10. The use according to claim 9, characterized in that In the step S205, the spectral data in the training set and the test set are processed differently according to different excitation modes of surface enhanced Raman spectroscopy: When single-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are sliding average spectra obtained under single-wavelength excitation; When dual-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under two wavelength excitations; When three-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under three wavelength excitation; When four-wavelength excitation surface enhanced Raman spectroscopy is used, the spectra in the training set and the test set are spliced ​​spectra obtained by splicing sliding average spectra obtained under four wavelength excitation.

Images