Multi-channel Raman spectrum analysis system for tumor marker detection

Abstract

The invention belongs to the technical field of marker detection, and discloses a multichannel Raman spectrum analysis system for tumor marker detection, which comprises a preprocessing module, an acquisition module, an enhancement processing module, an analysis module and an output module, the pretreatment module is used for uniformly coating the surface of a target tissue slice with an electromagnetic nano material; the acquisition module is used for acquiring a Raman spectrum of a target tissue slice, and the Raman spectrum comprises a plurality of Raman scattering lights with different wavelengths; the enhancement processing module is used for converting the Raman spectrum acquired by the acquisition module into an electric signal, and is also used for enhancing the electric signal; the analysis module is used for identifying a tumor marker by analyzing the enhanced electric signal; and the output module is used for marking a tumor marker according to an identification result of the analysis module, and is also used for outputting a target tissue pathological image marked with the tumor marker.

Description

Technical Field

[0001] The present invention belongs to the technical field of biomarker detection, and particularly relates to a multi-channel Raman spectroscopy analysis system for tumor biomarker detection. Background Art

[0002] Tumor biomarkers are substances synthesized and released by tumor cells or generated by the body's response to tumor cells during the occurrence and growth of tumors, including proteins, enzymes, gene products, metabolites, etc. The detection of tumor biomarkers is of crucial significance for the early diagnosis, treatment, and prognosis evaluation of malignant tumors.

[0003] At present, malignant tumors have become the leading cause of death among Chinese residents due to illness. Therefore, developing a method that can sensitively, accurately, and specifically detect tumor biomarkers while meeting the requirements of simple, rapid, and non-invasive diagnosis has extremely important clinical value. However, existing tumor biomarker detection methods such as enzyme-linked immunosorbent assay (ELISA), polymerase chain reaction (PCR), fluorescence method, electrochemical method, and mass spectrometry, although widely used in clinical practice, have a series of disadvantages, such as insufficient sensitivity, complex operation, long detection cycle, etc., and most of these methods are invasive detections, which are not conducive to the early screening of low-concentration tumor biomarkers.

[0004] Raman spectroscopy techniques, including surface-enhanced Raman scattering (SERS), tip-enhanced Raman spectroscopy (TERS), and stimulated Raman scattering (SRS), are powerful label-free tools that can provide excellent dynamic performance and high-spatial-resolution imaging in the fields of biology and medicine. Despite these advantages of Raman spectroscopy techniques, due to the weakness of spontaneous Raman signals, slow data acquisition speed, and high cost of equipment, these factors limit their wide application in clinical practice.

[0005] Protein molecules are an important type of tumor biomarker. However, surface-enhanced Raman spectroscopy (SERS) performs poorly in directly detecting high-molecular-weight biomolecules, manifested as the overlapping of spectral bands of complex biomacromolecules and limited selectivity for proteins.

[0006] In addition, current algorithms for Raman spectroscopy reconstruction are mainly based on the reconstruction of spectral reflectance curves, such as the pseudo-inverse method, Wiener estimation method, and finite-dimensional model method, etc. These methods are all established on the baseline of a linear system and have weak resistance to non-linear interference.

[0007] Therefore, it is particularly important to develop a calibration-reconstruction algorithm specifically applicable to Raman spectroscopy, which can further improve the accuracy and specificity of detection and achieve sensitive detection of trace tumor biomarkers. Summary of the Invention

[0008] In view of this, to solve the problems raised in the above background technology, the purpose of the present invention is to provide a multi-channel Raman spectroscopy analysis system for tumor marker detection.

[0009] To achieve the above purpose, the present invention provides the following technical solutions:

[0010] A multi-channel Raman spectroscopy analysis system for tumor marker detection, including a pretreatment module, a collection module, an enhancement processing module, an analysis module, and an output module;

[0011] The pretreatment module is used to modify the target tissue section by magnetic nanoparticle aptamer modification;

[0012] The collection module is used to collect the Raman spectrum of the target tissue section, and the Raman spectrum includes Raman scattered light of multiple different wavelengths;

[0013] The enhancement processing module is used to convert the Raman spectrum collected by the collection module into an electrical signal, and is also used to perform enhancement processing on the electrical signal;

[0014] The analysis module is used to identify tumor markers by analyzing the enhanced electrical signal;

[0015] The output module is used to mark tumor markers according to the identification result of the analysis module, and is also used to output the target tissue pathological image marked with tumor markers.

[0016] Preferably, the pretreatment steps of the pretreatment module include:

[0017] Select an aptamer based on the tumor markers in the target tissue section;

[0018] Modify magnetic nanoparticles with the aptamer;

[0019] Coat a layer of chitosan on the surface of the modified magnetic nanoparticles as a capture probe;

[0020] Incubate the target tissue section with the capture probe to form an aptamer-target tissue section-aptamer sandwich complex;

[0021] Magnetically separate and wash the sandwich complex with a buffer solution, and then transfer it to a SERS substrate.

[0022] Preferably, the step of modifying magnetic nanoparticles with the aptamer includes:

[0023] Rinse the streptavidin-coated magnetic nanoparticles with a buffer solution, and then resuspend them in the buffer solution to obtain a suspension;

[0024] Add the aptamer to the suspension and react for 30 min under smooth oscillation conditions at room temperature;

[0025] Magnetically separate to obtain magnetic nanoparticles with the aptamer immobilized on the surface.

[0026] Preferably, the acquisition module includes a spectral acquisition device provided with a plurality of independent optical channels, and a laser source, a spectroscope, a detector, and a TE cooling module are provided in each independent optical channel.

[0027] Preferably, the enhancement processing module enhances the electrical signal through a Raman signal separation algorithm based on a convolutional neural network.

[0028] Preferably, the expression of the Raman signal separation algorithm based on a convolutional neural network is:

[0029] R = L * B + N; where R is the measured spectral signal input, L represents the independent spectral peak signal of the Raman spectrum, B represents the instrument broadening signal and the background baseline signal, N represents the measurement noise, and "*" represents the convolution operation.

[0030] Preferably, the independent spectral peak signal L of the Raman spectrum is expressed as:

[0031]

[0032] where x represents the Raman shift, c represents the center of the Lorentz function, wL represents the full width at half maximum of the Lorentz function, and S represents the full spectral range area of the Lorentz function;

[0033] Preferably, the instrument broadening signal and the background baseline signal B are expressed as:

[0034]

[0035] where w G represents the full width at half maximum of the Gaussian function .

[0036] Preferably, the tumor markers to be identified include:

[0037] Construct a Raman spectrum database of tissue sections containing tumor markers, obtain a spectral data matrix, perform data normalization processing, calculate the eigenvectors of the Raman spectrum, and use the formula:

[0038] ;

[0039] Calculate the eigenvectors of the Raman spectrum , and construct an identification model based on a deep learning algorithm;

[0040] where represents the The eigenvector of the Raman spectrum, representing the Raman signal intensity of the th data point in the representing the average signal intensity of the representing the total number of sampling points of the Raman spectrum, representing the regularization parameter to prevent the denominator from approaching zero;

[0041] Training the recognition model using the database;

[0042] Inputting the electrical signal into the recognition model, using the formula:

[0043] ;

[0044] Calculating the specific signal feature, extracting the specific signal feature, and identifying the tumor marker according to the specific signal feature;

[0045] wherein, represents the specific signal feature value, representing the th processed electrical signal data value after enhancement, representing the mean value of the processed electrical signal data after enhancement, representing the th background signal data value, representing the mean value of the background signal data, representing the total number of processed electrical signal data after enhancement, representing the total number of background signal data, representing the index of the processed electrical signal data after enhancement, representing the index of the background signal data.

[0046] Preferably, the labeled tumor marker includes a spatial position label and a quantity concentration label.

[0047] As a general inventive concept, the present invention also provides the following technical solutions:

[0048] A multi-channel Raman spectroscopy analysis method for tumor marker detection, comprising the following steps:

[0049] S1. Hardware preprocessing

[0050] Calibrating the spectral acquisition device provided with a plurality of independent optical channels, and a laser source, a beam splitter, a detector and a TE cooling module are provided in each independent optical channel.

[0051] The size of the overall spectral acquisition system is 144×92×46 mm, and asFigure 3 As shown in the figure, the optical path of the spectrometer is built using the Czerny-Turner system. Two spherical mirrors are used as the collimating lens and the focusing lens respectively. The incident light path and the imaging light path cross each other, and the slit width is 25μm, so as to improve the light flux as much as possible while ensuring that the resolution is not affected too much.

[0052] Specifically, the mercury lamp method is used to calibrate and debug the spectral acquisition device. The results of multiple measurements all prove that the spectral coverage range of the spectral acquisition device of the present invention is 785nm - 1050nm, and the resolution is 6.45cm -1 (Full width at half maximum at 0.54nm and 912nm), and the resolution is improved on the premise of ensuring that the Raman spectrum can fully reflect the concentration information of tumor markers. At the same time, the signal-to-noise ratio of the system is higher than 6000:1, ensuring that the collected spectral signal can be used as a reliable information basis for quantitative analysis of tumor markers, and a silicon wafer is used as a standard substance to correct the wave number accuracy.

[0053] The laser source uses a quantum nano-pulse laser (QNPL). Each laser can generate a beam with a specific wavelength, which is used to excite the Raman scattering of the sample;

[0054] The beam splitter uses a "precision beam splitter (PBS)" and a "high-resolution spectral sorter (HSS)" to ensure the precise focusing of the laser and the clear separation of the spectral signals. The PBS is responsible for evenly distributing the incident laser to different regions of the sample, while the HSS is responsible for screening out useful spectral information from the complex scattered light; for example, a grating with 830 lines / mm.

[0055] The detector uses a "synchronous spectral sensor (SSS)". When the laser irradiates the sample, the scattered Raman spectrum is captured by the SSS to achieve efficient signal acquisition. In order to further improve the quality and accuracy of the data, the present invention also introduces a "dynamic spectral enhancement module (DSEM)" in the detector to amplify and enhance weak Raman signals; for example, a 2048-element back-illuminated linear array CCD with a pixel size of 14×900μm.

[0056] The refrigeration temperature of the TE refrigeration module is -50°C, so as to improve the signal-to-noise ratio and stability of the spectra in each optical channel.

[0057] S2. Sample pretreatment

[0058] Select aptamers based on the tumor markers in the target tissue section;

[0059] Wash the streptavidin-coated magnetic nanoparticles with buffer, and then resuspend them in buffer to obtain a suspension; add the aptamer to the suspension and react for 30min at room temperature under smooth oscillation conditions; magnetically separate to obtain magnetic nanoparticles with aptamers modified on the surface;

[0060] Coat the surface of the modified magnetic nanoparticles with a layer of chitosan as the capture probe;

[0061] Incubate the target tissue section with the capture probe to form an aptamer-target tissue section-aptamer sandwich complex;

[0062] Magnetically separate and wash the sandwich complex with buffer and then transfer it onto the SERS substrate.

[0063] The magnetic nanoparticles create an enhanced spectral reaction interface on the surface of the target tissue section. This interface significantly improves the efficiency of Raman scattering, enabling the effective detection of even extremely low concentrations of tumor markers. Using the multi-frequency resonance Raman scattering (MFRS) technique, by configuring a multi-wavelength laser system, multiple tumor markers in the sample are simultaneously excited. The laser of each wavelength is precisely adjusted to match the resonance frequency of a specific tumor marker, thereby maximizing the generation of Raman signals.

[0064] S3. Software preprocessing

[0065] Construct a Raman signal separation algorithm based on a convolutional neural network, a Raman spectroscopy database of tissue sections containing tumor markers, and an identification model based on a deep learning algorithm, and use the database to train the identification model.

[0066] S4. Use a spectral acquisition device to collect the Raman spectrum of the target tissue section. Specifically, multiple Raman scattered lights of different wavelengths are collected through multiple channels to improve the speed and efficiency of data collection.

[0067] S5. Convert the collected Raman spectrum into an electrical signal, and during this conversion process, the signal is also preliminarily processed by the spectral depth weaving technology (SDW), such as signal amplification, noise filtering, and signal stabilization.

[0068] S6. Perform enhancement processing on the electrical signal;

[0069] S7. Input the enhanced electrical signal into the identification model, extract specific signal features, and identify tumor markers based on the specific signal features;

[0070] S8. Construct an enhanced Raman image of the target tissue section based on the collected Raman spectrum and image reconstruction technology;

[0071] S9. Use the identification result of the tumor marker and the spatial analysis algorithm to mark the spatial position and quantity concentration of the tumor marker in the enhanced Raman image to obtain a target tissue pathological image marked with the tumor marker;

[0072] S10. Output the target histopathological image so that doctors can see the specific distribution of tumor markers in biological samples, thereby providing more accurate diagnostic information.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] (1) The high automation of the system of the present invention reduces the complexity of operations and the possibility of human errors. The standardized sample processing, spectral acquisition, data analysis, and result output processes greatly lower the operation threshold, enabling non-professionals to easily use it and improving the user-friendliness of the system.

[0075] (2) The system of the present invention improves the detection sensitivity of tumor markers by adopting multi-channel spectral acquisition and magnetic nano-enhancement, thereby ensuring effective detection of trace biomarkers related to tumors even at low concentrations.

[0076] (3) Through the design of multiple optical channels, the system of the present invention realizes the excitation and acquisition of Raman scattered light at multiple different wavelengths, thereby greatly enhancing the speed and efficiency of data acquisition; and the beam splitter of each optical channel uses a precision beam splitter and a high-resolution spectral sorter, thereby effectively ensuring the precise focusing of the laser and the high-quality output of the spectral signal.

[0077] (4) The system of the present invention optimizes and enhances the Raman spectral signal through the Raman signal separation processing of the convolutional neural network, thereby precisely eliminating the noise and interference introduced by sample preparation and environmental factors and ensuring the high consistency and repeatability of spectral data.

[0078] (5) The system of the present invention uses the directly acquired Raman spectrum as training data, does not involve complex spectral enhancement processing, and extracts and effectively identifies the features of tumor markers with a trained recognition model based on deep learning algorithms. Without antibody reagents and staining processes, it greatly shortens the pathological diagnosis cycle and reduces the diagnosis cost.

[0079] (6) The system of the present invention can present the distribution of tumor markers on the target tissue section in the form of an image through Raman imaging processing, providing a visual pathological image for doctors. This is seamless and transitional for pathologists, helping doctors obtain clinical interpretations and make treatment decisions, overcoming the disadvantage that existing spectroscopy detection technologies are not easily understood by doctors and improving the efficiency and accuracy of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a structural diagram of the multi-channel Raman spectroscopy analysis system for tumor marker detection of the present invention;

[0081] Figure 2Flow chart of the multi-channel Raman spectroscopy analysis method for tumor marker detection of the present invention;

[0082] Figure 3 Optical path structure diagram of the acquisition module of the present invention. Specific embodiments

[0083] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0084] Embodiment 1

[0085] A multi-channel Raman spectroscopy analysis system for tumor marker detection includes an acquisition module, an analysis module, and an output module.

[0086] Regarding the acquisition module - used to acquire the Raman spectrum of the target tissue section.

[0087] The acquisition module includes a spectrum acquisition device provided with a plurality of independent optical channels, and a laser source, a beam splitter, a detector, and a TE cooling module are provided in each independent optical channel.

[0088] The overall size of the spectrum acquisition system is 144×92×46 mm, and as shown in Figure 3 The Czerny-Turner system is used to build the optical path of the spectrometer, and two spherical mirrors are used as the collimating lens and the converging lens respectively. The incident light path and the imaging light path cross each other, and the slit width is 25μm, so as to improve the light flux as much as possible while ensuring that the resolution is not affected too much.

[0089] The laser source uses a quantum nano-pulse laser (QNPL), and each laser can generate a beam with a specific wavelength, which is used to excite the Raman scattering of the sample. Based on this, a plurality of independent optical channels can respectively excite a plurality of Raman scattering lights with different wavelengths.

[0090] The beam splitter uses a "precision beam splitter (PBS)" and a "high-resolution spectrum sorter (HSS)" to ensure the precise focusing of the laser and the clear separation of the spectral signals. The PBS is responsible for evenly distributing the incident laser to different regions of the sample, while the HSS is responsible for screening out useful spectral information from the complex scattered light; for example, a grating with 830 lines / mm.

[0091] The detector uses a "Synchronous Spectrum Sensor (SSS)". When the laser irradiates the sample, the scattered Raman spectrum is captured by the SSS, enabling efficient signal acquisition. To further improve the quality and accuracy of the data, the present invention also introduces a "Dynamic Spectrum Enhancement Module (DSEM)" in the detector to amplify and enhance weak Raman signals; for example, a 2048-element back-illuminated linear array CCD with a pixel size of 14×900μm.

[0092] The refrigeration temperature of the TE refrigeration module is -50°C to improve the signal-to-noise ratio and stability of the spectra in each optical channel.

[0093] Regarding the analysis module - identifying tumor markers in the target tissue section through Raman spectroscopy analysis.

[0094] The analysis module uses high-dimensional spectral analysis technology to analyze and process the collected Raman spectra. This technology combines composite multi-dimensional quantum analysis (such as quantum support vector mechanism) and deep learning algorithms. Specifically, through deep learning of the Raman spectrum database, precise extraction of the specific spectral features of tumor markers is achieved, thereby ensuring the extraction of trace information from complex biological samples and enabling ultra-high-sensitivity identification of tumor markers even at extremely low concentrations.

[0095] Regarding the output module - used to label tumor markers according to the identification results of the analysis module and also used to output the target tissue pathological image with tumor markers labeled.

[0096] Labeling: Introduce a biological marker spectral mapping technology that combines image reconstruction technology and spatial analysis algorithms. Specifically, by constructing an enhanced Raman image of the target tissue section and marking the spatial position and quantity concentration of tumor markers in the enhanced Raman image, the target tissue pathological image is obtained.

[0097] Output: Output the target tissue pathological image (indicating the spatial position and quantity concentration of tumor markers in the image). In addition, the possible clinical significance of these tumor markers can also be analyzed in this image, providing necessary decision-making support for doctors and reference for formulating early diagnosis and personalized treatment plans.

[0098] Embodiment 2

[0099] This embodiment is a multi-channel Raman spectroscopy analysis method for tumor marker detection provided based on the above Embodiment 1, specifically including the following steps:

[0100] S1. Hardware preprocessing

[0101] Calibrate the spectral acquisition device with multiple independent optical channels. Each independent optical channel is equipped with a laser source, a spectroscope, a detector, and a TE refrigeration module.

[0102] The spectral acquisition device is calibrated and debugged by the mercury lamp method. The results of multiple measurements all prove that the spectral coverage range of the spectral acquisition device of the present invention is 785 nm - 1050 nm, and the resolution is 6.45 cm -1 (full width at half maximum at 0.54 nm and 912 nm), and the resolution is improved on the premise of ensuring that the Raman spectrum can fully reflect the concentration information of tumor markers. At the same time, the signal-to-noise ratio of the system is higher than 6000:1, ensuring that the collected spectral signals can serve as a reliable information basis for the quantitative analysis of tumor markers, and using a silicon wafer as a standard to correct the wavenumber accuracy.

[0103] S2. Software preprocessing

[0104] Construct a Raman spectrum database of tissue sections containing tumor markers and an identification model based on a deep learning algorithm, and train the identification model using the database.

[0105] Construct a Raman spectrum database of tissue sections containing tumor markers, obtain a spectral data matrix, perform data normalization processing, calculate the eigenvectors of the Raman spectrum, and use the formula:

[0106] ;

[0107] Calculate the eigenvectors of the Raman spectrum , and construct an identification model based on a deep learning algorithm;

[0108] Among them, represents the eigenvector of the th Raman spectrum, represents the Raman signal intensity of the th data point in the th group of Raman spectra, represents the average signal intensity of the th group of Raman spectra, represents the total number of sampling points of the Raman spectrum, represents the regularization parameter to prevent the denominator from approaching zero;

[0109] Construct a Raman spectroscopy database of tissue sections containing tumor markers. First, collect the Raman spectroscopy data of the target tissue sections. Using a high-sensitivity Raman spectroscopy instrument, under standardized experimental conditions, such as a laser wavelength of 785 nanometers, a laser power of 100 milliwatts, and an integration time of 10 seconds, collect the spectral data of each sample. The obtained raw spectral data may contain baseline drift and noise. Therefore, it is necessary to preprocess the data. Use the method of polynomial fitting for baseline correction, select a third-order polynomial to fit the baseline, and then subtract the fitted baseline from the original spectrum. Next, apply the moving average filtering method to remove high-frequency noise, select data points with a window size of 5 for smoothing. After completing the preprocessing, normalize the spectral data. Use the method of maximum normalization, divide the intensity value of each spectrum by its maximum value, so that the intensity range of all spectra is between 0 and 1. The processed spectral data is stored in the database. Each spectral data entry contains relevant information of the sample, such as sample number, collection date, experimental conditions, etc. The database design uses a relational database structure, sets up tables to store the spectral data and its relevant information to ensure the retrievability and manageability of the data. Finally, construct an identification model based on deep learning algorithms.

[0110] In the above formula, represents the feature vector of the th Raman spectrum, represents the Raman signal intensity of the th data point in the th group of Raman spectra, represents the average signal intensity of the th group of Raman spectra, represents the total number of sampling points of the Raman spectrum, represents the regularization parameter to prevent the denominator from approaching zero.

[0111] To calculate these parameters, first, collect the data points of the th group of Raman spectra. Assume , indicating that each spectrum contains 1000 data points. Then, calculate the average signal intensity of the th group of Raman spectra. Its calculation formula is:

[0112] ;

[0113] Assume the sum of the signal intensities of the data points of the th group of spectra is 5000, then . Then, calculate the sum of the absolute values of the differences between each data point and the average value, that is . Assume the calculation result is 800. Then, calculate the sum of the squares of the differences between each data point and the average value, that is Assume the calculation result is 1200, and select the regularization parameter Substitute these values into the formula to obtain:

[0114] ;

[0115] Therefore, the eigenvector of the th group of Raman spectra is approximately 23.09.

[0116] The innovation of this formula lies in that by calculating the sum of the absolute values of the differences between the Raman spectrum data points and the average value, as well as the square root of the sum of the squares of these differences, and introducing the regularization parameter a comprehensive consideration of the deviation degree and fluctuation amplitude of the spectral signal generates an eigenvector which not only reflects the overall deviation degree of the spectral signal but also takes into account the severity of signal fluctuations, helping to more comprehensively characterize the characteristics of Raman spectra.

[0117] Compare the calculated eigenvector with a preset reference value. Assume the reference value is 20. Since is greater than the reference value, it indicates that the deviation degree and fluctuation amplitude of the Raman spectrum signal are relatively large, which may correspond to the high expression of tumor markers or other significant characteristics in the sample.

[0118] Input the electrical signal into the recognition model and use the formula:

[0119] ;

[0120] Calculate the specific signal characteristics, extract the specific signal characteristics, and identify the tumor markers based on the specific signal characteristics;

[0121] where represents the specific signal characteristic value, represents the th enhanced processed electrical signal data value, represents the average value of the enhanced processed electrical signal data, represents the th background signal data value, represents the average value of the background signal data, represents the total number of enhanced processed electrical signal data, represents the total number of background signal data, represents the index of the enhanced processed electrical signal data, represents the index of the background signal data.

[0122] The electrical signal is input into the recognition model. The electrical signal refers to a time-varying signal obtained by sensor detection, whose value changes continuously with time. In the specific implementation process, first, the original electrical signal data is acquired by a high-precision signal acquisition device, which needs to have a microvolt-level precision to capture weak bioelectrical signals. Subsequently, the electrical signal data is preprocessed, and the preprocessing process includes filtering and denoising. Among them, band-pass filtering is adopted for filtering, with the low-frequency cut-off frequency set at 0.5 Hz and the high-frequency cut-off frequency set at 100 Hz to ensure that the noise in the non-target frequency band is filtered out. In the denoising process, the neighborhood mean denoising method is adopted, calculating the mean value of the five points before and after the current signal point, and replacing the original value of the current point with the mean value to make the signal curve smoother, thereby reducing the interference of random noise. After filtering and denoising, amplitude normalization processing is performed on the electrical signal. The normalization method is linear normalization, that is, using the formula: ;

[0123] where, represents the normalized electrical signal value, is the original electrical signal value, and represent the minimum and maximum signal values within the current sampling time window respectively. This normalization operation ensures that electrical signals of different intensities can be analyzed within a unified numerical range. Subsequently, the normalized electrical signal is input into the recognition model. The recognition model is a neural network structure composed of a feature extraction unit and a classification unit. The feature extraction unit uses time-frequency analysis methods to extract features from the electrical signal. Specifically, first, the energy value of the electrical signal within a short-time window is calculated using the formula:

[0124] ;

[0125] where, represents the signal energy value within the short-time window, is the start time of the window, is the window length. This energy feature can effectively reflect the local change situation of the electrical signal. Subsequently, the calculated energy feature value is input into the classification unit. The classification unit uses multiple thresholds for judgment, setting thresholds , and . When , it is judged as a background noise signal. When , it is judged as a normal physiological signal. When , it is judged as a suspicious feature signal, and further analysis is carried out to determine whether it has tumor marker characteristics. Finally, specific signal characteristics are obtained, and tumor markers are identified based on the specific signal characteristics.

[0126] ;

[0127] where, Represents the specific signal eigenvalue;

[0128] Represents the th electrical signal data value after enhancement processing, and its value is derived from the normalized electrical signal;

[0129] Represents the mean value of the electrical signal data after enhancement processing, and its calculation formula is as follows:

[0130] ;

[0131] Where Represents the total number of electrical signal data after enhancement processing;

[0132] Represents the th background signal data value, and its source is the original background signal without enhancement processing;

[0133] Represents the mean value of the background signal data, and its calculation formula is as follows:

[0134] ;

[0135] Where Represents the total number of background signal data.

[0136] Parameter Assignment and Calculation

[0137] Set the number of electrical signal sampling data , the data is as follows:

[0138] ;

[0139] Calculate :

[0140] ;

[0141] Set the number of background signal sampling data , the data is as follows:

[0142] ;

[0143] Calculate :

[0144] ;

[0145] Calculate the numerator part:

[0146] ;

[0147] ;

[0148] Calculate the denominator part:

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] Calculate the final result:

[0155] ;

[0156] This result indicates that the current signal eigenvalue is at a relatively high level and may represent the characteristics of tumor markers.

[0157] Regarding the training of the recognition model:

[0158] Its deep learning algorithm uses a classic convolutional neural network model, represents each spectral sample as a one-dimensional array, uses two one-dimensional convolutional kernels with a kernel size of 2 in the first convolutional layer, then flattens the features extracted by the upper layer using Flatten, and finally uses a fully connected layer with 2 units to perform prediction output through the softmax activation function.

[0159] The present invention uses stochastic gradient descent to train the network. The input size of the network is m * 475 * 2, where m is the number of samples, 475 is the spectral length of the region of interest, the number of input channels is 2, and the output of the network is one-hot type data.

[0160] After the training is completed, it is verified through the validation set, and the convolutional neural network model with the minimum loss is used as the optimal recognition model.

[0161] S3. Use a spectral acquisition device to collect Raman spectra of the target tissue section.

[0162] S4. Convert the Raman spectra into electrical signals and input them into the recognition model, extract specific signal features, and identify tumor markers according to the specific signal features;

[0163] During the recognition process:

[0164] Keep the first dimension of the above model training data, cover all other dimensions, input it into the trained recognition model for prediction, and record the prediction recognition accuracy and LOSS value.

[0165] Next, change 1 to 1 + 1, repeat the above process, and select the 1 that maximizes the accuracy and minimizes the LOSS value to be added to the important region sequence.

[0166] Then, retain all the dimensions that appear in the important region in the input data, repeat the above process, and again select the 1 that maximizes the accuracy and minimizes the LOSS value to be added to the important region sequence. Keep iterating until N consecutive important points added to the important region sequence no longer change the prediction accuracy and then stop. For example, N = 5, and finally obtain the important region sequence;

[0167] After obtaining the important region sequence, continuousize the discrete data to obtain the important region of the final Raman spectrum. The density of points in the important region sequence represents the importance of the region where the point is located. Use the idea of progressive propagation to assign weight values to each point on the entire spectrum. Sequentially take out the important points from the important region sequence, assign the weight value w to this point, and then the weight values gradually decay to the leftmost and rightmost ends of the entire spectrum on the left and right of this point, that is, the weight value of the kth point on the left and right of this point is W×p k where p is the decay rate, for example, p = 0.9, and so on. Each point in the important region sequence will assign the weight value of the entire spectrum sequence once. After the important region sequence traversal is completed, add up the sequence weight values of each point in the entire spectrum to obtain the final spectrum region sequence weight value. The above process can ensure that when important points are adjacent, the weight values of these points and adjacent points are relatively high. Using the finally generated spectrum region weight value, specific signal features and corresponding tumor markers can be accurately identified.

[0168] S5. Construct an enhanced Raman image of the target tissue section according to the collected Raman spectrum and image reconstruction technology;

[0169] S6. Use the recognition result of the tumor marker and the spatial analysis algorithm to mark the spatial position and quantity concentration of the tumor marker in the enhanced Raman image to obtain the target tissue pathological image marked with the tumor marker;

[0170] S7. Output the target tissue pathological image.

[0171] In the above Embodiment 1 and Embodiment 2:

[0172] Through the design of multiple optical channels, the present invention realizes the excitation and collection of Raman scattered light of multiple different wavelengths, thereby greatly improving the speed and efficiency of data collection; and the beam splitter of each optical channel uses a precision beam splitter and a high-resolution spectral sorter, which effectively ensures the precise focusing of the laser and the high-quality output of the spectral signal.

[0173] The present invention uses directly collected Raman spectra as training data, without involving complex spectral enhancement processes. It uses a trained recognition model based on deep learning algorithms for feature extraction and effective recognition of tumor markers, eliminating the need for antibody reagents and staining processes, greatly shortening the pathological diagnosis cycle and reducing the diagnosis cost.

[0174] The present invention can present the distribution of tumor markers on the target tissue section in the form of an image through Raman imaging processing, providing a visual pathological image for doctors. This is seamless and transitional for pathologists, helping doctors obtain clinical interpretations and make treatment decisions, overcoming the drawback that existing spectroscopy detection techniques are not easily understood by doctors and improving the efficiency and accuracy of diagnosis.

[0175] Example 3

[0176] A multi-channel Raman spectroscopy analysis system for tumor marker detection includes an acquisition module, an enhancement processing module, an analysis module, and an output module.

[0177] Regarding the acquisition module - used to acquire the Raman spectra of the target tissue section.

[0178] The acquisition module includes a spectral acquisition device with multiple independent optical channels, and each independent optical channel is equipped with a laser source, a spectroscope, a detector, and a TE cooling module.

[0179] Regarding the enhancement processing module - converts the acquired Raman spectra into electrical signals and also uses a neural network to perform enhancement processing on the electrical signals.

[0180] The enhancement processing module specifically performs enhancement processing on the electrical signals through a Raman signal separation algorithm based on a convolutional neural network.

[0181] Regarding the analysis module - identifies the tumor markers of the target tissue section through Raman spectroscopy analysis.

[0182] The analysis module uses high-dimensional spectral analysis technology to analyze and process the acquired Raman spectra. This technology combines composite multi-dimensional quantum analysis (such as quantum support vector mechanism) with deep learning algorithms. Specifically, it realizes the precise extraction of specific spectral features of tumor markers through deep learning of the Raman spectroscopy database, so as to ensure the extraction of trace information from complex biological samples and achieve ultra-high sensitivity recognition of tumor markers even at extremely low concentrations.

[0183] Regarding the output module - used to mark the tumor markers according to the recognition results of the analysis module and also used to output the pathological image of the target tissue marked with tumor markers.

[0184] Marker: Introduce biomarker spectral mapping technology, which combines image reconstruction technology and spatial resolution algorithms. Specifically, by constructing an enhanced Raman image of the target tissue section and marking the spatial position and quantitative concentration of tumor markers in the enhanced Raman image, the pathological image of the target tissue is obtained.

[0185] Output: Output the pathological image of the target tissue (the spatial position and quantitative concentration of tumor markers are marked in the image). In addition, the possible clinical significance of these tumor markers can be analyzed in this image, providing necessary decision support for doctors and reference for formulating early diagnosis and personalized treatment plans.

[0186] Example 4

[0187] This example is a multi-channel Raman spectroscopy analysis method for tumor marker detection provided based on the above Example 3, specifically including the following steps:

[0188] S1. Hardware preprocessing

[0189] Calibrate the spectral acquisition device with multiple independent optical channels. Each independent optical channel is equipped with a laser source, a spectroscope, a detector, and a TE cooling module.

[0190] The overall size of the spectral acquisition system is 144×92×46 mm, and as Figure 3 shown, the optical path of the spectrometer is built using a Czerny-Turner system. Two spherical mirrors are used as the collimating lens and the focusing lens respectively. The incident light path and the imaging light path cross each other, and the slit width is 25μm, so as to improve the light flux as much as possible while ensuring that the resolution is not affected too much.

[0191] Specifically, the mercury lamp method is used to calibrate and debug the spectral acquisition device. The measurement results of multiple times all prove that the spectral coverage range of the spectral acquisition device of the present invention is 785nm - 1050nm, and the resolution is 6.45 cm -1 (the full width at half maximum at 0.54nm, 912nm). The resolution is improved on the premise of ensuring that the Raman spectrum can fully reflect the concentration information of tumor markers. At the same time, the signal-to-noise ratio of the system is higher than 6000:1, ensuring that the collected spectral signal can be used as a reliable information basis for the quantitative analysis of tumor markers, and using a silicon wafer as a standard to correct the wavenumber accuracy.

[0192] S2. Software preprocessing

[0193] Construct a Raman signal separation algorithm based on a convolutional neural network, a Raman spectrum database of tissue sections containing tumor markers, and an identification model based on a deep learning algorithm, and train the identification model using the database.

[0194] The expression of the Raman signal separation algorithm is as follows:

[0195] R = L * B + N; where R is the measured spectral signal input, L represents the independent spectral peak signal of the Raman spectrum, B represents the instrument broadening signal and the background baseline signal, N represents the measurement noise, and "*" represents the convolution operation;

[0196] ; where x represents the Raman shift, c represents the center of the Lorentz function, wL represents the full width at half maximum of the Lorentz function, and S represents the total spectral area of the Lorentz function;

[0197] ; where w G represents the full width at half maximum of the Gaussian function .

[0198] Regarding the training of the recognition model:

[0199] Its deep learning algorithm uses a classic convolutional neural network model. Each spectral sample is represented as a one-dimensional array. In the first convolutional layer, 2 one-dimensional convolutional kernels with a kernel size of 2 are used. Then, Flatten is used to flatten the features extracted by the upper layer. Finally, a fully connected layer with 2 units is used, and the prediction output is made through the softmax activation function.

[0200] The present invention uses stochastic gradient descent to train the network. The input size of the network is m * 475 * 2, where m is the number of samples, 475 is the spectral length of the region of interest, the number of input channels is 2, and the output of the network is one-hot type data.

[0201] After the training is completed, it is verified through the validation set, and the convolutional neural network model with the minimum loss is used as the optimal recognition model.

[0202] S3. Use a spectral acquisition device to collect Raman spectra of the target tissue section.

[0203] S4. Convert the collected Raman spectra into electrical signals. During this conversion process, the spectral depth weaving technology (SDW) is also synchronously used to preliminarily process the signals, such as signal amplification, noise filtering, and signal stabilization.

[0204] S5. Perform enhancement processing on the electrical signals;

[0205] When using the above Raman signal separation algorithm for electrical signal enhancement processing:

[0206] The Lorentz spectral peak signal I and its parameter vector β separated from the original spectrum, and the instrument broadening signal B and its half-width parameter W G ;

[0207] Set the initial value of the Lorentz spectral peak signal parameter vector β, β = ∅, and the initial value of the number k of Raman spectral peaks it contains is 0. Set the initial value of the instrument broadening signal half-width parameter WG, WG = 5.

[0208] From the above, calculate the initial value of the estimated spectrum:

[0209] Since β = ∅, thus , and the initial value of the residual spectrum is the original spectrum.

[0210] Add a new Raman spectral peak to the Lorentz spectral peak signal, update the number of Raman spectral peaks, k = k + 1. Set the initial values of the parameters of this newly added spectral peak, and the steps are as follows: Find the maximum value hk of the residual spectrum signal and the wavenumber ck where it is located. Then search from the wavenumber ck to both sides to find the closest xleft and xright to ck, such that Y(x left + 1) ≥ hk / 2 and Y(x left ) < hk / 2, and Y(x right - 1) ≥ hk / 2 and Y(x right ) < hk / 2. Set the initial value of the parameter vector βk, w LK = min{ck - x left , x right - ck}, SK = 2w LK h K , βk = (S K , Ck, w LK ). Expand the parameter vector, β = (β 1， β 2，…, β k-1， β k ).

[0211] Optimize β and wG to obtain the estimated spectrum after parameter update. According to the original spectrum and the estimated spectrum, calculate the residual spectrum signal r(v). The calculation method of the residual spectrum is as follows:

[0212] r(x) = Y(x) - R(x, β, w G ) for x = x b , x b+1 , ..., x e .

[0213] When the maximum value of the residual spectrum is less than the termination threshold or the number of iterations is greater than or equal to the maximum number of iterations, output the Lorentz spectral peak signal L and its parameters, the vector β, and the instrument broadening signal B and its half-width parameter; otherwise, continue the iteration.

[0214] S6. Input the processed electrical signal into the recognition model, extract specific signal features, and identify tumor markers based on the specific signal features;

[0215] S7. Construct an enhanced Raman image of the target tissue section based on the collected Raman spectrum and image reconstruction technology;

[0216] S8. Utilize the recognition result of the tumor marker and the spatial analysis algorithm to mark the spatial position and quantitative concentration of the tumor marker in the enhanced Raman image, obtaining a target tissue pathological image marked with the tumor marker;

[0217] S9. Output the target tissue pathological image.

[0218] Compared with Embodiment 1 and Embodiment 2, in the above-mentioned Embodiment 3 and Embodiment 4, an enhancement processing module is added, enabling the present invention to optimize the processing parameters for enhancing the Raman spectrum signal in real time through the alternating convergence of the neural network, thereby precisely eliminating the noise and interference introduced by sample preparation and environmental factors, and ensuring the high consistency and repeatability of the spectral data.

[0219] Embodiment 5

[0220] A multi-channel Raman spectroscopy analysis system for tumor marker detection includes a preprocessing module, a collection module, an enhancement processing module, an analysis module, and an output module.

[0221] Regarding the preprocessing module - used to modify the target tissue section by magnetic nanoparticle adaptation.

[0222] The magnetic nanoparticles include at least one of gold nanoparticles and silver nanoparticles.

[0223] Regarding the collection module - used to collect the Raman spectrum of the target tissue section.

[0224] The collection module includes a spectral collection device provided with a plurality of independent optical channels, and each independent optical channel is provided with a laser source, a spectroscope, a detector, and a TE cooling module.

[0225] Regarding the enhancement processing module - converting the collected Raman spectrum into an electrical signal and further using a neural network to perform enhancement processing on the electrical signal.

[0226] The enhancement processing module specifically performs enhancement processing on the electrical signal through a Raman signal separation algorithm based on a convolutional neural network.

[0227] Regarding the analysis module - identifying the tumor marker of the target tissue section through Raman spectroscopy analysis.

[0228] The analysis module analyzes and processes the collected Raman spectra using high-dimensional spectral analysis technology, which combines composite multi-dimensional quantum analysis (such as quantum support vector mechanism) and deep learning algorithms. Specifically, through the deep learning of the Raman spectrum database, the precise extraction of the specific spectral features of tumor markers is realized, so as to ensure that trace information can be extracted from complex biological samples, and the ultra-high sensitivity identification of tumor markers can be achieved even at extremely low concentrations.

[0229] Regarding the output module - it is used to label tumor markers according to the recognition results of the analysis module, and is also used to output the target tissue pathological image with tumor markers labeled.

[0230] Labeling: Introduce the biological marker spectral mapping technology, which combines image reconstruction technology and spatial analysis algorithm. Specifically, by constructing the enhanced Raman image of the target tissue section and marking the spatial position and quantity concentration of tumor markers in the enhanced Raman image, the target tissue pathological image is obtained.

[0231] Output: Output the target tissue pathological image (the spatial position and quantity concentration of tumor markers are indicated in the image). In addition, the possible clinical significance of these tumor markers can also be analyzed in this image, providing necessary decision support for doctors and reference for formulating early diagnosis and personalized treatment plans.

[0232] Example 6

[0233] This example is a multi-channel Raman spectroscopy analysis method for tumor marker detection provided based on the above Example 5, specifically including the following steps:

[0234] S1. Hardware preprocessing

[0235] Calibrate the spectral acquisition device with multiple independent optical channels. Each independent optical channel is equipped with a laser source, a spectroscope, a detector, and a TE cooling module.

[0236] The overall size of the spectral acquisition system is 144×92×46 mm, and as Figure 3 shown, the optical path of the spectrometer is built using a Czerny-Turner system. Two spherical mirrors are used as the collimating lens and the converging lens respectively. The incident light path and the imaging light path cross each other, and the slit width is 25μm, so as to improve the light flux as much as possible while ensuring that the resolution is not affected too much.

[0237] Specifically, the mercury lamp method is used to calibrate and debug the spectral acquisition device. The measurement results of multiple times all prove that the spectral coverage range of the spectral acquisition device of the present invention is 785nm - 1050nm, and the resolution is 6.45cm -1(Full width at half maximum at 0.54 nm and 912 nm), which improves the resolution on the premise of ensuring that the Raman spectrum can fully reflect the concentration information of tumor markers. At the same time, the signal-to-noise ratio of the system is higher than 6000:1, ensuring that the collected spectral signals can serve as a reliable information basis for the quantitative analysis of tumor markers, and using a silicon wafer as a standard to correct the wavenumber accuracy.

[0238] S2. Sample pretreatment

[0239] Select aptamers with high affinity and high specificity for breast cancer biomarkers;

[0240] Wash the streptavidin-coated magnetic nanoparticles with buffer, and then resuspend them in buffer. Biotinylated TBA15 is added and combined with the buffer, and the reaction is carried out for 30 min under gentle oscillation conditions at room temperature. TBA15 is immobilized on the surface of the magnetic nanoparticles through avidin-biotin interaction. Streptavidin and the small ligand biotin exhibit extremely high binding affinity and specificity. Use a magnetic separator and B&W buffer to wash away the unreacted reagents. Resuspend the magnetic nanoparticles in the buffer solution. Thus, the modification of aptamers on the surface of magnetic beads of magnetic nanoparticles is achieved;

[0241] Coat a layer of chitosan on the surface of the modified magnetic nanoparticles to improve biocompatibility and stability;

[0242] Use the aptamer-modified magnetic nanoparticles as capture probes;

[0243] Incubate the sample with the capture probe to form an aptamer-tumor marker-aptamer sandwich complex;

[0244] Use a magnetic separator to remove the unbound components, and wash the complex with buffer;

[0245] Transfer the sandwich complex to the SERS substrate and wait for Raman spectrum measurement.

[0246] The magnetic nanoparticles create an enhanced spectral reaction interface on the surface of the target tissue section. This interface significantly improves the efficiency of Raman scattering, enabling the effective detection of even extremely low concentrations of tumor markers. Using the multi-frequency resonance Raman scattering (MFRS) technique, by configuring a multi-wavelength laser system, multiple tumor markers in the sample are simultaneously excited. The laser of each wavelength is precisely adjusted to match the resonance frequency of a specific tumor marker, thereby maximizing the generation of Raman signals.

[0247] S3. Software pretreatment

[0248] Construct a Raman signal separation algorithm based on a convolutional neural network, a Raman spectroscopy database of tissue sections containing tumor markers, and an identification model based on a deep learning algorithm, and use the database to train the identification model.

[0249] S4. Use a spectral acquisition device to collect Raman spectra of the target tissue section. Specifically, multi-channel is used to collect Raman scattered light of multiple different wavelengths, thereby improving the speed and efficiency of data collection.

[0250] S5. Convert the collected Raman spectra into electrical signals, and during this conversion process, the signal is also preliminarily processed by the spectral depth weaving technology (SDW), such as signal amplification, noise filtering, and signal stabilization.

[0251] S6. Perform enhancement processing on the electrical signal;

[0252] S7. Input the enhanced electrical signal into the identification model, extract specific signal features, and identify tumor markers based on the specific signal features;

[0253] S8. Construct an enhanced Raman image of the target tissue section according to the collected Raman spectra and image reconstruction technology;

[0254] S9. Use the identification result of the tumor marker and the spatial analysis algorithm to mark the spatial position and quantity concentration of the tumor marker in the enhanced Raman image, and obtain a target tissue pathological image marked with the tumor marker;

[0255] S10. Output the target tissue pathological image, so that doctors can see the specific distribution of tumor markers in biological samples, and further provide more accurate diagnostic information.

[0256] In the above-mentioned Example 5 and compared with Example 6, compared with Example 3 and Example 4, a pretreatment module is added, so that the present invention can improve the detection sensitivity of tumor markers by means of magnetic nanoparticles, thereby further ensuring that trace biological markers related to tumors can be effectively detected at low concentrations.

[0257] To sum up, the present invention is also illustrated by taking breast cancer biomarkers as an example. Breast cancer biomarkers can be represented by circulating tumor cells (CTCs), microRNA, and specific proteins:

[0258] S1. Collect samples

[0259] Collect serum samples from patients, pretreat the samples with buffer reagents to remove impurities and interfering substances, and store them at low temperature.

[0260] S2. Sample processing

[0261] Select aptamers with high affinity and high specificity for breast cancer biomarkers;

[0262] Wash streptavidin-coated magnetic nanoparticles with buffer and then resuspend them in buffer. Biotinylated TBA15 is added and combined with the buffer, and the reaction is carried out for 30 min under gentle shaking conditions at room temperature. TBA15 is immobilized on the surface of magnetic nanoparticles through avidin-biotin interaction. Streptavidin and the small ligand biotin exhibit extremely high binding affinity and specificity. Unreacted reagents are washed away using a magnetic separator and B&W buffer. The magnetic nanoparticles are resuspended in the buffer solution. Thus, the modification of aptamers on the surface of magnetic beads magnetic nanoparticles is achieved;

[0263] Coat a layer of chitosan on the surface of the modified magnetic nanoparticles to improve biocompatibility and stability;

[0264] Use the aptamer-modified magnetic nanoparticles as capture probes;

[0265] Incubate the sample with the capture probe to form a sandwich complex of aptamer-tumor marker-aptamer;

[0266] Use a magnetic separator to remove unbound components and wash the complex with buffer;

[0267] Transfer the sandwich complex to the SERS substrate and wait for Raman spectroscopy measurement.

[0268] S3. Use a spectral acquisition device to collect Raman spectra of the processed sample.

[0269] S4. Perform a series of preprocessing on the collected Raman spectra, such as cosmic ray muon noise removal, convolutional smoothing filtering, fluorescence background correction, and normalization.

[0270] S5. Convert the Raman spectra into electrical signals.

[0271] S6. Perform enhancement processing on the electrical signals.

[0272] Specifically, a Raman signal separation algorithm based on a convolutional neural network is proposed for enhancement processing.

[0273] The algorithm expression is R = L * B + N; where R is the measured spectral signal input, L represents the independent spectral peak signal of the Raman spectrum, B represents the instrument broadening signal and the background baseline signal, N represents the measurement noise, and "*" represents the convolution operation;

[0274] ; where x represents the Raman shift, c represents the center of the Lorentz function, wL represents the full width at half maximum of the Lorentz function, and S represents the full spectral range area of the Lorentz function;

[0275] ; where w G represents the full width at half maximum of the Gaussian function .

[0276] When performing electro - signal enhancement processing using the above algorithm:

[0277] The Lorentz spectral peak signal I separated from the original spectrum, its parameter vector β, the instrument broadening signal B, and its full width parameter W G ;

[0278] Set the initial value of the Lorentz spectral peak signal parameter vector β, β = ∅, and the initial value k of the number of Raman spectral peaks it contains is 0. Set the initial value of the instrument broadening signal full width parameter WG, WG = 5.

[0279] Thus, calculate the initial value of the estimated spectrum:

[0280] Since β = ∅, so , and the initial value of the residual spectrum is the original spectrum.

[0281] Add a new Raman spectral peak to the Lorentz spectral peak signal, update the number of Raman spectral peaks, k = k + 1. Set the initial values of the parameters of the newly added spectral peak as follows: Find the maximum value hk and the wavenumber ck of the residual spectrum signal. Then search from the wavenumber ck to both sides to find the closest xleft and xright to ck, such that Y(x left + 1) ≥ hk / 2 and Y(x left ) < hk / 2 and Y(x right - 1) ≥ hk / 2 and Y(x right ) < hk / 2. Set the initial value of the parameter vector βk, w LK = min{ck - x left , x right - ck}, SK = 2w LK h K , βk = (S K , Ck, w LK ). Expand the parameter vector, β = (β 1， β 2，…, β k-1， β k ).

[0282] Optimize β and wG to obtain the estimated spectrum after parameter update. According to the original spectrum and the estimated spectrum, calculate the residual spectrum signal r(v). The calculation method of the residual spectrum is as follows:

[0283] r(x) = Y(x) - R(x, β, w G ) for x = x b , x b+1, ..., x e 。

[0284] When the maximum value of the residual spectrum is less than the termination threshold or the number of iterations is greater than or equal to the maximum number of iterations, output the Lorentz spectral peak signal L and its parameters, the vector β, and the instrument broadening signal B and its half-width parameter; otherwise, continue the iteration.

[0285] S7. Construct a Raman spectroscopy database of tissue sections containing tumor markers and an identification model based on a deep learning algorithm; it should be known in the technical field that the Raman spectroscopy database in this application can be a Raman spectroscopy database of a biomarker, a classified Raman spectroscopy database of all biomarkers of a certain type of cancer, or a hierarchical classification Raman spectroscopy database of all biomarkers of multiple types of cancer;

[0286] S8. Use the database to train the identification model. Divide the database into a training set and a validation set, specifically train the identification model through the training set, and verify the trained identification model through the validation set. If the verification is passed, it is an available identification model;

[0287] Regarding the training of the identification model:

[0288] Its deep learning algorithm uses a classic convolutional neural network model. Each spectral sample is represented as a one-dimensional array. In the first convolutional layer, 2 one-dimensional convolutional kernels with a kernel size of 2 are used. Then, Flatten is used to flatten the features extracted by the upper layer. Finally, a fully connected layer with 2 units is used, and the prediction output is performed through the softmax activation function.

[0289] The present invention uses stochastic gradient descent to train the network. The input size of the network is m * 475 * 2, where m is the number of samples, 475 is the spectral length of the region of interest, the number of input channels is 2, and the output of the network is one-hot type data.

[0290] After completion of training, verify through the validation set, and use the convolutional neural network model with the minimum loss as the optimal identification model.

[0291] S9. Input the enhanced electrical signal into the identification model, extract the specific signal features of breast cancer biomarkers, and identify tumor markers (such as circulating tumor cells (CTCs), microRNA, and specific proteins) according to the specific signal features;

[0292] During the identification process:

[0293] Keep the first dimension of the above model training data, cover all other dimensions, input it into the trained identification model for prediction, and record the prediction identification accuracy and LOSS value.

[0294] Next, change 1 to 1 + 1, repeat the above process, and select the 1 that maximizes the accuracy and minimizes the LOSS value to be added to the important region sequence.

[0295] Then, retain all the dimensions that appear in the important region in the input data, repeat the above process, and again select the 1 that maximizes the accuracy and minimizes the LOSS value to be added to the important region sequence. Keep iterating until the prediction accuracy no longer changes when N consecutive important points are added to the important region sequence, for example, N = 5, and finally obtain the important region sequence;

[0296] After obtaining the important region sequence, continuousize the discrete data to obtain the important region of the final Raman spectrum. The density of points in the important region sequence indicates the importance of the region where the point is located. Use the idea of progressive propagation to assign weight values to each point on the entire spectrum. Successively take out the important points from the important region sequence, assign the weight value w to this point, and then the weight values gradually decay to the leftmost and rightmost ends of the entire spectrum on the left and right of this point, that is, the weight value of the kth point on the left and right of this point is W×p k , where p is the decay rate, for example, p = 0.9, and so on. Each point in the important region sequence will assign a weight value to the entire spectrum sequence once. After the important region sequence traversal is completed, add up the sequence weight values of each point in the entire spectrum to obtain the final spectrum region sequence weight value. The above process can ensure that when important points are adjacent, the weight values of these points and adjacent points are relatively high. Using the finally generated spectrum region weight value, specific signal features and corresponding tumor markers can be accurately identified.

[0297] S10. Construct an enhanced Raman image of the breast tissue section sample according to the collected Raman spectrum and image reconstruction technology;

[0298] S11. Use the recognition result of the tumor marker and the spatial analysis algorithm to mark the spatial position and quantity concentration of the tumor marker in the enhanced Raman image (color brightness can be used for marking), and obtain the breast tissue pathological image marked with the tumor marker;

[0299] S12. Output the breast tissue pathological image. The breast tissue pathological image indicates the spatial position and quantity concentration of circulating tumor cells, microRNA, and specific proteins. In addition, the possible clinical significance of circulating tumor cells, microRNA, and specific proteins can also be analyzed in the breast tissue pathological image.

[0300] In addition, to achieve the above object, an embodiment of the present invention also proposes a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, it implements the multi-channel Raman spectrum analysis system for tumor marker detection according to the embodiment of the present invention.

[0301] Those skilled in the art should understand that the embodiments of the present invention may be provided as a method, a system, or a computer program product. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0302] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal devices generate a system for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0303] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction system that implements the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0304] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0305] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0306] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, replacements, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-channel Raman spectroscopy analysis system for tumor marker detection, characterized in that: It includes acquisition module, enhanced processing module, analysis module and output module; The acquisition module is used to acquire the Raman spectrum of the target tissue slice, and the Raman spectrum includes Raman scattered light of multiple different wavelengths; The enhancement processing module is used to convert the Raman spectrum collected by the collection module into an electrical signal, and is also used to perform enhancement processing on the electrical signal; The analysis module is used to identify tumor markers by analyzing the enhanced processed electrical signals; The output module is used to mark the tumor marker according to the recognition result of the analysis module, and is also used to output the target tissue pathology image marked with the tumor marker.

2. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 1, characterized in that: It also includes a pre-processing module; The pretreatment module is used to modify the target tissue slice by adapting magnetic nanoparticles.

3. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 2, characterized in that: The preprocessing steps of the preprocessing module include: selecting an aptamer based on a tumor marker in the target tissue section; Modifying magnetic nanoparticles by the aptamer; A layer of chitosan is coated on the surface of the modified magnetic nanoparticles as a capture probe; incubating the target tissue slice with the capture probe to form a sandwich complex of aptamer-target tissue slice-aptamer; The sandwich complex was separated magnetically and washed with buffer before being transferred to a SERS substrate.

4. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 3, characterized in that: The step of modifying the magnetic nanoparticles by the aptamer comprises: The streptavidin-coated magnetic nanoparticles are washed with a buffer solution, and then resuspended in the buffer solution to obtain a suspension solution; The aptamer was added to the suspension and reacted for 30 min at room temperature under smooth shaking conditions; Magnetic separation is used to obtain magnetic nanoparticles with aptamers immobilized on the surface.

5. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 1, characterized in that: The acquisition module includes a spectrum acquisition device with multiple independent optical channels, and each independent optical channel is provided with a laser source, a spectrometer, a detector and a TE cooling module.

6. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 1, characterized in that: The enhancement processing module performs enhancement processing on the electrical signal by using a Raman signal separation algorithm based on a convolutional neural network.

7. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 6, characterized in that: The Raman signal separation algorithm based on convolutional neural network is expressed as: R=L*B+N; where R is the measured spectral signal input, L is the independent peak signal of the Raman spectrum, B is the instrument broadening signal and the background baseline signal, N is the measurement noise, and "*" represents the convolution operation.

8. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 7, characterized in that: The independent peak signal L of the Raman spectrum is expressed as: Where x represents the Raman shift, c represents the center of the Lorentz function, wL represents the half-width at half-height of the Lorentz function, and S represents the full spectrum area of ​​the Lorentz function; The instrument broadening signal and background baseline signal B are expressed as: Among them, w G The half-height and half-width of the Gaussian function .

9. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 1, characterized in that: The tumor markers identified include: A Raman spectral database of tissue sections containing tumor markers was constructed, and the spectral data matrix was obtained. After data normalization, the eigenvector of the Raman spectrum was calculated using the formula: ; Calculate Raman spectrum feature vector , a recognition model is built based on deep learning algorithm; among them, Representative The characteristic vector of the Raman spectrum, Representative The Raman spectrum of the group The Raman signal intensity of each data point is Representative The average signal intensity of the Raman spectrum of the group, Represents the total number of sampling points of the Raman spectrum, represents the regularization parameter that prevents the denominator from approaching zero; Using a database to train the recognition model; The electrical signal is input into the recognition model using the formula: ; Calculate specific signal features, extract specific signal features, and identify tumor markers based on the specific signal features; wherein, represents the characteristic value of the specific signal, Representative The electrical signal data value after enhancement processing, represents the mean value of the electrical signal data after enhancement processing, Representative Background signal data values, represents the mean value of background signal data, Represents the total amount of electrical signal data after enhanced processing, Represents the total number of background signal data, Represents the index of the electrical signal data after enhanced processing, Represents the index of the background signal data.

10. The multi-channel Raman spectroscopy analysis system for tumor marker detection according to claim 1, characterized in that: The labeled tumor markers include spatial position markers and quantitative concentration markers.

Images