Raman spectrum system application method and system based on ICCD (Intensified Charge Coupled Device) and optical fiber gating

Abstract

The invention discloses a Raman spectrum system application method and system based on ICCD (Intensified Charge Coupled Device) and optical fiber gating, and the method comprises the following steps: S1, a control unit switches the working state of a laser light source to generate laser for exciting a Raman signal of a target tissue; s2, the laser reaches a to-be-detected part through a flexible optical fiber probe, and a control unit switches the working states of a gating unit and an ICCD detector so as to collect a Raman spectrum signal of the target tissue through the optical fiber probe; s3, the control unit outputs the collected Raman spectrum signals to a spectral analysis module, and the spectral analysis module processes the collected Raman spectrum signals based on a built-in deep learning model so as to analyze and evaluate the malignancy degree of the collected tissue. The invention provides a Raman spectrum system application method and system based on ICCD (Intensified Charge Coupled Device) and optical fiber type gating, by combining ICCD and gating technologies, the Raman signal intensity is improved, the fluorescence background is reduced, and the detection sensitivity and efficiency are further improved.

Description

Technical Field

[0001] The present invention relates to the field of medical detection technology, and more specifically, to an application method and system of a Raman spectroscopy system based on ICCD and fiber-optic gating. Background Art

[0002] Cancer is a common problem facing mankind. Early diagnosis of cancer is crucial to improving treatment effectiveness and patient survival. Traditional histopathological testing methods usually require surgical biopsy or puncture biopsy, which are invasive, may cause complications, and are time-consuming.

[0003] Of course, non-invasive detection is also used in the existing technology. For example, the detection of certain parts (such as breast) is achieved by imaging (such as ultrasound, etc.) technology. For example, the patent application number is 202110038828.6, and the patent name is Construction of a cancer clinical indicator evaluation system based on imaging genomics qualitative algorithm, which is used to provide non-invasive, accurate and non-subjective judgment for the early diagnosis, prognosis and drug efficacy subtypes of cancer. However, the problem with this technology is that the system can only realize tumor morphological judgment by extracting multiple imaging features, and tumors have strong heterogeneity and will have multiple morphological manifestations. Imaging genomics analyzes from one dimension, and it is difficult to fully and accurately reflect the molecular biological characteristics, immunological characteristics, etc. of the tumor, thereby affecting the accuracy of diagnosis.

[0004] In addition, some parts use test paper products to make cancer judgments, such as the application number: 202211408042.X, the patent name is a gastric cancer diagnostic product and diagnostic model, which is used to diagnose early gastric cancer and has very high sensitivity and specificity. It can be used for early screening of gastric cancer to reduce the mortality rate of gastric cancer, but the problem is that this test paper product is an in vitro diagnostic product, which cannot achieve real-time monitoring or accurately locate the location of cancerous tissue, and requires further confirmation with the help of other detection methods.

[0005] As a non-invasive molecular imaging technology, Raman spectroscopy can analyze the relative content and distribution of proteins, lipids, nucleic acids and other components in cancerous cells at the molecular level, and has a deeper understanding of tumor heterogeneity. It can be used to identify and classify tumors, and the diagnosis is more accurate. At the same time, combining Raman spectroscopy with optical fiber probes and other equipment can obtain Raman spectral information of tissues in real time during surgery, helping doctors to quickly determine the boundaries and properties of tumors, so as to perform surgical resection more accurately. However, when traditional Raman spectrometers are used for in vivo detection, their detection sensitivity and efficiency are limited due to the fluorescence background of biological tissues themselves and the low Raman signal intensity. That is, traditional Raman spectrometers use continuous lasers as excitation sources and use detectors with exposure times in the ms range (or even longer) to collect signals. This leads to problems such as strong fluorescence interference and large background noise. In order to reduce the excitation of natural fluorescence and lower the intensity of fluorescence signals, a longer laser wavelength (such as 785nm) is usually selected. The Raman scattering cross section is inversely proportional to the fourth power of the laser wavelength, which leads to further weakening of the Raman signal intensity. The weak signal and strong background make the rapid collection of Raman signals from biological samples a major difficulty. Therefore, there is currently no mature technology to apply Raman spectrometers to the early diagnosis of cancer. Summary of the invention

[0006] An object of the present invention is to solve at least the above problems and / or disadvantages and to provide at least the advantages which will be described hereinafter.

[0007] In order to achieve these purposes and other advantages of the present invention, a method for applying a Raman spectroscopy system based on ICCD and fiber-type gating is provided, comprising:

[0008] S1, the control unit switches the working state of the laser light source to generate laser light that excites the Raman signal of the target tissue;

[0009] S2, the laser reaches the part to be detected through the flexible optical fiber probe, and the control unit switches the working state of the gating unit and the ICCD detector to collect the Raman spectrum signal of the target tissue through the optical fiber probe;

[0010] S3, the control unit outputs the collected Raman spectrum signal to the spectrum analysis module, and the spectrum analysis module processes the collected Raman spectrum signal based on the built-in deep learning model to analyze and evaluate the malignancy of the collected tissue;

[0011] S4. The control unit outputs and / or displays the processing result of the spectrum analysis module.

[0012] Preferably, in S2, the method for collecting Raman spectrum signals of the target tissue is: randomly selecting 10 points within a predetermined area of ​​the tissue to be collected to collect Raman spectra, and the collection time for each point is maintained at least 10 seconds.

[0013] Preferably, in S2, the gate control unit is used to control the exposure time of the ICCD to ensure that the shutter speed of the ICCD is 500 ps and the gate width is 5 ns.

[0014] Preferably, in S3, the deep learning model is obtained in the following manner:

[0015] S30, preprocessing and feature extraction of the collected Raman spectrum signal to obtain a collected tissue component spectrum;

[0016] S31, learning the standard Raman spectra of basic components through the neural network model, and using the collected tissue component spectra and normal tissue spectra as training data for the neural network model;

[0017] The trained neural network model is used to analyze the contribution of the basic components in the collected tissue component spectrum to give the proportional relationship of the basic components in the tissue;

[0018] S32. Based on the characteristics of Raman spectral data, a fusion neural network model based on a convolutional neural network and a long short-term memory network is constructed and trained, so as to evaluate the performance of the neural network model through a cross-validation method, and then the weights W1 and W2 of the convolutional neural network and the long short-term memory network are assigned respectively to obtain a trained deep learning model;

[0019] The basic components include DNA, fat, histone, collagen and actin.

[0020] Preferably, in S3, the trained deep learning model is compressed and deployed on the spectral analysis module to achieve real-time analysis and feedback of the collected Raman spectral data.

[0021] Preferably, in S30, the preprocessing includes: performing noise reduction, baseline correction, normalization and spectrum distortion correction operations on the collected Raman spectrum.

[0022] The feature extraction method comprises:

[0023] S301, extracting effective feature information in the preprocessed Raman spectrum signal graph from multiple dimensions;

[0024] S302, extracting key information with biological significance for tumor diagnosis from effective feature information using a spectrum peak feature extraction algorithm based on continuous wavelet transform, thereby obtaining spectrum peak feature data;

[0025] The effective characteristic information refers to characteristic spectral peak fragments, so as to characterize the spectral peak attribution with biological significance for tumor diagnosis through characteristic spectral peak fragments;

[0026] The key information includes the peak position, peak intensity and peak width of the spectrum peak in the characteristic spectrum peak fragment.

[0027] Preferably, in S301, the attribution of the spectral peaks with biological significance for tumor diagnosis is characterized by the following characteristic spectral peak fragments:

[0028] Pass 616-626cm -1 The characteristic peak fragments characterize amide IV;

[0029] Pass 637-647cm -1 The characteristic peak fragments represent the CS stretching vibration;

[0030] Pass 785-795cm -1 The characteristic peak fragments represent DNA;

[0031] Pass 855-865cm -1 The characteristic peak fragments characterize lipids;

[0032] Through 998-1008cm -1 The characteristic peak fragments characterize phenylalanine;

[0033] Pass 1075-1085cm -1 The characteristic peak fragments represent DNA;

[0034] Through 1235-1245cm -1 The characteristic peak fragments characterize amide III;

[0035] Pass 1318-1328cm -1 The characteristic peak fragments represent the CH deformation vibration;

[0036] Through 1444-1454cm -1 The characteristic peak fragments represent the CH2 bending vibration;

[0037] Through 1547-1557cm -1 The characteristic peak fragments characterize amide II;

[0038] Through 1611-1621cm -1 The characteristic peak fragments represent tryptophan;

[0039] Through 1664-1674cm -1 The characteristic peak fragments characterize amine I.

[0040] Preferably, in S31, the neural network model is trained to determine the ratio of the basic components in the tissue in the following manner:

[0041] S310, obtaining a simulated tissue spectrum S by the following formula:

[0042] s=a0s0+a1s1+a2s2+a3s3+a4s4+a5s5

[0043] In the above formula, s0 is the normal tissue spectrum, s1-s5 are the spectra of DNA, lipids, histones, collagen, and actin, respectively, and a0-a5 are the coefficients corresponding to the normal tissue spectrum, DNA spectrum, lipid spectrum, histone spectrum, collagen spectrum, and actin spectrum, respectively, and the values ​​of a0-a5 satisfy the following formula:

[0044]

[0045] S311, superimpose random noise on the simulated tissue spectrum S by the following formula to obtain the generated spectrum s m :

[0046] s m =s+n

[0047] In the above formula, n is random noise;

[0048] S312, use the convolutional neural network Alexnet to build a model to generate spectrum s m As the input data of the convolutional neural network model, the output data are the coefficients of each component in the spectrum.

[0049] Preferably, in S32, the fusion neural network model is based on the Boosting model to complete the fusion of the convolutional neural network and the long short-term memory network, and the output F of the fusion neural network model is obtained by the following formula:

[0050] F = ReLU(O CNN W1+O LSTM W2)

[0051] Among them, O CNN and O LSTM are the outputs of the CNN model and the LSTM model respectively, ReLU(·) represents the rectified linear unit, and ReLU(x)=max(0,x).

[0052] A Raman spectroscopy detection system is applied to a Raman spectroscopy system application method based on ICCD and fiber-optic gating, comprising:

[0053] A laser light source for generating Raman signals for exciting tissues;

[0054] ICCD detector for rapid acquisition of Raman spectroscopy signals;

[0055] A control unit that is communicatively connected to the laser light source and the ICCD detector;

[0056] a spectrum analysis module for providing spectrum analysis results to the control unit;

[0057] Wherein, the laser light source and ICCD detector are respectively combined into a fiber optic probe through corresponding optical fibers;

[0058] The end of the optical fiber probe is designed as a bendable structure.

[0059] The present invention has at least the following beneficial effects:

[0060] Firstly, the application method of the present invention provides a non-invasive tumor detection method, which improves the Raman signal intensity and reduces the fluorescence background by combining ICCD and gating technology, thereby improving the detection sensitivity and efficiency.

[0061] Secondly, the present invention provides a set of spectral analysis algorithms based on deep learning, which can achieve rapid and accurate analysis results of tumors and their malignancy under non-invasive conditions in vivo, and can serve as an auxiliary means to help doctors diagnose tumors.

[0062] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 The schematic diagram of the Raman spectrum detection system based on ICCD and fiber-optic gating of the present invention is shown in FIG.

[0064] Figure 2 A comparison diagram of tissue spectra collected by the Raman spectrum detection system of the present invention and a conventional optical fiber Raman system;

[0065] Figure 3 A flow chart of an application method of an ICCD-based fiber-type gated Raman spectrometer according to the present invention;

[0066] Figure 4 A schematic diagram of a deep learning model used in the present invention for spectral analysis;

[0067] Figure 5 In order to obtain the standard Raman spectra of the five basic components in the tissue during the application of the deep learning model of the present invention;

[0068] Figure 6 A schematic diagram showing the comparison of Raman spectra of gastric malignant tumor tissue and normal gastric body tissue collected by the Raman spectrum detection system of the present invention;

[0069] Figure 7 The figure is a schematic diagram of the analysis results of the components of malignant tumor tissue and normal gastric body tissue obtained by using the algorithm of the present invention. DETAILED DESCRIPTION

[0070] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0071] This patent aims to develop a Raman spectrometer based on ICCD (enhanced charge-coupled device) and fiber-optic gated Raman spectrometer. The ICCD is first applied to the fiber-optic Raman spectrometer, and the high sensitivity and fast readout capability of the ICCD are used to improve the collection efficiency and quality of the Raman signal. At the same time, by combining ICCD and gating technology, the Raman signal intensity is increased, the fluorescence background is reduced, and the detection sensitivity and efficiency are improved.

[0072] Example 1

[0073] like Figure 1 As shown, the Raman spectrometer of the present invention mainly includes:

[0074] (1) Fiber optic probe 1: It consists of a fiber optic bundle and is used to transmit the excitation laser and the Raman spectrum signal at the target tissue to the spectrometer. The end of the fiber optic probe is designed to be bendable so that it can adapt to different in vivo detection paths.

[0075] (2) Laser light source 2: The laser used to generate the excitation tissue Raman signal is a repetition rate pulsed laser light source with a wavelength of 785 nm, a maximum power of 300 mW, a pulse width of 5 ns, and a repetition rate of 1 MHz.

[0076] (3) ICCD detector 3: The highly sensitive ICCD detector is used to quickly collect Raman spectral signals. It should be noted that ICCD technology, namely enhanced charge coupled device technology, is a highly sensitive photoelectric detector technology that enhances the signal by adding a gain layer on the basis of the traditional CCD sensor. ICCD has the characteristics of high sensitivity, fast readout capability, low noise performance, high dynamic range, and good linear response. This scheme mainly utilizes the fast electronic gating capability of the ICCD detector, which can collect a large number of photons in a short time, thereby improving the signal-to-noise ratio of the signal.

[0077] (4) Gating unit 4: It should be noted that gating technology is a selection detection technology implemented on a time scale. By matching and synchronizing with the laser in time, it can detect effective Raman signals and compress background noise signals. The present solution mainly uses gating technology to control the exposure time of the detector. It can distinguish between Raman signals and fluorescence signals in terms of time resolution to suppress the fluorescence background. The shutter speed is 500ps and the gate width is 5ns. That is, the present invention uses gating technology to open the camera shutter every time a Raman signal appears, and sets the shutter width to allow only the Raman signal to enter the detector. The strong background signal is suppressed by gating, and the signal-to-noise ratio of the Raman signal is improved, thereby realizing Raman testing under a strong background.

[0078] (5) Spectral Analysis Module 5: Built-in deep learning-based algorithm for processing and analyzing collected Raman spectral data. This module includes a high-performance computing unit for running deep learning models.

[0079] (6) Control unit 6: used to control the operation of the entire system, including the position adjustment of the optical fiber probe, the working mode of the ICCD detector and the gating unit, and the data processing of the spectrum analysis module.

[0080] It should be noted that since the generation and decay speed of Raman signals are much faster than those of fluorescence signals, by accurately controlling the gating time, the fluorescence background can be effectively suppressed and the quality of Raman spectra can be improved. Therefore, this scheme can not only enhance the Raman signal but also reduce the fluorescence background through the combination of ICCD and gating technology, thereby significantly improving the detection sensitivity.

[0081] In addition, the fast readout capability of ICCD also helps to increase the speed of Raman spectrum acquisition and further improve detection efficiency. The design of the fiber-optic Raman spectrometer allows the device to be easily inserted into the body tissue for detection, and the application of ICCD and gating technology ensures that high-quality Raman spectral data can be obtained in the in vivo environment, making it possible for real-time, non-invasive in vivo detection, thereby greatly promoting the clinical application of Raman spectrometers.

[0082] like Figure 2 As shown, by comparing the tissue Raman spectra collected by the Raman spectrometer of the present invention with those collected by a conventional non-ICCD gating system, it can be seen that the fluorescence background in the Raman spectra collected by the present invention is significantly suppressed, and the tissue characteristic peaks are more obvious.

[0083] Furthermore, the present invention provides a set of Raman spectrometer application methods based on deep learning, which can realize rapid and accurate analysis of tumors and their malignancy under non-invasive conditions in vivo, and provide assistance for doctors' diagnosis.

[0084] Example 2

[0085] like Figure 3 As shown, the Raman spectrometer application method of the present invention mainly includes the following steps:

[0086] S1. Preparation and start stage: Place the fiber optic probe in contact with the part to be tested directly or through a visual imaging guidance system, ensuring that the end of the fiber optic probe is close to the target tissue (the distance is less than 1 cm).

[0087] S2. Raman spectrum acquisition: Start the gating unit and ICCD detector, collect the Raman spectrum signal of the target tissue through the optical fiber probe, and use the high-speed readout capability of the ICCD detector to greatly shorten the signal acquisition time. However, in actual applications, the Raman spectrum acquisition here refers to randomly selecting 10 points within a certain area of ​​the tissue to collect the Raman spectrum, and the acquisition time for each point is 10 seconds.

[0088] S3. Spectral data processing and analysis after signal transmission: The collected Raman spectrum signal is transmitted to the spectrum analysis module through the optical fiber probe, ICCD detector and control unit.

[0089] After receiving the signal, the spectral analysis module uses the built-in deep learning algorithm to process the spectral data. The deep learning model is trained to identify the specific Raman spectral features of tumor cells, assess their malignancy, and provide component analysis results.

[0090] It should be noted that the deep learning spectral analysis algorithm used in S3 needs to adapt to the characteristics of in vivo tissues and Raman spectra collected by ICCD and gating systems, and have high adaptability, high accuracy and high efficiency.

[0091] Specifically, if Figure 3 As shown, the processing method of the deep learning model includes the specific construction and implementation methods:

[0092] S31, data preprocessing module: perform preprocessing operations such as noise reduction, baseline correction, normalization and spectral distortion correction on the collected Raman spectra. Since the Raman spectra collected by ICCD and gating system may contain noise, it is necessary to first reduce the noise of the spectral data through the Savitzky-Golay filtering algorithm. Secondly, Raman spectra are often affected by the environment, instruments and sample preparation, resulting in baseline drift. Baseline correction is performed through methods such as polynomial fitting and iterative least squares to eliminate this effect. At the same time, since the spectral intensity may be different in different tissues and under different acquisition conditions, the spectrum needs to be normalized to eliminate the impact of this difference on model training. Finally, the collected spectrum may be distorted due to the ICCD itself, and wavelength calibration technology is needed to reduce the impact of distortion on the analysis results.

[0093] S32, feature extraction module: extract effective feature information from Raman spectra from multiple dimensions. Combine expert experience and theoretical knowledge to select characteristic peak fragments in tissue Raman spectra that have biological significance for tumor diagnosis, as shown in the following table:

[0094]

[0095]

[0096] Then, the peak feature extraction algorithm based on continuous wavelet transform is used to accurately extract the key information such as the peak position, peak intensity and peak width of the spectrum peaks in these spectral fragments. After such processing, the complex spectral data with thousands of channels is simplified to the peak feature data of only dozens of channels:

[0097] X s →X f

[0098] PFE(X f )→X c

[0099] Where X s is the original spectral data, X f is the selected spectral fragment, PFE is the peak feature extraction algorithm, X c is the extracted spectral peak characteristic data.

[0100] S33, tissue composition analysis module: build a neural network model to learn the standard Raman spectra of five basic components such as DNA, fat, histone, collagen, and actin, such as Figure 5 The trained model can analyze the contribution of these spectra in the tissue spectrum and then give the proportion of these components in the tissue, providing support for more in-depth pathological analysis.

[0101] The specific construction process of the model is as follows:

[0102] Collect tissue component spectra and a small amount of normal tissue spectra to generate model training data;

[0103] The spectrum of each tissue component and the spectrum of normal tissue are multiplied by a certain coefficient to obtain the simulated tissue spectrum:

[0104] s=a0s0+a1s1+a2s2+a3s3+a4s4+a5s5

[0105] Among them, s0 is the normal tissue spectrum, s1-s5 are the spectra of DNA, lipid (triolein), histone, collagen, and actin, respectively. All spectra are normalized. The six coefficients satisfy

[0106]

[0107] To increase the robustness of the model, a certain degree of random noise is superimposed on the generated spectrum:

[0108] s m =s+n

[0109] According to the results of existing literature research, in malignant tumors, the ratio of DNA and protein increases, and the ratio of lipids decreases. Therefore, the coefficient range of each component is set as follows: DNA: 0-0.05, lipid: 0.05-0.8, collagen: 0.05-0.4, actin: 0.01-0.4, and the rest is histone.

[0110] The convolutional neural network Alexnet is used to build the model. During training, the input data is the generated spectrum, and the output data is the coefficients of each component in the spectrum. During the training process, the neural network will learn the relationship between the presented tissue spectrum and the coefficients of each component in the spectrum. For the trained model, a tissue spectrum is input, and the proportional coefficients of each tissue component reflected by the spectrum can be analyzed and output.

[0111] S34, model training module: According to the characteristics of Raman spectroscopy data, a neural network model that integrates convolutional neural network (CNN) and long short-term memory network (LSTM) is constructed. Among them, CNN is used to implement the tissue component analysis function described in S43. It can automatically learn the deep features of Raman spectra, which are more effective in distinguishing different tissues and tumor types. LSTM is used to establish the correlation between the features extracted in S42 to improve the recognition accuracy and robustness of the model. The prediction results of the two models are combined through the Boosting model fusion technology to improve the final prediction accuracy:

[0112] F = ReLU(O CNN W1+O LSTM W2)

[0113] Where F is the final output result of the dual model fusion algorithm, O CNN and O LSTM are the outputs of the CNN model and the LSTM model, respectively, and W1 and W2 are the weights of the two models. The two weight values ​​of ReLU (full name of English: Rectified Linear Unit) are trainable parameters, which are determined by the neural network model through the evaluation and optimization module described in S45.

[0114] S35, Model evaluation and optimization module: Use cross-validation methods to evaluate the performance of the model to ensure that the model can achieve good performance on data sets of different organ tissues. Adjust the model's hyperparameters through grid search, random search, or Bayesian optimization to further improve the model's performance.

[0115] S36. For the application scenario of real-time detection at the surgical site, the algorithm is required to have real-time prediction capabilities. In order to achieve real-time prediction, the trained model needs to be compressed to reduce the computational complexity and memory usage of the model. The model is deployed on a Raman spectrometer to achieve real-time analysis and feedback of the collected Raman spectral data.

[0116] S4. Result output: The control unit outputs the processing results for display. The doctor can quickly determine whether the target tissue is a tumor and the malignancy of the tumor based on the displayed Raman spectrum and the analysis results of the deep learning algorithm. The doctor makes the final diagnosis decision based on the results provided by the system and combined with clinical information.

[0117] The supporting deep learning algorithm developed in this embodiment 2 is based on the characteristics of in vivo tissues and ICCD gated Raman spectrometers collecting spectra, which makes the identification of in vivo tumors and the assessment of malignancy more accurate and rapid. In addition, the use of a fiber-optic flexible probe design allows the device to reduce invasiveness and trauma to patients.

[0118] Verification example:

[0119] The Raman spectrometer based on ICCD and fiber-gated control of the present invention was used to complete the Raman detection and classification implementation case of gastric tissue biopsy samples. The specific results are as follows:

[0120] like Figure 6 As shown, the Raman spectra of gastric malignant tumor tissue and normal gastric body tissue were detected by using the present invention. It can be seen that the Raman spectra collected by the present invention have distinguishable differences, that is, the adaptability meets the detection requirements.

[0121] like Figure 7 As shown, the algorithm model of the present invention can realize the judgment of whether a tissue is a tumor, and can also give the results of tissue composition analysis. At the same time, according to the results, the DNA and protein content in the tumor tissue is higher, and the fat content is lower, which is consistent with the conclusions of existing research results and can fully verify the effectiveness of the present invention.

[0122] The above solution is only an illustration of a preferred embodiment, but is not limited thereto. When implementing the present invention, appropriate replacement and / or modification can be performed according to user needs.

[0123] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and the embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily realized. Therefore, without departing from the general concept defined by the claims and equivalent scope, the present invention is not limited to the specific details and the illustrations shown and described here.

Claims

1. A method for applying a Raman spectroscopy system based on ICCD and fiber-optic gating, characterized in that: include: S1, the control unit switches the working state of the laser light source to generate laser light that excites the Raman signal of the target tissue; S2, the laser reaches the part to be detected through the flexible optical fiber probe, and the control unit switches the working state of the gating unit and the ICCD detector to collect the Raman spectrum signal of the target tissue through the optical fiber probe; S3, the control unit outputs the collected Raman spectrum signal to the spectrum analysis module, and the spectrum analysis module processes the collected Raman spectrum signal based on the built-in deep learning model to analyze and evaluate the malignancy of the collected tissue; S4. The control unit outputs and / or displays the processing result of the spectrum analysis module.

2. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 1, characterized in that: In S2, the method for collecting Raman spectrum signals of the target tissue is: randomly selecting 10 points within a predetermined area of ​​the tissue to be collected to collect Raman spectra, and the collection time of each point is kept at least 10 seconds.

3. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 1, characterized in that: In S2, the gate control unit is used to control the exposure time of the ICCD to ensure that the shutter speed of the ICCD is 500 ps and the gate width is 5 ns.

4. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 1, characterized in that: In S3, the deep learning model is obtained in the following manner: S30, preprocessing and feature extraction of the collected Raman spectrum signal to obtain a collected tissue component spectrum; S31, learning the standard Raman spectra of basic components through the neural network model, and using the collected tissue component spectra and normal tissue spectra as training data for the neural network model; The trained neural network model is used to analyze the contribution of the basic components in the collected tissue component spectrum to give the proportional relationship of the basic components in the tissue; S32. Based on the characteristics of Raman spectral data, a fusion neural network model based on a convolutional neural network and a long short-term memory network is constructed and trained, so as to evaluate the performance of the neural network model through a cross-validation method, and then the weights W1 and W2 of the convolutional neural network and the long short-term memory network are assigned respectively to obtain a trained deep learning model; The basic components include DNA, fat, histone, collagen and actin.

5. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 4, characterized in that: In S3, the trained deep learning model is compressed and deployed on the spectral analysis module to achieve real-time analysis and feedback of the collected Raman spectral data.

6. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 4, characterized in that: In S30, the preprocessing includes: performing noise reduction, baseline correction, normalization and spectrum distortion correction operations on the collected Raman spectrum. The feature extraction method comprises: S301, extracting effective feature information in the preprocessed Raman spectrum signal graph from multiple dimensions; S302, extracting key information with biological significance for tumor diagnosis from effective feature information using a spectrum peak feature extraction algorithm based on continuous wavelet transform, thereby obtaining spectrum peak feature data; The effective characteristic information refers to characteristic spectral peak fragments, so as to characterize the spectral peak attribution with biological significance for tumor diagnosis through characteristic spectral peak fragments; The key information includes the peak position, peak intensity and peak width of the spectrum peak in the characteristic spectrum peak fragment.

7. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 6, characterized in that: In S301, the attribution of spectral peaks with biological significance for tumor diagnosis is characterized by the following characteristic spectral peak fragments: Pass 616-626cm -1 The characteristic peak fragments characterize amide IV; Pass 637-647cm -1 The characteristic peak fragments represent the CS stretching vibration; Pass 785-795cm -1 The characteristic peak fragments represent DNA; Pass 855-865cm -1 The characteristic peak fragments characterize lipids; Through 998-1008cm -1 The characteristic peak fragments characterize phenylalanine; Pass 1075-1085cm -1 The characteristic peak fragments represent DNA; Through 1235-1245cm -1 The characteristic peak fragments characterize amide III; Pass 1318-1328cm -1 The characteristic peak fragments represent the CH deformation vibration; Through 1444-1454cm -1 The characteristic peak fragments represent the CH2 bending vibration; Through 1547-1557cm -1 The characteristic peak fragments characterize amide II; Through 1611-1621cm -1 The characteristic peak fragments represent tryptophan; Through 1664-1674cm -1 The characteristic peak fragments characterize amine I.

8. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 6, characterized in that: In S31, the neural network model is trained to determine the ratio relationship of the basic components in the tissue as follows: S310, obtaining a simulated tissue spectrum S by the following formula: s=a0s0+a1s1+a2s2+a3s3+a4s4+a5s5 In the above formula, s0 is the normal tissue spectrum, s1-s5 are the spectra of DNA, lipids, histones, collagen, and actin, respectively, and a0-a5 are the coefficients corresponding to the normal tissue spectrum, DNA spectrum, lipid spectrum, histone spectrum, collagen spectrum, and actin spectrum, respectively, and the values ​​of a0-a5 satisfy the following formula: S311, superimpose random noise on the simulated tissue spectrum S by the following formula to obtain the generated spectrum s m : s m =s+n In the above formula, n is random noise; S312, use the convolutional neural network Alexnet to build a model to generate spectrum s m As the input data of the convolutional neural network model, the output data are the coefficients of each component in the spectrum.

9. The application method of the Raman spectroscopy system based on ICCD and fiber gating as claimed in claim 4, characterized in that: In S32, the fusion neural network model is based on the Boosting model to complete the fusion of the convolutional neural network and the long short-term memory network, and the output F of the fusion neural network model is obtained by the following formula: F=ReLU(O CNN W1+O LSTM W2) Among them, O CNN and O LSTM are the outputs of the CNN model and the LSTM model, respectively. ReLU(·) represents the rectified linear unit, and ReLU(x)=max(0,x) 。 10. A Raman spectroscopy detection system, applied to the application method of the Raman spectroscopy system based on ICCD and fiber-optic gating as claimed in any one of claims 1 to 9, characterized in that: include: A laser light source for generating Raman signals for exciting tissues; ICCD detector for rapid acquisition of Raman spectroscopy signals; A control unit that is communicatively connected to the laser light source and the ICCD detector; a spectrum analysis module for providing spectrum analysis results to the control unit; Wherein, the laser light source and ICCD detector are respectively combined into a fiber optic probe through corresponding optical fibers; The end of the optical fiber probe is designed as a bendable structure.

Images