Method and apparatus for detecting MGMT promoter methylation in glioblastoma

Abstract

A method and apparatus for detecting MGMT promoter methylation in glioblastoma, relating to the field of Raman spectroscopic detection. The method comprises: performing Raman spectral scanning on a glioma tissue sample to be detected to obtain Raman spectral data; preprocessing the Raman spectral data to obtain normalized Raman spectral data; extracting intensity values at characteristic shifts from the normalized Raman spectral data; and inputting the intensity values at the characteristic shifts into a classification model to calculate a classification prediction value, the classification prediction value being used for assisting in determining the MGMT promoter methylation state of said glioma tissue sample. The effect of the present application is that low-cost, non-destructive and accurate detection of the MGMT promoter methylation state of glioblastoma can be achieved.

Description

A method and apparatus for detecting methylation of the MGMT promoter in glioblastoma.

[0001] This application claims priority to Chinese Patent Application No. 202411523298.4, filed on October 30, 2024, entitled "A Method and Apparatus for Detecting Methylation of Glioblastoma MGMT Promoter", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of Raman spectroscopy detection, and in particular to a method and apparatus for detecting methylation of the MGMT promoter in glioblastoma. Background Technology

[0003] Gliomas are the most common tumors of the central nervous system, accounting for approximately 70% of all primary malignant brain tumors. They are among the most serious brain diseases with high mortality and disability rates, with the median survival of patients with highly malignant glioblastoma being only 12–14 months. To date, there is no satisfactory treatment for glioblastoma, and resistance to chemotherapy drugs is a significant reason for poor treatment outcomes.

[0004] The methylation status of the O6-methylguanine-DNA methyltransferase (MGMT) promoter is significantly associated with chemotherapy resistance in gliomas. MGMT promoter methylation is not only an important basis for evaluating the sensitivity of glioblastoma patients to alkylating agents, thus enabling the development of personalized precision treatment plans, but also a reference indicator for assessing prognosis and differentiating recurrence from pseudoprogression. Patients with high-grade gliomas often exhibit pseudoprogression on imaging after radiotherapy combined with concurrent temozolomide (TMZ) chemotherapy. Notably, the incidence of pseudoprogression is significantly higher in MGMT-methylated patients than in unmethylated patients, a phenomenon that actually predicts a better prognosis. Glioblastoma patients with MGMT promoter methylation are more sensitive to chemotherapy and radiotherapy, and therefore have relatively longer survival. For glioblastoma patients over 70 years of age, if the Karnofsky Performance Status (KPS) score is below 70, even under tolerable conditions, temozolomide treatment can effectively delay recurrence, prolong overall survival, and improve quality of life. If these patients also have MGMT promoter methylation, the treatment effect of temozolomide is even better. Currently, the detection of MGMT promoter methylation relies on surgically obtained pathological tissue for gene analysis, which has drawbacks such as high cost, invasiveness, and inaccurate results, making the high testing fees generally unacceptable to patients. Summary of the Invention

[0005] The purpose of this application is to provide a method and apparatus for detecting MGMT promoter methylation in glioblastoma, which can achieve low-cost, non-destructive, and accurate detection of MGMT promoter methylation status in glioblastoma.

[0006] To achieve the above objectives, this application provides the following solution.

[0007] In a first aspect, this application provides a method for detecting methylation of the MGMT promoter in glioblastoma, comprising: performing Raman spectroscopy on a glioma tissue sample to be tested to obtain Raman spectral data; preprocessing the Raman spectral data to obtain normalized Raman spectral data; and extracting the intensity value at a characteristic shift from the normalized Raman spectral data; wherein the characteristic shift includes a Raman shift of 1518 cm⁻¹. -1 1664cm -1 2895cm -1 2945cm -1 and 3176cm -1 The intensity value at the characteristic displacement is input into the classification model to calculate the classification prediction value. The classification model represents the relationship between the classification prediction value and the intensity value at the characteristic displacement. The classification prediction value is used to assist in determining the MGMT promoter methylation status of the glioma tissue sample to be tested. The MGMT promoter methylation status includes MGMT promoter methylation positive and MGMT promoter methylation negative. If the classification prediction value is greater than or equal to a predetermined threshold, the MGMT promoter methylation status is determined to be positive. If the classification prediction value is less than the predetermined threshold, the MGMT promoter methylation status is determined to be negative.

[0008] In one exemplary embodiment, preprocessing the Raman spectral data to obtain normalized Raman spectral data includes: removing fluorescence background from the Raman spectral data to obtain Raman spectral data with removed fluorescence background; smoothing and denoising the Raman spectral data with removed fluorescence background to obtain smoothed and denoised Raman spectral data; and normalizing the smoothed and denoised Raman spectral data to obtain normalized Raman spectral data.

[0009] In an exemplary embodiment, the classification model is determined based on a logistic regression algorithm and a maximum correlation minimum redundancy algorithm. The process of determining the classification model includes: acquiring several Raman spectral samples labeled with the methylation state of the MGMT promoter, preprocessing them, and extracting feature spectral information to construct a sample dataset; the sample dataset includes feature spectral information corresponding to several normalized Raman spectral samples and the methylation state of the MGMT promoter; the feature spectral information includes intensity values ​​at several Raman shifts; the MGMT promoter methylation state includes 1 and 0, where 1 represents MGMT promoter methylation positive and 0 represents MGMT promoter methylation negative; using the MGMT promoter methylation state as the target variable and the feature spectral information as the feature variable, a linear regression model is constructed; based on the sample dataset, the maximum correlation minimum redundancy algorithm is used to filter feature shifts, and the logistic regression algorithm is used to train the linear regression model to obtain the classification model.

[0010] In one exemplary embodiment, the expression of the classification model is: s = -0.2675 + (-1.9564 × I) 1518 )+0.7958×I 1664 +0.5511×I 2895 +1.3966×I 2945 +(-1.4758×I 3176 ); where s represents the regression algorithm output value; σ() represents the sigmoid function; e represents the natural constant; P represents the classification prediction value, when P≥0.5, it is judged as glioblastoma with MGMT promoter methylation positive, when P<0.5, it is judged as glioblastoma with MGMT promoter methylation negative; I 1518 This indicates a Raman displacement of 1518 cm. -1 Intensity value at; I 1664 This indicates a Raman displacement of 1664 cm. -1 Intensity value at; I 2895 This indicates a Raman displacement of 2895 cm. -1 Intensity value at; I 2945 This indicates a Raman displacement of 2945 cm. -1 Intensity value at; I 3176 This indicates a Raman displacement of 3176 cm. -1 The intensity value at each characteristic displacement; the regression coefficient before the intensity value at each characteristic displacement represents the importance of that characteristic displacement.

[0011] In one exemplary embodiment, removing the fluorescence background from the Raman spectral data to obtain Raman spectral data with removed fluorescence background includes: performing fluorescence background removal processing on the Raman spectral data using an adaptive iterative reweighted penalized least squares method to obtain Raman spectral data with removed fluorescence background.

[0012] In one exemplary embodiment, the Raman spectral data with removed fluorescence background is smoothed and denoised to obtain smoothed and denoised Raman spectral data, including: using the Savitzky-Golay smoothing algorithm to smooth and denoise the Raman spectral data with removed fluorescence background to obtain smoothed and denoised Raman spectral data.

[0013] In one exemplary embodiment, normalizing the smoothed and denoised Raman spectral data to obtain normalized Raman spectral data includes: normalizing the smoothed and denoised Raman spectral data using a max-min normalization algorithm to obtain normalized Raman spectral data.

[0014] Secondly, this application provides a glioblastoma MGMT promoter methylation detection device, employing the aforementioned glioblastoma MGMT promoter methylation detection method, comprising: a laser for emitting laser light onto a glioma tissue sample to be tested; a spectrometer for collecting Raman light scattered by the glioma tissue sample to obtain Raman spectral data; and a computer connected to the spectrometer for preprocessing the Raman spectral data to obtain normalized Raman spectral data, extracting intensity values ​​at characteristic shifts from the normalized Raman spectral data, and inputting the intensity values ​​at characteristic shifts into a classification model to calculate a classification prediction value; wherein if the classification prediction value is greater than or equal to a predetermined threshold, the MGMT promoter methylation status is determined to be positive; and if the classification prediction value is less than the predetermined threshold, the MGMT promoter methylation status is determined to be negative.

[0015] In an exemplary embodiment, the glioblastoma MGMT promoter methylation detection device further includes: an optical fiber probe connected to the laser and the spectrometer, respectively, for transmitting laser light emitted by the laser to the surface of the glioma tissue sample to be tested, and transmitting Raman light scattered from the surface of the glioma tissue sample to the spectrometer.

[0016] In an exemplary embodiment, the glioblastoma MGMT promoter methylation detection device further includes: a display connected to the computer for displaying the classification prediction value and auxiliary determination result; the auxiliary determination result is determined based on the classification prediction value.

[0017] In an exemplary embodiment, the glioblastoma MGMT promoter methylation detection device further includes: a detection platform for placing the glioma tissue sample to be tested, and for adjusting the position and height of the glioma tissue sample to be tested.

[0018] In one exemplary embodiment, the laser has an excitation wavelength of 532 nm and an integration time of 3 seconds.

[0019] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method and apparatus for detecting MGMT promoter methylation in glioblastoma. First, Raman spectroscopy is performed on the glioma tissue sample to be tested to obtain Raman spectral data. Then, the Raman spectral data is preprocessed and features are extracted to obtain the intensity values ​​at characteristic shifts. Finally, the intensity values ​​at the characteristic shifts are input into a pre-trained classification model to calculate a classification prediction value, which is used to assist in determining the MGMT promoter methylation status of the glioma tissue sample to be tested. Compared with the prior art, this application utilizes Raman spectroscopy technology to perform Raman spectral analysis on the glioma tissue sample to be tested, combined with a trained classification model, to achieve low-cost, non-destructive, and accurate detection of MGMT promoter methylation status, which can partially replace high-cost tumor gene detection and analysis functions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 is a flowchart of the glioblastoma MGMT promoter methylation detection method provided in this application.

[0022] Figure 2 is a schematic diagram of the average methylated and unmethylated spectra of the glioblastoma MGMT promoter provided in this application.

[0023] Figure 3 is a curve showing the predictive efficacy of the classification model provided in this application for the methylation state of the MGMT promoter.

[0024] Figure 4 is a structural diagram of the glioblastoma MGMT promoter methylation detection device provided in this application.

[0025] Figure labels: Laser-1, Spectrometer-2, Computer-3, Monitor-4, Detection Platform-5. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] This application provides a method and apparatus for detecting MGMT promoter methylation in glioblastoma. The method utilizes Raman spectroscopy to identify the methylation status of the MGMT gene promoter by analyzing the Raman spectrum of the glioma tissue sample, thereby enabling rapid detection of glioblastomas with positive MGMT promoter methylation and guiding the formulation of treatment plans.

[0028] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] In one exemplary embodiment, this application provides a method for detecting MGMT promoter methylation in glioblastoma. This method is executed by a computer device, specifically a terminal or server, or both. As shown in FIG1, the method for detecting MGMT promoter methylation in glioblastoma provided in this application includes the following steps S1 to S4.

[0030] Step S1: Perform Raman spectroscopy on the glioma tissue sample to be tested to obtain Raman spectral data.

[0031] The glioma tissue sample to be tested can be collected directly during the operation without any sectioning. Simply take a tissue fragment with a long diameter of about 2 to 20 mm and place it on a glass slide.

[0032] For example, a Raman spectroscopy excitation device (such as a laser with an excitation wavelength of 532 nm) was used to perform Raman spectroscopy scanning on the glioma tissue sample to be tested, and the Raman spectral data was acquired using a spectrometer. The integration time of the Raman spectroscopy scan was controlled within 3 seconds.

[0033] Step S2: Preprocess the Raman spectral data to obtain normalized Raman spectral data.

[0034] For example, firstly, the Raman spectral data is processed to remove the fluorescence background to obtain Raman spectral data with the fluorescence background removed; secondly, the Raman spectral data with the fluorescence background removed is processed to smooth and reduce noise to obtain smooth and denoised Raman spectral data; finally, the smooth and denoised Raman spectral data is processed to normalize to obtain normalized Raman spectral data.

[0035] Preferably, the fluorescence background removal process employs the Adaptive Iterative Re-weighted Penalized Least Squares (airPLS) method.

[0036] The airPLS algorithm removes background by iteratively adjusting the weights of the signal. In each iteration, it calculates the difference between the signal and the background and updates the weights so that the background can better fit the data. The airPLS algorithm includes initialization and iterative processes.

[0037] Initialization: Set the initial weights and input parameters. For example, the initial weight w = 1, and the input parameters include the signal g, the penalty parameter μ (4 in this application), and the maximum number of iterations itermax (15 in this application).

[0038] Iterative process: For each iteration, background smoothing, residual calculation, weight update, and convergence check are performed sequentially.

[0039] Background smoothing: The background z = WhittakerSmooth(g,w,μ) is solved using the Whittaker smoothing algorithm; where WhittakerSmooth() is the function of the Whittaker smoothing algorithm, which is achieved by minimizing the following objective function: min||w·(gz)|| 2 +μ·||Δz|| 2 ; where Δ represents the difference operator, applied to the background z to increase smoothness.

[0040] Calculate the residual: calculate the difference d between the signal and the background. k =g k -z k Where k represents the index of the data point, g k Let z represent the signal at the k-th data point. k d represents the background of the k-th data point. k This represents the difference between the signal at the k-th data point and the background, i.e., the residual value.

[0041] Update weights: For the portion of the signal that is higher than or equal to the background (i.e., d) k For data points ≥0), set the corresponding weight to 0 to ignore them, i.e., w k =0if d k ≥0; for the portion of the signal below the background (i.e., d k If <0), update the weights based on the residual values: if d k <0, where w kLet represent the weight of the k-th data point, and dssn represent the sum of the absolute values ​​of the negative residuals, i.e. t represents the number of iterations, and exp() represents an exponential function with the natural constant e as the base.

[0042] Convergence check: If the sum of the absolute values ​​of the negative residuals, dssn, is sufficiently small, or the number of iterations reaches the upper limit, then exit the iteration.

[0043] Preferably, the smoothing and noise reduction process uses the Savitzky-Golay smoothing algorithm, and the Python function name is savgol_filter.

[0044] The Savitzky-Golay smoothing algorithm is a filtering method for smoothing and differentiating digital signals. It achieves smoothing by performing polynomial fitting on the data within a local window. The specific Python function is: scipy.signal.savgol_filter(a,window_length,polyorder).

[0045] Where 'a' represents the signal to be smoothed. 'window_length' is the window width, which is an odd number and cannot exceed the length len(a). A larger value results in a more pronounced smoothing effect, while a smaller value results in a closer approximation of the original curve. The window width used in this application is 7. 'polyorder' is the order of the polynomial fitting. A smaller value results in a more pronounced smoothing effect, while a larger value results in a closer approximation of the original curve. The order used in this application is 2.

[0046] The formula for the Savitzky-Golay smoothing algorithm can be expressed as: Among them, b u For the smoothed data points, a u For the original data points, a u+v For a u The data points are moved by v units, m is half the window width (i.e., the window width is 2m+1), c v The coefficients are those of the polynomial fit.

[0047] Preferably, the normalization process employs a maximum-minimum normalization algorithm. The formula is: x' = (x - min(x)) / (max(x) - min(x)); where x' is the normalized Raman spectral data, x is the smoothed and denoised Raman spectral data, min(x) is the minimum value of the smoothed and denoised Raman spectral data, and max(x) is the maximum value of the smoothed and denoised Raman spectral data.

[0048] Step S3: Extract the intensity values ​​at characteristic shifts from the normalized Raman spectral data. Characteristic shifts include the Raman shift of 1518 cm⁻¹. -11664cm -1 2895cm -1 2945cm -1 and 3176cm -1 .

[0049] Step S4: Input the intensity value at the feature shift into the classification model to calculate the classification prediction value, which is used to help determine the MGMT promoter methylation status of the glioma tissue sample to be tested. The MGMT promoter methylation status includes MGMT promoter methylation positive and MGMT promoter methylation negative. If the classification prediction value is greater than or equal to a predetermined threshold, the MGMT promoter methylation status is determined to be positive; if the classification prediction value is less than the predetermined threshold, the MGMT promoter methylation status is determined to be negative. The classification model characterizes the relationship between the classification prediction value and the intensity value at the feature shift. The classification model is determined using machine learning or pattern recognition algorithms.

[0050] Preferably, this application determines the classification model based on the Logistic Regression (LR) algorithm and the maximal relevance and minimal redundancy (mRMR) algorithm. The determination process includes: acquiring several Raman spectral samples labeled with the methylation state of the MGMT promoter, preprocessing them and extracting feature spectral information to construct a sample dataset; the sample dataset includes feature spectral information corresponding to several normalized Raman spectral samples and the methylation state of the MGMT promoter; the feature spectral information includes intensity values ​​at several Raman shifts; the MGMT promoter methylation state includes 1 and 0, where 1 represents MGMT promoter methylation positive (i.e., MGMT promoter methylation) and 0 represents MGMT promoter methylation negative (i.e., MGMT promoter unmethylation); using the MGMT promoter methylation state as the target variable and the feature spectral information as the feature variable, a linear regression model is constructed; based on the sample dataset, the mRMR algorithm is used to screen feature shifts, and the LR algorithm is used to train the linear regression model to obtain the classification model.

[0051] The mRMR algorithm aims to select features from a sample dataset that are most relevant to the target variable and have minimal redundancy among them. Its objective function is typically expressed as: Where S is the selected feature subset; I(f i c) is a feature f i Mutual information with the target variable c (representing correlation); I(f i ,f j ) is a feature f i With feature f jMutual information between them (representing redundancy); max() indicates taking the maximum value. This application is based on the pymrmr library in Python, selecting 5 features that are most relevant to the target variable and have the least redundancy, specifically including the Raman shift of 1518cm. -1 1664cm -1 2895cm -1 2945cm -1 and 3176cm -1 This facilitates the recognition of methylation of the MGMT promoter in gliomas.

[0052] In this embodiment, based on spectral data of 443 cases from 104 retrospectively included patients (including 178 MGMT promoter methylation spectra and 265 MGMT promoter unmethylation spectra, the average spectrum is shown in Figure 2), the patients were randomly divided into training and validation groups at a ratio of 8:2 for training and validation. The mRMR algorithm and LR algorithm were used to generate a classification model to predict the methylation status of the MGMT promoter.

[0053] The trained classification model performs excellently on the validation set, with the area under the curve (AUC) reaching up to 86.4% and the sensitivity reaching 82.4%, as shown in Figure 3.

[0054] The main possible characteristic peaks involved in the process of using the mRMR algorithm for spectral feature screening are as follows.

[0055] 1518cm -1 This peak position is the characteristic peak position of β-carotene.

[0056] 1664cm -1 This peak position often reflects peptide bond vibrations, namely the stretching vibrations of C=O and NH, and can provide information about protein structure and conformation.

[0057] 2895cm -1 The peak positions in this region represent the Raman characteristic peak positions of protein content, caused by the stretching vibration of the CH bond in the methyl group.

[0058] 2945cm -1 This peak position usually represents the characteristic peak of the methyl bond CH3, reflecting the lipid content in the tissue.

[0059] 3176cm -1 The peak position in this region is usually associated with the Amide B vibration in the protein, which involves the stretching of free NH bonds and represents a characteristic of the protein's secondary structure.

[0060] Specifically, the expression for the classification model is: s = -0.2675 + (-1.9564 × I) 1518 )+0.7958×I 1664 +0.5511×I 2895 +1.3966×I 2945 +(-1.4758×I 3176 ).

[0061] Where s represents the output value of the regression algorithm; σ() represents the sigmoid function; e represents the natural constant; P represents the classification prediction value, i.e., the classification prediction probability. When P≥0.5, it is determined to be glioblastoma with MGMT promoter methylation positive; when P<0.5, it is determined to be glioblastoma with MGMT promoter methylation negative. 1518 This indicates a Raman displacement of 1518 cm. -1 Intensity value at; I 1664 This indicates a Raman displacement of 1664 cm. -1 Intensity value at; I 2895 This indicates a Raman displacement of 2895 cm. -1 Intensity value at; I 2945 This indicates a Raman displacement of 2945 cm. -1 Intensity value at; I 3176 This indicates a Raman displacement of 3176 cm. -1 The intensity value at each characteristic displacement. The regression coefficient preceding the intensity value at each characteristic displacement represents the importance of that characteristic displacement.

[0062] In one exemplary embodiment, this application provides a device for detecting MGMT promoter methylation in glioblastoma. As shown in FIG4, the device for detecting MGMT promoter methylation in glioblastoma provided in this application includes: a laser 1, a spectrometer 2, and a computer 3. The laser 1 emits a laser (i.e., 532nm excitation light) towards the glioma tissue sample to be tested; the spectrometer 2 collects the Raman light scattered by the glioma tissue sample to obtain Raman spectral data; the computer 3 is connected to the spectrometer 2 and is used to preprocess the Raman spectral data to obtain normalized Raman spectral data, extract the intensity value at the characteristic shift from the normalized Raman spectral data, and input the intensity value at the characteristic shift into a classification model to calculate a classification prediction value. If the classification prediction value is greater than or equal to a predetermined threshold, the MGMT promoter methylation status is determined to be positive; if the classification prediction value is less than the predetermined threshold, the MGMT promoter methylation status is determined to be negative.

[0063] Preferably, the excitation wavelength of laser 1 is 532 nm and the integration time is 3 seconds.

[0064] Furthermore, the aforementioned glioblastoma MGMT promoter methylation detection device also includes an optical fiber probe. The optical fiber probe is connected to the laser 1 and the spectrometer 2, respectively, and is used to transmit the laser emitted by the laser 1 to the surface of the glioma tissue sample to be tested, and to transmit the Raman light scattered from the surface of the glioma tissue sample to the spectrometer 2.

[0065] Furthermore, the aforementioned glioblastoma MGMT promoter methylation detection device also includes a display 4. The display 4 is connected to the computer 3 and is used to display classification prediction values ​​and auxiliary determination results; the auxiliary determination results are determined based on the classification prediction values.

[0066] Furthermore, the aforementioned glioblastoma MGMT promoter methylation detection device also includes a detection platform 5. The detection platform 5 is used to place the glioma tissue sample to be tested and to adjust the position and height of the glioma tissue sample.

[0067] The glioblastoma MGMT promoter methylation detection method and apparatus provided in this application have the following advantages.

[0068] 1. Fast: Utilizing resonance Raman spectroscopy, the detection time is short and the results are reported quickly.

[0069] 2. Accuracy: Utilizing characteristic spectral information for analysis, the methylation state of the MGMT promoter can be accurately identified.

[0070] 3. Non-destructive: No additional processing of tissue samples is required, maintaining the integrity of the samples.

[0071] 4. Simple: It is easy to operate and does not require complicated experimental conditions or professional skills.

[0072] In summary, this application provides a method and device for detecting MGMT promoter methylation in glioblastoma, which has the advantages of being rapid, accurate, and non-destructive, and is suitable for clinical medicine and scientific research.

[0073] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0074] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for detecting methylation of the MGMT promoter in glioblastoma, characterized in that, include: Raman spectroscopy was performed on the glioma tissue sample to be tested to obtain Raman spectral data; The Raman spectral data are preprocessed to obtain normalized Raman spectral data; Intensity values ​​at characteristic shifts are extracted from the normalized Raman spectral data; the characteristic shifts include a Raman shift of 1518 cm⁻¹. -1 1664cm -1 2895cm -1 2945cm -1 and 3176cm -1 ; The intensity value at the characteristic displacement is input into the classification model to calculate the classification prediction value; The classification model characterizes the relationship between the classification prediction value and the intensity value at the feature shift; the classification prediction value is used to assist in determining the MGMT promoter methylation status of the glioma tissue sample to be tested; the MGMT promoter methylation status includes MGMT promoter methylation positive and MGMT promoter methylation negative; If the classification prediction value is greater than or equal to a predetermined threshold, the methylation status of the MGMT promoter is determined to be positive; and if the classification prediction value is less than the predetermined threshold, the methylation status of the MGMT promoter is determined to be negative.

2. The method for detecting methylation of the MGMT promoter in glioblastoma according to claim 1, characterized in that, The Raman spectral data is preprocessed to obtain normalized Raman spectral data, including: The Raman spectral data are subjected to fluorescence background removal processing to obtain Raman spectral data with fluorescence background removed; The Raman spectral data with removed fluorescence background is smoothed and denoised to obtain smoothed and denoised Raman spectral data. The smoothed and denoised Raman spectral data are normalized to obtain normalized Raman spectral data.

3. The method for detecting methylation of the MGMT promoter in glioblastoma according to claim 1, characterized in that, The classification model is determined based on the logistic regression algorithm and the maximum correlation minimum redundancy algorithm; the process of determining the classification model includes: A number of Raman spectral samples labeled with the methylation state of the MGMT promoter were obtained, and preprocessed and feature spectral information was extracted to construct a sample dataset. The sample dataset includes feature spectral information and MGMT promoter methylation state corresponding to several normalized Raman spectral samples. The feature spectral information includes intensity values ​​at several Raman shifts. The MGMT promoter methylation state includes 1 and 0, where 1 indicates MGMT promoter methylation positive and 0 indicates MGMT promoter methylation negative. A linear regression model was constructed using the methylation state of the MGMT promoter as the target variable and the characteristic spectral information as the characteristic variable. Based on the sample dataset, the maximum correlation minimum redundancy algorithm is used to filter feature shifts, and the logistic regression algorithm is used to train the linear regression model to obtain a classification model.

4. The method for detecting methylation of the MGMT promoter in glioblastoma according to claim 1, characterized in that, The expression for the classification model is: s = -0.2675 + (-1.9564 × I) 1518 )+0.7958×I 1664 +0.5511×I 2895 +1.3966×I 2945 + (-1.4758×I 3176 ); Where s represents the regression algorithm output value; σ() represents the sigmoid function; e represents the natural constant; P represents the classification prediction value, when P≥0.5, it is judged as glioblastoma with MGMT promoter methylation positive, and when P<0.5, it is judged as glioblastoma with MGMT promoter methylation negative; I 1518 This indicates a Raman displacement of 1518 cm. -1 Intensity value at; I 1664 This indicates a Raman displacement of 1664 cm. -1 Intensity value at; I 2895 This indicates a Raman displacement of 2895 cm. -1 Intensity value at; I 2945 This indicates a Raman displacement of 2945 cm. -1 Intensity value at; I 3176 This indicates a Raman displacement of 3176 cm. -1 The intensity value at each characteristic displacement; the regression coefficient before the intensity value at each characteristic displacement represents the importance of that characteristic displacement.

5. The method for detecting methylation of the MGMT promoter in glioblastoma according to claim 2, characterized in that, The Raman spectral data were processed to remove fluorescence background using an adaptive iterative reweighted penalized least squares method. The Savitzky-Golay smoothing algorithm was used to smooth and denoise the Raman spectral data after removing the fluorescence background; the max-min normalization algorithm was used to normalize the smoothed and denoised Raman spectral data.

6. A device for detecting MGMT promoter methylation in glioblastoma, employing the method for detecting MGMT promoter methylation in glioblastoma as described in claim 1, characterized in that, include: A laser used to emit laser light onto a glioma tissue sample to be tested; A spectrometer is used to collect the Raman light scattered by the glioma tissue sample to be tested, and to obtain Raman spectral data; A computer, connected to the spectrometer, is used to preprocess the Raman spectral data to obtain normalized Raman spectral data, extract the intensity values ​​at characteristic shifts from the normalized Raman spectral data, and input the intensity values ​​at characteristic shifts into a classification model to calculate the classification prediction value. If the classification prediction value is greater than or equal to a predetermined threshold, the methylation status of the MGMT promoter is determined to be positive; and if the classification prediction value is less than the predetermined threshold, the methylation status of the MGMT promoter is determined to be negative.

7. The glioblastoma MGMT promoter methylation detection device according to claim 6, characterized in that, Also includes: An optical fiber probe is connected to the laser and the spectrometer, respectively, for transmitting the laser emitted by the laser to the surface of the glioma tissue sample to be tested, and for transmitting the Raman light scattered from the surface of the glioma tissue sample to the spectrometer.

8. The glioblastoma MGMT promoter methylation detection device according to claim 6, characterized in that, Also includes: A display, connected to the computer, is used to display the classification prediction values ​​and auxiliary judgment results; The auxiliary determination result is determined based on the classification prediction value.

9. The glioblastoma MGMT promoter methylation detection device according to claim 6, characterized in that, Also includes: The testing platform is used to place the glioma tissue sample to be tested and to adjust the position and height of the glioma tissue sample to be tested.

10. The glioblastoma MGMT promoter methylation detection device according to claim 6, characterized in that, The laser has an excitation wavelength of 532 nm and an integration time of 3 seconds.

Images