A circulating tumor cell detection system and electrochemical sensor preparation method

Abstract

The present invention relates to a circulating tumor cell detection system and an electrochemical sensor preparation method, which relate to the field of computer vision. The system includes an acquisition module for acquiring cell images using an electrochemical sensor; a segmentation and extraction module for performing segmentation and extraction based on the cell images using a cell image segmentation model to obtain segmentation results; the cell image segmentation model is obtained by training a U-Net network using a transfer learning algorithm; and a Grad-CAM algorithm processing module for obtaining a tumor cell heat map based on the segmentation results using the Grad-CAM algorithm. The present invention can achieve accurate identification of CTCs.

Description

Technical Field

[0001] The present invention relates to the field of computer vision, and in particular to a circulating tumor cell detection system and an electrochemical sensor preparation method. Background Art

[0002] Because the early symptoms of cancer are very hidden, and existing cancer diagnosis methods have shortcomings such as high cost, time-consuming screening, and limitations on tumor size, many patients have missed the best treatment period when they are diagnosed. Therefore, the improvement of early cancer diagnosis technology is a key means to improve patient cure and survival rates. Currently, the most promising new method for early cancer diagnosis is the detection of circulating tumor cells (CTCs), also known as liquid biopsy technology. Moreover, the morphological characteristics of CTCs in the blood can indicate the nature of the primary tumor, and high CTC counts may be related to the progression of tumor development and recurrence, thereby providing predictive and prognostic information about recurrence, survival, and treatment efficiency. Therefore, there is an urgent need to develop biosensors that can observe CTCs from a morphological perspective.

[0003] Traditional liquid biopsy techniques have drawbacks such as complex processes and high costs. Electrochemical detection technology, however, has garnered widespread attention in a variety of fields, including clinical medicine, drug screening, and disease diagnosis, thanks to its advantages of rapid detection, real-time performance, and high sensitivity. However, the high false positive and missed diagnosis rates of single-dimensional wave signals present significant challenges in the clinical application of electrochemical sensing technology. To address this challenge, the introduction of computer vision technology promises to offer a broad range of applications. Specifically, in research using artificial intelligence for medical image processing, deep learning algorithm models have demonstrated outstanding overall performance in medical image detection and disease diagnosis tasks, with accuracy rates comparable to, or even exceeding, those of real pathologists in many cases.

[0004] Despite its significant advantages, CTC detection presents significant challenges in early-stage cancer detection due to the extremely low CTC count in the blood (1-10 cells / mL). Therefore, it is necessary to combine the advantages of electrochemical sensors and artificial intelligence-assisted medical image processing to achieve accurate CTC identification and detection. Summary of the Invention

[0005] The purpose of the present invention is to provide a circulating tumor cell detection system and an electrochemical sensor preparation method to achieve accurate identification of CTCs.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A circulating tumor cell detection system, comprising:

[0008] an acquisition module, for acquiring cell images using an electrochemical sensor;

[0009] A segmentation and extraction module is used to perform segmentation and extraction based on the cell image using a cell image segmentation model to obtain a segmentation result; the cell image segmentation model is obtained by training a U-Net network using a transfer learning algorithm;

[0010] The Grad-CAM algorithm processing module is used to obtain a tumor cell heat map using the Grad-CAM algorithm according to the segmentation results.

[0011] Optionally, the segmentation and extraction module includes a backbone feature extraction network, a reinforcement feature extraction network, and a prediction network connected in sequence;

[0012] The backbone feature extraction network is used to perform preliminary extraction on the cell image to obtain a plurality of preliminary effective features;

[0013] The enhanced feature extraction network is used to perform upsampling, feature fusion and convolution on multiple preliminary effective features to obtain output features;

[0014] The prediction network is used to perform convolution on the output features to obtain a segmentation result.

[0015] Optionally, a pre-processing module is further included, and the pre-processing module is used to crop the cell image.

[0016] Optionally, the backbone feature extraction network is a VGG16 network.

[0017] A method for preparing an electrochemical sensor, wherein the electrochemical sensor obtained by the preparation method is applied to the circulating tumor cell detection system described in the above technical solution, and the electrochemical sensor preparation method comprises:

[0018] The conductive substrate is sequentially subjected to hydroxylation, epoxidation and carboxylation treatments in a solution;

[0019] Immersing the obtained surface-carboxylated conductive substrate in a dendritic polymer solution to carry out a first coupling reaction, coupling the dendritic polymer to the surface of the conductive substrate; the dendritic polymer solution comprises an organic condensation reagent, a dendritic polymer, and a buffer solution, wherein the dendritic polymer contains an amino group;

[0020] The obtained conductive substrate with the surface coupled to the dendritic polymer is immersed in an antibody solution for a second coupling reaction, and the antibody is coupled to the surface of the conductive substrate to obtain the electrochemical sensor; the antibody solution includes antibodies, an organic condensation reagent and a buffer solution.

[0021] Optionally, the hydroxylation treatment comprises the following steps:

[0022] The conductive substrate is immersed in an alkaline solution and heated to perform surface hydroxylation; the alkaline solution includes NH3, and the mass concentration of NH3 in the alkaline solution is 28%; the heating temperature for the hydroxylation is 85° C., and the heating and holding time is 30 minutes.

[0023] Optionally, the epoxidation treatment comprises the following steps:

[0024] The obtained surface hydroxylated conductive substrate was immersed in a 3-(2,3-epoxypropoxy)propyltrimethoxysilane solution and heated to perform surface epoxidation; the heating temperature for the surface epoxidation was 110° C., and the heating and holding time was 1 hour.

[0025] Optionally, the carboxylation treatment comprises the following steps:

[0026] The obtained surface epoxidized conductive substrate is immersed in an iminodiacetic acid solution and heated for surface carboxylation; the molar concentration of the iminodiacetic acid solution is 0.75 mol / L; the heating temperature for the surface carboxylation is 70° C., and the heating and holding time is 1 hour.

[0027] Optionally, the dendritic polymer is a polyamide-amine dendritic polymer; the pH value of the first coupling reaction is 7.4±0.5; and the reaction time of the first coupling reaction is 30 min.

[0028] The present invention provides an electrochemical sensor prepared by the preparation method described in the above technical solution, wherein the electrochemical sensor includes a conductive substrate and an antibody connected to the surface of the conductive substrate by a chemical bond; the electrochemical sensor is used in the circulating tumor cell detection system described in the above technical solution.

[0029] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0030] The present invention includes an acquisition module for acquiring cell images using an electrochemical sensor; a segmentation and extraction module for segmenting and extracting the cell images using a cell image segmentation model to obtain a segmentation result; the cell image segmentation model is obtained by training a U-Net network using a transfer learning algorithm; and a Grad-CAM algorithm processing module for obtaining a tumor cell heat map based on the segmentation result using the Grad-CAM algorithm. By training the U-Net network using a transfer learning algorithm to obtain the cell image segmentation model and implementing the Grad-CAM algorithm to obtain the tumor cell heat map, accurate CTC identification can be achieved.

[0031] The present invention provides a method for preparing an electrochemical sensor. The obtained electrochemical sensor is applied to the circulating tumor cell detection system described in the above technical solution. The method for preparing the electrochemical sensor comprises: sequentially subjecting a conductive substrate to hydroxylation, epoxidation, and carboxylation treatments in a solution; immersing the obtained conductive substrate with the carboxylated surface in a dendritic polymer solution for a first coupling reaction, thereby coupling the dendritic polymer to the surface of the conductive substrate; the dendritic polymer solution comprises an organic condensation reagent, a dendritic polymer, and a buffer solution, wherein the dendritic polymer contains amino groups; immersing the obtained conductive substrate with the dendritic polymer coupled to the surface in an antibody solution for a second coupling reaction, thereby coupling the antibody to the surface of the conductive substrate, thereby obtaining the electrochemical sensor; the antibody solution comprises an antibody, an organic condensation reagent, and a buffer solution. The preparation method provided by the present invention sequentially hydroxylates, epoxidizes, and carboxylates a conductive substrate, then couples a dendritic polymer through reactions between carboxyl and amino groups, and finally couples an antibody that can specifically recognize cancer cells to the structure of the dendritic polymer. The present invention synergizes the network structure characteristics of the dendritic polymer with the specific recognition of antibodies to achieve specific recognition and efficient capture of CTCs at extremely low levels in the blood in the early stages of cancer, creating a foundation for the combined detection and processing of CTCs by electrochemical sensors and artificial intelligence-assisted medical images, thereby achieving accurate recognition and detection of CTCs. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 This is a microscopic image of the ITO electrode provided by the present invention;

[0034] Figure 2 A comparison chart of cell capture results provided by the present invention;

[0035] Figure 3 This is a structural diagram of the H1299 cell image segmentation model based on the U-Net network provided by the present invention;

[0036] Figure 4 The loss function diagram of the algorithm model provided by the present invention;

[0037] Figure 5 The average intersection-union ratio map of the H1299 cell image segmentation model provided by the present invention;

[0038] Figure 6 An example of cell image segmentation provided by the present invention;

[0039] Figure 7 This is an example of a H1299 cell heat map provided by the present invention;

[0040] Figure 8 Schematic diagram of the circulating tumor cell detection system provided by the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] The purpose of the present invention is to provide a circulating tumor cell detection system and an electrochemical sensor preparation method to achieve accurate identification of CTCs.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] like Figure 8 As shown, the present invention provides a circulating tumor cell detection system, comprising:

[0045] The acquisition module 1 is used to acquire cell images using an electrochemical sensor.

[0046] The segmentation and extraction module 2 is used to perform segmentation and extraction based on the cell image using a cell image segmentation model to obtain a segmentation result; the cell image segmentation model is obtained by training a U-Net network using a transfer learning algorithm.

[0047] The Grad-CAM algorithm processing module 3 is configured to obtain a tumor cell heat map using the Grad-CAM algorithm according to the segmentation results.

[0048] In practical applications, the segmentation and extraction module 2 includes a backbone feature extraction network, an enhanced feature extraction network, and a prediction network connected in sequence. The backbone feature extraction network is used to perform preliminary extraction on the cell image to obtain multiple preliminary valid features. The enhanced feature extraction network is used to perform upsampling, feature fusion, and convolution on the multiple preliminary valid features to obtain output features. The prediction network is used to convolve the output features to obtain segmentation results. The backbone feature extraction network is a VGG16 network.

[0049] The circulating tumor cell detection system further includes a preprocessing module, which is used to crop the cell image.

[0050] In this embodiment, the cell segmentation model is an H1299 cell image segmentation model. The specific steps for constructing the H1299 cell image segmentation model are as follows:

[0051] Step 1: Create an H1299 cell image dataset.

[0052] Raw cell images were acquired using an electrochemical sensor. Photoshop was used to crop the images to a size of 512×512 pixels. Labeling of the cell images was performed using the Labelme tool. This application directly generated a JSON file corresponding to the image, which was then converted to a PNG file containing the corresponding labels. The VOC dataset was used as the dataset for the model.

[0053] Step 2: Use the U-Net network framework to construct the backbone feature extraction network, enhanced feature extraction network and prediction network.

[0054] The network block diagram of the H1299 cell image segmentation model is as follows Figure 3 As shown in the figure, the VGG16 network is used as the backbone feature extraction network, which consists of a convolution layer and a maximum pooling layer. The H1299 image size of the input backbone feature extraction network is 512×512×3. First, two convolutions with 64 channels are performed to obtain a preliminary effective feature layer F1 with a size of 512×512×64. Then, a maximum pooling and two convolutions with 128 channels are performed to obtain a feature layer F2 with a size of 256×256×128. After convolution and maximum pooling operations, a total of five preliminary effective feature layers are obtained. Feature layers: F1~F5; the enhanced feature extraction network uses five preliminary effective feature layers to continuously perform upsampling, feature fusion and convolution, and finally obtains an output feature layer of size 512×512×64. In the enhanced feature extraction part, the crop image operation is not performed. The purpose is to make the output final effective feature layer have the same size as the input H1299 image and increase the versatility of the network. In the prediction network part, a 1×1 convolution is used to adjust the number of channels of the input feature layer to the number of distinguished categories to obtain the prediction result.

[0055] Step 3: Use transfer learning method to train the model, obtain H1299 cell image segmentation model, and calculate model evaluation indicators.

[0056] 46 images are used as the training dataset, and the ratio of training set to validation set is set to 9:1. The model is trained using the pre-trained weights of the VGG16 network in the ImageNet project, and the number of training rounds is 20. The loss function after training is shown in the figure below. Figure 4 As shown, Figure 4 (a) in the figure is the loss function diagram of the training set. Figure 4 (b) in the figure is the loss function diagram of the validation set, which is given by Figure 4 It can be seen that the loss functions of the training set and the validation set gradually converge; the evaluation indicators such as Figure 5 As shown in Figure 2, the average intersection-over-union ratio of the model reached 95.32%. The cell image segmentation effect of this embodiment is shown in Figure 2. Figure 6 As shown, Figure 6 (a) is the H1299 cell segmentation diagram. Figure 6 (b) is the A549 cell segmentation diagram. Figure 6 (c) in the figure is the Hela cell segmentation diagram; Figure 6 As can be seen in the figure, almost all cells are completely segmented, and the model shows excellent segmentation performance. In addition, A549 lung cancer cell images and Hela cervical cancer cell images are used to verify the generalization performance of the model.

[0057] Step 4: Use the Grad-CAM algorithm to obtain the Grad-CAM output of the last feature layer, and then post-process it to create a heatmap. Use the PyTorch callback function to obtain the back-propagated gradient information of the model's last feature layer. After obtaining the weights for each channel in this feature layer, perform a weighted sum to obtain the Grad-CAM output, and then post-process it to create a heatmap. The feature map represents the segmented H1299 cells.

[0058] Load the trained model weights and specify the output of the last convolutional layer as the target feature layer; use Pytorch's callback function to obtain the back-propagation gradient information of the target feature layer and calculate the Grad-CAM output corresponding to the feature layer; take the average of the back-propagation gradient information of the target layer in the height and width dimensions, return the weight of each channel for the forward propagation information, then perform weighting and summation, set all values ​​less than 0 in the output to 0, and obtain the Grad-CAM output data of a single channel; scale the Grad-CAM output to the [0,1] interval, resize it, adjust it to the same size as the original input image, convert the obtained Grad-CAM grayscale image into uint8 image data format, call OpenCV's applyColorMap method to convert the Grad-CAM grayscale image into a color image; scale the input original image to the [0,1] interval and fuse it with the color image to obtain the following: Figure 7The heat map shown intuitively shows the key areas that the model focuses on during cell segmentation.

[0059] The present invention also provides a method for preparing an electrochemical sensor. The method for preparing the electrochemical sensor is applied to the circulating tumor cell detection system described in the above technical solution. The method for preparing the electrochemical sensor comprises:

[0060] The conductive substrate is sequentially subjected to hydroxylation, epoxidation and carboxylation treatments in a solution;

[0061] Immersing the obtained surface-carboxylated conductive substrate in a dendritic polymer solution to carry out a first coupling reaction, coupling the dendritic polymer to the surface of the conductive substrate; the dendritic polymer solution comprises an organic condensation reagent, a dendritic polymer, and a buffer solution, wherein the dendritic polymer contains an amino group;

[0062] The obtained conductive substrate with the surface coupled to the dendritic polymer is immersed in an antibody solution for a second coupling reaction, and the antibody is coupled to the surface of the conductive substrate to obtain the electrochemical sensor; the antibody solution includes antibodies, an organic condensation reagent and a buffer solution.

[0063] In the present invention, unless otherwise specified, all raw materials or components used are commercially available products well known to those skilled in the art.

[0064] The present invention sequentially performs hydroxylation, epoxidation and carboxylation treatments on the conductive substrate in a solution.

[0065] In the present invention, the material of the conductive substrate is preferably indium tin oxide (ITO).

[0066] In the present invention, the hydroxylation treatment preferably includes the following steps: immersing the conductive substrate in an alkaline solution and heating it to perform surface hydroxylation.

[0067] In the present invention, the alkaline solution preferably includes aqueous ammonia, and the mass concentration of NH3 in the aqueous ammonia is preferably 28%.

[0068] In the present invention, the alkaline solution preferably further comprises hydrogen peroxide, and the mass concentration of H2O2 in the hydrogen peroxide is preferably 30%.

[0069] In the present invention, the role of the hydrogen peroxide solution is to oxidatively remove dirt and impurities on the surface of the conductive substrate, so that the surface hydroxylation is fully carried out.

[0070] In the present invention, the volume ratio of ammonia water, hydrogen peroxide and water in the alkaline solution is preferably 1:1:6.

[0071] In the present invention, the water is preferably deionized water.

[0072] In the present invention, the heating temperature for the surface hydroxylation is preferably 85° C., and the heating and holding time is preferably 30 minutes.

[0073] In the present invention, the epoxidation treatment preferably comprises the following steps:

[0074] The obtained surface hydroxylated conductive substrate is immersed in a 3-(2,3-epoxypropoxy)propyltrimethoxysilane solution and heated to perform surface epoxidation.

[0075] In the present invention, the 3-(2,3-epoxypropoxy)propyltrimethoxysilane solution is preferably a toluene solution of 3-(2,3-epoxypropoxy)propyltrimethoxysilane.

[0076] In the present invention, the volume ratio of 3-(2,3-epoxypropyloxy)propyltrimethoxysilane to toluene in the toluene solution of 3-(2,3-epoxypropyloxy)propyltrimethoxysilane is preferably 1:1.

[0077] In the present invention, the heating temperature for the surface epoxidation is preferably 110° C., and the heating and holding time is preferably 1 hour.

[0078] In the present invention, the carboxylation treatment comprises the following steps:

[0079] The obtained surface-epoxidized conductive substrate is immersed in an iminodiacetic acid solution and heated to perform surface carboxylation.

[0080] In the present invention, the molar concentration of the iminodiacetic acid solution is preferably 0.75 mol / L.

[0081] In the present invention, the iminodiacetic acid solution preferably further comprises sodium chloride and sodium carbonate. The molar concentration of sodium chloride in the iminodiacetic acid solution is preferably 0.34 mol / L, and the molar concentration of sodium carbonate in the iminodiacetic acid solution is preferably 2.0 mol / L.

[0082] In the present invention, the heating temperature for the surface carboxylation is preferably 70° C., and the heating and holding time is preferably 1 hour.

[0083] The present invention immerses a conductive substrate with a carboxyl group on its surface into a dendritic polymer solution to carry out a first coupling reaction, thereby coupling the dendritic polymer on the surface of the conductive substrate; the dendritic polymer solution includes an organic condensation reagent (hereinafter referred to as the first organic condensation reagent), the dendritic polymer and a buffer solution (hereinafter referred to as the first buffer solution); the dendritic polymer contains amino groups.

[0084] In the present invention, before the conductive substrate with carboxylated surface is immersed in the dendrimer solution, the conductive substrate with carboxylated surface is preferably rinsed with deionized water.

[0085] In the present invention, the dendritic polymer is preferably a polyamidoamine dendritic polymer (PAMAM).

[0086] In the present invention, the first organic condensation reagent preferably includes 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS).

[0087] In the present invention, the first buffer is preferably 4-morpholineethanesulfonic acid buffer (MES buffer).

[0088] In the present invention, in the dendrimer solution, the ratio of the volume of the dendrimer to the volume of the first buffer solution is preferably 0.1:10.

[0089] In the present invention, in the dendrimer solution, the ratio of the mass of the EDC to the volume of the first buffer solution is preferably 1.92 mg:1 mL, and the ratio of the mass of the NHS to the volume of the first buffer solution is preferably 2.87 mg:1 mL.

[0090] In the present invention, the pH value of the first coupling reaction is preferably 7.4. In the present invention, when performing the first coupling reaction, the pH value of the solution of the dendritic polymer dissolved with sodium hydroxide is preferably adjusted to 7.4.

[0091] In the present invention, the reaction temperature of the first coupling reaction is preferably room temperature.

[0092] In the present invention, the reaction time of the first coupling reaction is preferably 30 min.

[0093] The present invention immerses the obtained conductive substrate with the surface coupled to the dendritic polymer into an antibody solution for a second coupling reaction, and couples the antibody on the surface of the conductive substrate to obtain the electrochemical sensor; the antibody solution includes the antibody, an organic condensation reagent (hereinafter referred to as the second organic condensation reagent) and a buffer (hereinafter referred to as the second buffer).

[0094] In the present invention, before the conductive substrate with the surface coupled to the dendritic polymer is immersed in the antibody solution, the conductive substrate with the surface coupled to the dendritic polymer is preferably rinsed with deionized water.

[0095] In the present invention, the antibody is preferably a PD-L1 antibody.

[0096] In the present invention, the second organic condensation reagent preferably includes EDC and NHS.

[0097] In the present invention, the second buffer is preferably MES buffer.

[0098] In the present invention, in the antibody solution, the ratio of the volume of the antibody to the volume of the first buffer solution is preferably 0.01:10.

[0099] In the present invention, in the antibody solution, the ratio of the mass of the EDC to the volume of the first buffer solution is preferably 0.192 mg:1 mL, and the ratio of the mass of the NHS to the volume of the first buffer solution is preferably 0.287 mg:1 mL.

[0100] In the present invention, the pH value of the second coupling reaction is preferably 7.4. In the present invention, when performing the second coupling reaction, the pH value of the solution in which the antibody is dissolved using sodium hydroxide is preferably adjusted to 7.4.

[0101] In the present invention, the reaction temperature of the second coupling reaction is preferably room temperature.

[0102] In the present invention, the reaction time of the second coupling reaction is preferably 30 min.

[0103] In the present invention, after the second coupling reaction, the present invention preferably further comprises performing a surface blocking treatment on the conductive substrate with the surface-coupled antibody to obtain the electrochemical sensor. In the present invention, the surface blocking treatment preferably comprises immersing the conductive substrate with the surface-coupled antibody in a solution of polyethylene glycol dicarboxylic acid (PEG-COOH) to perform a third coupling reaction to obtain the electrochemical sensor. The PEG-COOH solution comprises a third organic condensation reagent (hereinafter referred to as the third organic condensation reagent) and a buffer (hereinafter referred to as the third buffer).

[0104] In the present invention, before the conductive substrate with the surface coupled antibodies is immersed in the PEG-COOH solution, the conductive substrate with the surface coupled antibodies is preferably rinsed with deionized water.

[0105] In the present invention, the third organic condensation reagent preferably includes EDC and NHS.

[0106] In the present invention, the third buffer solution is preferably a phosphate buffer solution.

[0107] In the present invention, in the PEG-COOH solution, the mass concentration of the PEG-COOH is 1 mg / mL.

[0108] In the present invention, in the PEG-COOH solution, the mass of the EDC is 1.92 mg / mL, and the mass of the NHS is 2.87 mg / mL.

[0109] In the present invention, the reaction temperature of the third coupling reaction is preferably room temperature.

[0110] In the present invention, the reaction time of the third coupling reaction is preferably 30 min.

[0111] In the present invention, when the antibody is preferably a PD-L1 antibody, the electrochemical sensor prepared by the above preparation method provided by the present invention can target H1299 cells.

[0112] The present invention provides an electrochemical sensor prepared by the preparation method described in the above technical solution, wherein the electrochemical sensor includes a conductive substrate and an antibody connected to the surface of the conductive substrate by a chemical bond; the electrochemical sensor is used in the circulating tumor cell detection system described in the above technical solution.

[0113] The present invention enables rapid and risk-free detection and measurement of single-level cancer cells. The present invention provides a solution comprising an electrochemical sensor targeting H1299 cells and an H1299 cell image segmentation model. The electrochemical sensor captures H1299 cells using PD-L1 antibodies on the surface of an ITO electrode. After cell capture, the electrode impedance measured by an electrochemical workstation significantly increases, and the sensor exhibits good electrochemical detection performance for H1299 cells. The H1299 cell image segmentation model is constructed using a U-Net network framework and trained using a transfer learning method to improve the model's learning efficiency. The model demonstrates excellent segmentation performance on a self-made H1299 cell image dataset, with an average intersection-over-union ratio of 95.32%. The model also exhibits good generalization performance. The Grad-CAM algorithm is used to visualize the model, and the heat map drawn intuitively displays the key areas of focus of the model during cell segmentation.

[0114] The specific preparation method of the electrochemical sensor targeting H1299 cells is as follows:

[0115] Example 1

[0116] (1) ITO was pre-cleaned with anhydrous ethanol and deionized water for 10 min and dried with nitrogen.

[0117] (2) ITO was immersed in a mixture of 28% ammonia: 30% hydrogen peroxide: water (volume ratio: 1:1:6) and heated at 85°C for 30 min to obtain a hydroxylated surface. The ITO was washed with deionized water and dried with nitrogen.

[0118] (3) Immerse the ITO obtained in step (2) in 20 mL of a 1:1 solution of 3-(2,3-epoxypropoxy)propyltrimethoxysilane and toluene and heat at 110°C for 1 h to form an epoxy monolayer on the surface. Rinse the ITO with plenty of acetone.

[0119] (4) Immerse the ITO obtained in step (3) in 20 mL of 0.75 mol / L iminodiacetic acid solution (containing 0.34 mol / L sodium chloride and 2.0 mol / L calcium carbonate) and heat at 70°C for 1 h to obtain a carboxylated surface. Rinse the ITO with plenty of deionized water.

[0120] (5) Add 100 μL of dendritic PAMAM, 19.2 mg of EDC, and 28.7 mg of NHS to 10 mL of MES buffer, mix well, and react at room temperature for 5 min. Raise the pH of the mixture to 7.4 using NaOH solution. Immerse the ITO obtained in step (4) in the solution for 30 min and rinse with plenty of deionized water.

[0121] (6) Add 10 μL of PD-L1 antibody, 1.92 mg of EDC, and 2.87 mg of NHS to 10 mL of MES buffer, mix well, and react at room temperature for 5 minutes. Use NaOH solution to raise the pH of the mixture to 7.4. Immerse the ITO obtained in step (5) in the solution for 30 minutes and rinse with plenty of deionized water.

[0122] (7) Prepare 1 mg / mL PEG-COOH (containing 1.92 mg / mL EDC and 2.87 mg / mL NHS) in PBS. Immerse the ITO obtained in step (6) in the solution for 30 min to perform surface blocking. Rinse with deionized water and dry with nitrogen to obtain the PD-L1 antibody-modified ITO electrode.

[0123] Comparative Example 1

[0124] (1) ITO was pre-cleaned with anhydrous ethanol and deionized water for 10 min and dried with nitrogen;

[0125] (2) ITO was immersed in a mixture of 28% ammonia: 30% hydrogen peroxide: water (volume ratio: 1:1:6) and heated at 85°C for 30 min to obtain a hydroxylated surface. The ITO was washed with deionized water and dried with nitrogen.

[0126] (3) Immerse the ITO obtained in step (2) in 20 mL of a 1:1 volume ratio 3-(2,3-epoxypropoxy)propyltrimethoxysilane:toluene solution and heat at 110°C for 1 h to form an epoxy monolayer on the surface. Rinse the ITO with sufficient acetone.

[0127] (4) Immerse the ITO obtained in step (3) in 20 mL of 0.75 mol / L iminodiacetic acid solution (containing 0.34 mol / L sodium chloride and 2.0 mol / L calcium carbonate) and heat at 70°C for 1 h to obtain a carboxylated surface. Rinse the ITO with plenty of deionized water.

[0128] (5) Add 100 μL of dendritic PAMAM, 19.2 mg of EDC, and 28.7 mg of NHS to 10 mL of MES buffer, mix well, and react at room temperature for 5 min. Raise the pH of the mixture to 7.4 using NaOH solution. Immerse the ITO obtained in step (4) in the solution for 30 min and rinse with plenty of deionized water.

[0129] (6) Prepare 1 mg / mL PEG-COOH (containing 1.92 mg EDC and 2.87 mg NHS) in PBS. Immerse the ITO obtained in step (5) in the solution for 30 min to perform surface blocking. Rinse with deionized water and dry with nitrogen to obtain an ITO electrode modified only with the dendrimer.

[0130] In this comparative example, except that the PD-L1 antibody was not modified, the remaining materials and steps were the same as those in Example 1.

[0131] Test example

[0132] Microscope image of the ITO electrode surface Figure 1 As shown, Figure 1 (a) is a bare ITO electrode. Figure 1 (b) is a dendrimer-modified ITO electrode. The cell capture experiment was carried out using the ITO electrodes obtained in Example 1 and Comparative Example 1. The experimental results are shown in FIG. Figure 2 As shown, Figure 2 (a) is the impedance spectrum of antibody-modified ITO. Figure 2 (b) is the impedance spectrum of ITO modified with dendrimer; Figure 2 (c) is a microscope image of antibody-modified ITO. Figure 2 (d) shows a microscopic image of dendrimer-modified ITO. The antibody-modified ITO electrode (Example 1) shows a significant increase in electrode impedance after cell capture, and H1299 cells can be observed adsorbed on the electrode surface from the microscopic image. However, the ITO electrode modified with only the dendrimer (Comparative Example 1) shows no significant change in impedance before and after cell capture, and no cells can be observed adsorbed on the electrode surface from the microscopic image. This demonstrates that the PD-L1 antibody-modified ITO electrode has the ability to capture H1299 cells.

[0133] The present invention is used for medical image detection. Specifically, the present invention constructs a micro-nano analysis system, which specifically includes two parts: an electrochemical sensor targeting H1299 cells and an H1299 cell image segmentation model. By modifying the ITO electrode, the PAMAM dendrimer and the PD-L1 antibody are fixed on the surface of the ITO to construct an electrochemical sensor for capturing H1299 cells. The H1299 cell image segmentation model is constructed by using the U-Net network framework and the transfer learning method, and the model is visualized using the Grad-CAM algorithm. The sensor of the present invention shows good electrochemical detection performance for H1299 cells, and the algorithm model of the present invention shows excellent segmentation performance for H1299 cell images, with an average intersection-over-union ratio of 95.32%. The model has good generalization performance, and the key areas of focus of the model during cell segmentation are intuitively displayed in the form of a heat map.

[0134] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0135] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A circulating tumor cell detection system, characterized in that: include: An acquisition module is used to acquire cell images using an electrochemical sensor and a microscope; the electrochemical sensor captures H1299 cells using a PD-L1 antibody on the surface of an ITO electrode. After cell capture, the electrode impedance measured by an electrochemical workstation significantly increases, showing good electrochemical detection performance for H1299 cells; the preparation method of the electrochemical sensor comprises: sequentially subjecting a conductive substrate to hydroxylation, epoxidation, and carboxylation in a solution; the conductive substrate is an ITO electrode; the resulting surface-carboxylated conductive substrate is immersed in a dendritic polymer solution for a first coupling reaction, thereby coupling the dendritic polymer to the surface of the conductive substrate; the dendritic polymer solution comprises an organic condensation reagent, a dendritic polymer, and a buffer solution, wherein the dendritic polymer contains amino groups; the resulting conductive substrate with the surface-coupled dendritic polymer is immersed in an antibody solution for a second coupling reaction, thereby coupling the antibody to the surface of the conductive substrate to obtain the electrochemical sensor; the antibody solution comprises an antibody, an organic condensation reagent, and a buffer solution; the antibody is a PD-L1 antibody; A segmentation and extraction module is used to perform segmentation and extraction based on the cell image using a cell image segmentation model to obtain a segmentation result; the cell image segmentation model is obtained by training a U-Net network using a transfer learning algorithm; The Grad-CAM algorithm processing module is used to obtain a tumor cell heat map using the Grad-CAM algorithm according to the segmentation results.

2. The circulating tumor cell detection system according to claim 1, characterized in that The segmentation and extraction module includes a backbone feature extraction network, an enhanced feature extraction network and a prediction network connected in sequence; The backbone feature extraction network is used to perform preliminary extraction on the cell image to obtain a plurality of preliminary effective features; The enhanced feature extraction network is used to perform upsampling, feature fusion and convolution on multiple preliminary effective features to obtain output features; The prediction network is used to perform convolution on the output features to obtain a segmentation result.

3. The circulating tumor cell detection system according to claim 1, characterized in that It also includes a pre-processing module, which is used to crop the cell image.

4. The circulating tumor cell detection system according to claim 2, characterized in that The backbone feature extraction network is the VGG16 network.

5. The circulating tumor cell detection system according to claim 1, characterized in that The hydroxylation treatment comprises the following steps: The conductive substrate is immersed in an alkaline solution and heated to perform surface hydroxylation; the alkaline solution includes ammonia water, and the mass concentration of NH3 in the alkaline solution is 28%; the heating temperature for the hydroxylation is 85°C, and the heating and holding time is 30 minutes.

6. The circulating tumor cell detection system according to claim 5, characterized in that The epoxidation treatment comprises the following steps: The obtained surface hydroxylated conductive substrate was immersed in a 3-(2,3-epoxypropoxy)propyltrimethoxysilane solution and heated to perform surface epoxidation; the heating temperature for the surface epoxidation was 110° C., and the heating and holding time was 1 hour.

7. The circulating tumor cell detection system according to claim 6, characterized in that The carboxylation treatment comprises the following steps: The obtained surface epoxidized conductive substrate is immersed in an iminodiacetic acid solution and heated for surface carboxylation; the molar concentration of the iminodiacetic acid solution is 0.75 mol / L; the heating temperature for the surface carboxylation is 70° C., and the heating and holding time is 1 hour.

8. The circulating tumor cell detection system according to claim 1, characterized in that The dendritic polymer is a polyamide-amine dendritic polymer; the pH value of the first coupling reaction is 7.4±0.5; and the reaction time of the first coupling reaction is 30 minutes.

Images