Exosome capture and label-free detection platform based on SERS (Surface Enhanced Raman Scattering)

Abstract

The invention discloses an exosome capture and label-free detection platform based on SERS (Surface Enhanced Raman Scattering), and belongs to the technical field of biomedicine. The specific preparation method comprises the following steps: preparing a gold nano island array substrate, modifying the gold nano island array substrate with SH-PEG and CP05 polypeptide in different proportions, and washing to obtain the functionalized SERS chip. The chip can efficiently capture exosomes in 100 microliters of serum samples within one hour, is directly used for rapid detection of exosome Raman fingerprint signals, and is combined with a deep learning algorithm for spectral analysis. Clinical verifications show that the method shows relatively high sensitivity and specificity on the detection of early cancers, and the average diagnosis accuracy of ten cancer types reaches 93.89%. The SERS chip prepared by the invention has the characteristics of automation, low cost and high throughput, is suitable for early screening of cancers, and has high application value and economic benefit.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical detection technology and relates to an integrated exosome capture / label-free detection chip based on surface-enhanced Raman scattering (SERS) technology, as well as a preparation method and system thereof. In particular, it relates to a system and method for realizing automatic exosome capture / label-free detection and early cancer diagnosis by using a gold nanoisland array SERS chip functionalized with CP05 polypeptide. Background Art

[0002] Exosomes are nanoscale extracellular vesicles produced by various cells through a process of endocytosis, fusion, and secretion. They typically range in diameter from 30 to 150 nanometers and are widely present in various bodily fluids, including blood, urine, saliva, and pleural and ascites fluids. Exosomes are formed by the release of multivesicular bodies (MVBs) following fusion with the cell membrane. Their outer membrane, a typical lipid bilayer, encapsulates a variety of bioactive molecules, including proteins, lipids, mRNA, miRNA, and lncRNA. Exosomes are important mediators of intercellular communication and participate in regulating various biological processes, including cell proliferation, apoptosis, immune regulation, and angiogenesis. In the field of oncology, tumor-derived exosomes carry a rich array of molecular markers. Their surface expression of transmembrane proteins such as CD63, CD9, and CD81 is high, and their interiors are rich in specific information closely related to tumorigenesis, invasion, and metastasis. Therefore, they are considered highly promising non-invasive biomarkers and are widely used in early cancer screening, diagnosis, and treatment efficacy monitoring.

[0003] Although the research on exosomes is becoming increasingly in-depth, it still faces many technical difficulties in the process of separation and detection. Currently, the commonly used methods for extracting exosomes, such as ultracentrifugation, density gradient centrifugation, membrane filtration and immunomagnetic beads, generally have problems such as complex operation, long cycle, low separation efficiency, and insufficient purity, which makes it difficult to meet the clinical demand for rapid, high-throughput and standardized operations. For example, CN117916817A extracts exosomes by size exclusion chromatography. This method has low purity, may contain protein impurities, and has limited processing volume and low throughput. CN110118765A extracts exosomes by kit, which is difficult to support high-throughput experiments and has a high economic burden. CN107741416B uses ultracentrifugation, which has complex operation, long cycle and low separation efficiency. CN105628672B uses a magnetic enrichment method, but the steps are relatively cumbersome and there is serious interference from foreign proteins. In terms of detection, traditional methods such as Western blotting, ELISA, and flow cytometry require high sample volumes, have limited sensitivity, and rely heavily on labeling, making rapid, non-invasive, and highly accurate exosome analysis difficult. Therefore, there is an urgent need to develop a label-free, rapid, simple, and adaptable exosome detection strategy suitable for complex biological sample backgrounds to enable its widespread promotion and application in clinical practice.

[0004] Surface-enhanced Raman scattering (SERS) technology, due to its ultra-high sensitivity, molecular specificity, and ease of operation, has shown great potential in bioanalysis, particularly in disease marker detection. Based on the localized surface plasmon resonance effect induced by noble metal nanostructures, SERS can amplify Raman signals by millions or even billions of times, enabling the detection of extremely low-abundance target molecules. Requiring no complex manipulations such as fluorescence or enzyme labeling, SERS can directly identify molecular fingerprints in complex backgrounds. The detection process is rapid and highly reproducible, making it suitable for high-throughput and clinical scenarios. In recent years, with the continuous advancement of functionalized SERS substrate preparation technology, SERS has gradually evolved into a comprehensive platform that can simultaneously capture, identify, and enhance signals, providing strong technical support for the precise quantification and typing analysis of complex particles such as exosomes.

[0005] The application of SERS technology to exosome detection has significant advantages. By designing a functionalized SERS substrate with targeted recognition capabilities, it is not only possible to selectively capture exosomes of the target source, but also to achieve high-sensitivity signal reading at the same time. However, the solution proposed by CN119198677A uses Raman reporter molecule labeling detection. The exosome biological information detected by this method is limited, and it is impossible to extract the exosome biological fingerprint information; CN116265919A uses a substrate material with a Raman background signal, which causes the collected exosome SERS spectrum to be interfered with, masking the inherent fingerprint spectral characteristics of the biological sample to be tested, which is not conducive to disease marker analysis and diagnosis. This solution can complete the efficient detection of exosome surface proteins, internal nucleic acid components or overall spectral characteristics without the need for pre-labeling Raman dyes, and the collected exosome SERS fingerprint spectrum is free of substrate background interference, which is suitable for label-free Raman rapid detection of exosomes in tumor liquid biopsy. The high sensitivity, high selectivity and rapid response characteristics of the SERS platform in this solution provide new solutions for the clinical application of exosomes in early screening, personalized diagnosis and treatment, and prognosis monitoring of major diseases such as cancer. It has important research value and broad industrial prospects. Summary of the Invention

[0006] The present invention provides a system and method for fully automatic, high-throughput capture and label-free detection of exosomes based on a CP05 functionalized SERS chip for precise cancer diagnosis. This system aims to address the low accuracy and specificity of existing cancer serum diagnostics. Furthermore, as a new biomarker, the detection of exosomes is cumbersome, has a low degree of automation, and lacks diagnostic accuracy, among other technical bottlenecks.

[0007] A method for preparing a functionalized SERS chip, comprising: Step 1: Ultrasonic cleaning of the glass slide with acetone, ethanol, and deionized water in sequence; Step 2: Then immerse in chloroauric acid solution; Step 3: After taking it out, immerse it in NaBH4 solution to form a gold seed layer; Step 4: Transfer the glass slide to the growth solution for reaction to prepare the gold nanoisland array substrate; Step 5: Incubate the gold nanoisland array substrate with molecules that can specifically bind to exosomes (CP05 polypeptide, CD63 antibody, membrane fusion molecules SH-PEG-DSPE, SH-PEG-CHOL, SH-DNA-CHOL) and SH-PEG in a PBS solution; then remove and rinse with PBS to obtain a functionalized SERS chip.

[0008] Preferably, the interlayer spacing of the gold nano-island array substrate is 10-50 nm, and the height is 20-80 nm.

[0009] Preferably, the glass support is ultrasonically cleaned in acetone, ethanol, and deionized water, sequentially, with each solvent treatment lasting 3 to 10 minutes to remove organic impurities and particulate contaminants from the support surface. The cleaned glass slide is then immersed in a 0.005-0.02% (w / v) chloroauric acid solution for 3 to 10 minutes. The cleaned glass slide is then immediately transferred to a freshly prepared 0.05-0.2 mol / L sodium borohydride solution for reduction for 30 seconds to 2 minutes, forming a uniform gold seed layer on the surface of the glass slide. Following the formation of the gold seeds, the glass slide is transferred to a growth solution consisting of 1-2 mmol / L chloroauric acid and 1-2 mmol / L hydroxylamine hydrochloride for 3 to 10 minutes at 20-30°C to complete the growth of the gold nanoislands. After the reaction, the glass slide is thoroughly rinsed with deionized water and dried at room temperature to obtain a gold nanoisland array substrate with surface-enhanced properties. The method is simple to operate, has mild reaction conditions, and the surface structure of the obtained substrate is uniform, which is suitable for large-scale and reproducible preparation.

[0010] Preferably, the pre-prepared gold nanoisland array substrate is immersed in phosphate buffered saline (PBS, pH 7.4) containing 25 μM CP05 peptide and 50 μM thiol-polyethylene glycol (SH-PEG) and incubated at 4°C for 4 hours. The substrate is then removed and rinsed three times with PBS to remove unbound molecules, thereby obtaining a functionalized SERS chip. In this step, the CP05 peptide and SH-PEG simultaneously form stable gold-sulfur bonds with the gold nanoisland array surface through the thiol groups, constructing a co-modified monolayer structure with specific functions. The gold nanoisland array provides a substrate for surface-enhanced Raman scattering (SERS) signals; the CP05 peptide specifically recognizes CD63 protein expressed on the surface of exosomes, constituting the targeted capture functional unit of the chip; and the SH-PEG occupies the remaining vacancies on the gold surface, forming a hydrophobic antifouling barrier against nonspecific adsorption, effectively suppressing background interference in complex biological samples such as serum. This co-modified structure organically combines signal enhancement, targeted capture, and antifouling functions, providing a stable and reliable surface platform for highly sensitive and selective label-free SERS detection of exosomes.

[0011] The present invention also discloses a CP05 polypeptide functionalized SERS chip, which is characterized by: being prepared by the above method; the CP05 polypeptide sequence is CRHSQMTVTSRL, forming an Au-S bond with the gold nano-island array through the thiol group of the terminal cysteine, and the molar ratio of the CP05 polypeptide to SH-PEG is 1:2.

[0012] The present invention also discloses a SERS-based exosome capture and label-free detection platform, wherein the above-mentioned SERS chip is mounted on the platform; the platform is characterized in that it includes the following modules: Dropping module: The sample is accurately dispensed to the designated position of the reagent kit through a high-precision micropipette pump. It supports multi-channel synchronous operation and multi-well layout, and can automatically switch to cleaning mode after the drop is completed; Oscillation module: uses an adjustable frequency vortex oscillator to ensure uniform distribution of the sample by setting the oscillation intensity and time parameters, and has a splash-proof design to prevent liquid overflow; Cleaning and aspiration module: Driven by a multi-axis robotic arm, it integrates a controllable flow nozzle and a high-efficiency vacuum aspiration device, and supports a default two-round PBS automatic cleaning program with adjustable times; Sample delivery and positioning module: A high-precision conveyor device is used to accurately move the test kit under the Raman microscope head and seamlessly connect it to its motorized sample platform. Optical or laser sensors are used to achieve real-time position feedback and coordinate matching. Control module: Based on PLC or industrial computer architecture, supports Modbus and TCP / IP communication protocols, and is deeply linked to the Raman microscopy detection system;

[0013] Analysis module: CNN-MLP integrated algorithm is used to analyze the SERS information of exosome fingerprints.

[0014] The present invention also discloses a method for automated exosome capture and label-free detection, based on the aforementioned detection platform. The method features automated modules for the detection process, including precise sample addition to designated locations in the reagent chamber via a high-precision micropipette pump, supporting multi-channel simultaneous operation. Subsequently, the sample is mixed using a frequency-adjustable vortex oscillator to ensure uniform sample distribution and prevent liquid splashing. A multi-axis robotic arm with an integrated nozzle and vacuum aspiration device automatically performs PBS washes, supporting multiple cycles. After sample processing, the reagent chamber is automatically positioned under the Raman microscope platform via a high-precision transport system, and real-time position calibration is achieved using optical or laser sensors. During Raman acquisition, the system randomly scans five different locations within the reagent reaction zone, with an acquisition range of 10 μm × 10 μm and a step interval of 1 μm. The excitation current is adjustable to 5–10 mA to ensure representative spectral data and acquisition stability, thereby achieving efficient and accurate Raman detection.

[0015] The present invention also discloses a terminal device, characterized in that the terminal device includes: a processor, a memory, a communication interface and a bus; the processor, the memory and the communication interface are connected through the bus and complete communication with each other; the memory stores executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the above-mentioned method.

[0016] Beneficial effects

[0017] (1) The CP05 functionalized SERS chip has ultra-high sensitivity and can achieve fully automatic, high-throughput capture and detection of exosomes.

[0018] (2) The fully automated detection system shortens the traditional 6-8 hour exosome detection process to within 1 hour, and the manual operation time is less than 5 minutes, significantly improving the detection efficiency.

[0019] (3) The deep learning algorithm used in this study performed well in a validation study of 655 clinical samples: the sensitivity for detecting stage I cancer was 99.56%, the specificity was 92.50%, and the average diagnostic accuracy for ten cancer types was 93.89%. This improved the bottleneck problem of low accuracy and sensitivity in clinical serum cancer detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Scanning electron microscopy images of serum exosomes captured by the functionalized SERS chip; Figure 2Based on the SERS fingerprint information of serum exosome samples of various cancer types, the CNN-MLP algorithm was used to obtain multi-cancer diagnosis results; Figure 3 ROC curve of the CNN-MLP algorithm for multi-cancer classification model; Figure 4 Different functionalization methods of SERS chips and their effects on preventing nonspecific adsorption of albumin; Figure 5 Capture effect of fluorescently labeled exosomes on SERS chips functionalized with CP05:PEG (0.5:1) and CP05:PEG (0:1); Figure 6 Schematic diagram of the multifunctional kit structure; Figure 7 Scanning electron microscope image of SERS chip and photo of the kit; Figure 8 Clinical validation data (label-free SERS fingerprint spectra of serum exosomes from cancer patients). DETAILED DESCRIPTION

[0021] See also Figure 1-8 As shown, the following examples are given.

[0022] Example 1:

[0023] Functionalized SERS chip preparation method.

[0024] The slides were ultrasonically cleaned with acetone, ethanol, and deionized water for 15 minutes in sequence to effectively remove organic impurities, oil stains, and particulate matter, ensuring that the slide surface was clean and had good lyophilicity, providing an ideal surface state for subsequent gold seed attachment.

[0025] The glass slide was then briefly immersed in a 0.01% chloroauric acid solution for 2 minutes to form a preliminary adsorbed gold ion layer. The concentration and reaction time were optimized to stably control the seed density and ensure uniform nucleation. Immediately thereafter, it was transferred to a freshly prepared 0.1 M NaBH4 solution for 1 minute. The gold ions were quickly reduced by a strong reducing agent to generate a dense and uniform gold nanoseed layer, and to avoid excessive growth or agglomeration of the nanoparticles, thereby forming a good subsequent growth template. The treated glass slide was transferred to a growth solution consisting of 1.5 mM HAuCl4 and 1.5 mM hydroxylamine hydrochloride and reacted at 25 °C for 5 minutes. This concentration ratio and temperature control have been screened by a large number of preliminary experiments and can stably induce the directional growth of gold nanoparticles to form a uniformly distributed gold nanoisland array with rich nanoscale gap structures, which is conducive to the generation of strong "hot spots" to achieve surface-enhanced Raman signal amplification. The reaction time is controlled within 5 minutes, which can effectively avoid the merging of island structures and maintain the optimal surface roughness and enhancement effect. The schematic diagram and electron microscope image are shown in the figure. Figure 6 , Figure 7 As shown. The prepared gold nanoisland array substrate was immersed in PBS buffer (pH 7.4) containing 25 μM CP05 polypeptide and 50 μM SH-PEG and incubated at 4 °C for 4 hours. CP05 is a polypeptide that specifically recognizes exosomes. Its concentration is set at 25 μM to ensure that a capture layer is fully covered on the surface and improve the binding efficiency; SH-PEG concentration is 50 μM. As an anti-fouling modifier, it forms a stable bond with the gold surface through the thiol group to construct a hydrophobic anti-adsorption layer, which effectively reduces non-specific binding. The concentration ratio of the two is optimized to take into account the synergistic performance of the functional layer and the anti-fouling layer, as shown in FIG. Figure 4 As shown. A long, low-temperature incubation (7 h at 4°C) ensures slow and uniform assembly of molecules on the gold surface, improving the stability and uniformity of the functional layer. Three rinsing cycles with PBS after incubation fully remove unbound molecules and minimize background interference. The optimal number of rinses is determined to eliminate residual molecules without disrupting the bound structure.

[0026] Example 2

[0027] The slide pretreatment steps were consistent with those in Example 1. Subsequently, the slide was briefly immersed in a 0.01% chloroauric acid solution for 2 minutes to form a preliminary adsorbed gold ion layer. This step employed relatively low reducing conditions to evaluate the effects of low concentration and short time on seed density regulation. The slide was then immediately transferred to a freshly prepared 0.05 M NaBH4 solution for reduction for 30 seconds. This represents the lower limit of the defined range, and experiments have shown that it can rapidly generate a small, uniformly distributed gold nanoseed layer suitable for subsequent controlled growth, with no apparent agglomeration. The treated slide was transferred to a growth solution consisting of 1.0 mM HAuCl4 and 1.0 mM hydroxylamine hydrochloride and reacted at 20°C for 3 minutes. Under these conditions, the gold nanoislands were small in size but evenly arranged, with ample inter-island gaps and good SERS activity. Functionalization procedures were consistent with those in Example 1. The slides were immersed in PBS buffer containing 25 μM CP05 peptide and 50 μM SH-PEG, incubated at 4°C for 4 hours, and then rinsed three times with PBS to obtain a functionalized substrate under low-parameter conditions. This substrate exhibited excellent background control and exosome binding capacity.

[0028] Example 3

[0029] The slide pretreatment steps were consistent with those in Example 1. Subsequently, the slide was briefly immersed in a 0.01% chloroauric acid solution for 2 minutes to pre-adsorb gold ions and provide the same starting state for high-concentration reduction experiments. The slide was then immediately transferred to a 0.2 M NaBH₄ solution for 2 minutes of reduction. This condition represents the upper limit for the reducing agent concentration and reaction time, ensuring rapid reduction while forming a dense gold seed layer. No significant agglomeration was observed experimentally, indicating that it is suitable for the formation of large-particle island structures. The treated slide was transferred to a growth solution consisting of 2.0 mM HAuCl₄ and 2.0 mM hydroxylamine hydrochloride and reacted at 30°C for 10 minutes. The gold nanoislands formed under these conditions were larger in size, with slightly smaller inter-island spacing, while still retaining nanoscale voids. The structure was stable and possessed a certain degree of enhancement, making it suitable for detection scenarios requiring more demanding enhancement effects. Functionalization was also performed using a PBS buffer solution containing 25 μM CP05 and 50 μM SH-PEG. The solution was incubated at 4°C for 4 hours and then rinsed three times with PBS. Although the gold islands are relatively large, their surface modification is uniform, their binding ability is good, and they have stable optical response.

[0030] Example 4

[0031] A CP05 polypeptide-functionalized SERS chip is characterized by being prepared using the preparation method of Example 1; the CP05 polypeptide sequence is CRHSQMTVTSRL, and a stable Au-S covalent bond is formed with the gold nano-island array through the terminal thiol group, and the molar ratio of the CP05 polypeptide to SH-PEG is 1:2.

[0032] The CP05 peptide sequence, CRHSQMTVTSRL, was derived from the screening and optimization of specific binding sites on the exosome surface. It exhibits high selectivity and affinity, effectively identifying and capturing target exosomes and ensuring detection specificity and sensitivity. The peptide is terminally modified with a thiol (–SH) group, which forms a strong and stable gold-sulfur (Au–S) covalent bond with the gold nanoisland surface. This ensures the peptide is firmly fixed to the SERS substrate, preventing it from falling off during detection and improving the chip's stability and reusability.

[0033] The molar ratio of CP05 to SH-PEG is designed to be 1:2, mainly based on the following considerations: CP05 as the capture layer undertakes the specific binding task, while SH-PEG as the anti-fouling layer reduces nonspecific adsorption by forming dense polyethylene glycol chains, reducing background signals and impurity interference. This ratio can fully utilize the anti-fouling properties of PEG while ensuring sufficient capture molecule coverage, thereby enhancing the overall biocompatibility and stability of the chip surface. Experimental verification shows that the functionalized chip at this molar ratio exhibits excellent capture efficiency and signal stability, with a high enhancement factor and good batch-to-batch reproducibility, meeting the requirements of high-sensitivity and high-specificity detection, such as Figure 1 , Figure 5 shown.

[0034] Example 5

[0035] The SERS-based exosome capture and label-free detection platform achieves efficient and automated exosome capture and Raman detection through the integrated integration of the aforementioned functional modules. The sample is first dispensed into the designated wells of the reagent kit using a high-precision micropipette pump at a preset volume using the dropper module. Multi-channel parallel processing improves throughput while preventing cross-contamination. The oscillation module then activates a variable-frequency vortex oscillator, uniformly mixing the sample within the kit according to software-defined speed and time parameters, preventing sample precipitation and uneven distribution and ensuring detection sensitivity and accuracy. The cleaning and aspiration module uses a multi-axis robotic arm to control the nozzle and vacuum aspiration device for automated PBS rinsing. Multiple cycles effectively remove residual impurities and ensure a pure environment for the next step of detection. The sample delivery and positioning module precisely transfers the kit to the detection platform of the Raman microscope. Real-time position correction using optical or laser sensors ensures accurate alignment of the Raman laser with the collection area.

[0036] The entire inspection process is centrally managed by a control module. Based on a PLC or industrial computer architecture, the control module leverages Modbus and TCP / IP communication protocols for deep integration with various hardware modules. The software interface supports one-click start and parameter customization, greatly simplifying the operation process. Upon completion of the inspection, the control module automatically triggers Raman spectral data acquisition, ensuring synchronized data storage and preventing errors caused by human intervention.

[0037] The collected SERS spectral data is intelligently processed by the CNN-MLP integrated algorithm in the analysis module. First, the dynamic baseline correction module automatically identifies and corrects spectral baseline drift, eliminating background interference and improving the signal-to-noise ratio. Subsequently, intensity normalization is performed on specific spectral regions to standardize the spectral scale across samples, ensuring fairness and stability in subsequent analysis. The preprocessed data is then fed into a convolutional neural network (CNN) component, which comprises three convolutional layers employing 16, 32, and 64 filters, respectively. This automatically extracts key features and local patterns in the spectrum, capturing subtle Raman signal differences. The extracted features are then passed to a multilayer perceptron (MLP) component, consisting of two hidden layers, each containing 64 neurons. This layer integrates and discriminates complex features through nonlinear transformations. Finally, the integrated algorithm outputs a detection conclusion, including the presence and classification of exosomes.

[0038] This intelligent analysis process greatly improves the accuracy and robustness of detection, reduces the subjectivity and limitations of traditional manual feature extraction, and realizes an automated closed loop from spectral acquisition to result output. It is suitable for the needs of high-throughput and accurate exosome fingerprint information detection in clinical and scientific research.

[0039] Example 6

[0040] A method for automatic capture and detection of exosomes, based on the SERS-based exosome capture and label-free detection platform described in Example 3, specifically comprising the following steps:

[0041] Step 1: Sample preparation: Pre-treat the biological sample to be tested (such as serum, saliva, or urine) at room temperature and, if necessary, dilute it appropriately to achieve the volume and concentration range required by the platform detection.

[0042] Step 2: Sample loading: The pretreated sample is automatically added to the designated well of the kit pre-loaded with the CP05 peptide-functionalized SERS chip using a high-precision micropipette pump, ensuring that the sample volume is accurately 0.1 mL to avoid sample waste and cross-contamination.

[0043] Step 3: Mix the sample evenly: Start the vortex shaker and shake the sample evenly according to the set speed (50–300 rpm) and time (0–12h) parameters to promote efficient binding of exosomes and functionalized peptides while preventing liquid splashing.

[0044] Step 4: Automatic cleaning: The nozzle and vacuum aspiration device controlled by a multi-axis robotic arm are used to automatically rinse the wells of the reagent kit with PBS buffer to remove unbound impurities and interferences, thereby improving the detection signal-to-noise ratio. The cleaning program can set the number of cycles as needed.

[0045] Step 5: Sample positioning and spectrum acquisition: The reagent kit is accurately moved to the Raman microscope detection position through the platform's high-precision conveyor device. The optical sensor alignment feedback ensures that the Raman laser is precisely aligned with the surface of the functionalized SERS chip. During acquisition, 5 different sampling points on the chip are randomly selected. The acquisition area of each point is 10 μm × 10 μm, the step spacing is 1 μm, and the acquisition condition current is set in the range of 5–10 mA to obtain representative spectral data such as Figure 8 shown.

[0046] Step 6: Spectral data preprocessing: The system automatically performs dynamic baseline correction and regional intensity normalization on the collected Raman spectral data to eliminate the influence of environmental background and instrument drift and improve data quality.

[0047] Step 7: Intelligent Analysis and Detection: The preprocessed spectral data is fed into an integrated CNN-MLP deep learning model. The convolutional neural network layer automatically extracts spectral features, and the multi-layer perceptron layer performs feature fusion and classification, ultimately outputting the detection results, including the presence and classification of exosomes.

[0048] Step 8. Result output and storage: The platform displays the test conclusions in real time through the software interface, and automatically stores the test data and result reports. It supports exporting to multiple formats for subsequent analysis and archiving.

[0049] Through the above steps, this method achieves automated, efficient capture and high-sensitivity detection of exosomes, significantly improving the accuracy and repeatability of detection, and is suitable for clinical application and scientific research promotion.

[0050] Example 5: The steps and methods are essentially the same as those in Example 1, except that the gold substrate is replaced with a silver substrate. First, prepare a silver colloidal solution: add a predetermined amount of silver nitrate (AgNO₃) to 100 mL of deionized water. Heat to boiling, then quickly add sodium citrate solution as a reducing agent. Allow to react for approximately 30 minutes under continuous stirring and heating until the solution exhibits a uniform pale yellow or brownish-yellow color, indicating successful formation of silver nanoparticles. The resulting silver sol serves as the colloidal solution for subsequent assembly. Next, ultrasonically clean a 0.5 cm × 2 cm glass slide in acetone and deionized water to remove surface impurities. The cleaned slide is then immersed in a 2% polyvinylpyrrolidone (PVP) ethanol solution for approximately 2 hours for surface modification. After treatment, the slide is repeatedly rinsed with anhydrous ethanol and air-dried. The treated slide is then allowed to stand in the silver sol for approximately 6 hours, allowing the silver nanoparticles to form a uniform monolayer self-assembled structure on its surface.

[0051] Example 7: A SERS fingerprint database was constructed using a deep learning algorithm. This method first used a functionalized SERS chip to collect Raman spectra of standard biological samples (e.g., exosomes) under uniform detection parameters, ensuring sufficient representativeness and reproducibility for each sample type. After acquisition, the raw spectral data was preprocessed with dynamic baseline correction, denoising, and normalization to construct a standard SERS fingerprint database.

[0052] Based on this database, a CNN-MLP ensemble model was trained. The CNN module extracts spectral features, while the MLP module performs classification and discrimination. The model uses supervised learning, with processed spectral data as input and sample class labels as output. Dropout and early stopping were introduced during training to prevent overfitting. The Adam optimizer was used, and the cross-entropy loss function was used.

[0053] The resulting model can automatically identify unknown samples and determine their categories, such as Figure 2 , Figure 3 As shown in Figure 2, intelligent SERS fingerprint analysis is achieved. This database has the ability to be continuously expanded and updated, and can be widely used in high-throughput detection and precise identification of complex biological samples such as exosomes.

[0054] The present invention has three major technological breakthroughs: 1) CP05 polypeptide specifically recognizes CD63 protein on the surface of exosomes, achieving efficient capture; 2) Integrated chip design enables in situ detection of exosomes, eliminating the traditional separation step; 3) Integration of fully automated sampling, cleaning, and sample delivery modules, control and software systems, SERS detection modules, and data analysis modules enables fully automated enrichment and detection of exosomes from serum samples; 4) Integrating a deep learning algorithm to construct a SERS fingerprint database enables accurate diagnosis of common multiple cancers with a single test.

[0055] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for preparing a functionalized SERS chip, characterized in that: include: Step 1: Ultrasonic cleaning of the glass slide with acetone, ethanol, and deionized water in sequence; Step 2: Then immerse in chloroauric acid solution; Step 3: After taking it out, immerse it in NaBH4 solution to form a gold seed layer; Step 4: Transfer the glass slide to the growth solution for reaction to prepare the gold nanoisland array substrate; Step 5: Incubate the gold nanoisland array substrate with a PBS solution of molecules that specifically bind to exosomes and SH-PEG; remove and rinse with PBS to obtain a functionalized SERS chip.

2. The method for preparing a functionalized SERS chip according to claim 1, wherein: The interlayer spacing of the gold nanoisland array substrate is 10-50 nm and the height is 20-80 nm; the molecules of the exosomes are CP05 polypeptide, CD63 antibody, membrane fusion molecules SH-PEG-DSPE, SH-PEG-CHOL, and SH-DNA-CHOL.

3. The method for preparing a functionalized SERS chip according to claim 1, wherein: The glass carrier was ultrasonically cleaned in acetone, ethanol, and deionized water in sequence, with each solvent treatment time being 3 to 10 minutes. The cleaned glass slide was then immersed in a 0.005-0.02% (w / v) chloroauric acid solution for reaction for 3 to 10 minutes, and then immediately transferred to a freshly prepared 0.05-0.2 mol / L sodium borohydride solution for reduction for 30 seconds to 2 minutes to form a uniform gold seed layer on the surface of the glass slide. After the gold seeds were formed, the glass slide was transferred to a growth solution consisting of 1-2 mmol / L chloroauric acid and 1-2 mmol / L hydroxylamine hydrochloride and reacted at 20-30°C for 3 to 10 minutes to complete the growth process of gold nanoislands. After the reaction, the glass slide was thoroughly rinsed with deionized water and dried at room temperature to obtain a gold nanoisland array substrate with surface enhanced properties.

4. The method for preparing a functionalized SERS chip according to claim 3, wherein: The pre-prepared gold nanoisland array substrate was immersed in phosphate buffer containing 25 μM CP05 peptide and 50 μM thiol polyethylene glycol (SH-PEG) and incubated at 4°C for 4 hours. It was then removed and rinsed three times with PBS buffer to remove unbound molecules to obtain a functionalized SERS chip.

5. A CP05 polypeptide-functionalized SERS chip, characterized by being prepared by the method of claim 4; the CP05 polypeptide has a sequence of CRHSQMTVTSRL, forms an Au-S bond with the gold nanoisland array through the terminal thiol group, and has a molar ratio of 1:2 to SH-PEG.

6. A SERS-based exosome capture and label-free detection platform, wherein the CP05 polypeptide functionalized SERS chip according to claim 5 is mounted on the platform; characterized in that: Includes the following modules: Dropping module: The sample is accurately dispensed to the designated position of the reagent kit through a high-precision micropipette pump, supporting multi-channel synchronous operation and multi-well layout, and automatically switches to cleaning mode after the drop is completed; Oscillation module: uses an adjustable frequency vortex oscillator to ensure uniform distribution of the sample by setting the oscillation intensity and time parameters, and has a splash-proof design to prevent liquid overflow; Cleaning and aspiration module: Driven by a multi-axis robotic arm, it integrates a controllable flow nozzle and a high-efficiency vacuum aspiration device, and supports a default two-round PBS automatic cleaning program with adjustable times; Sample delivery and positioning module: A high-precision conveyor device is used to accurately move the test kit under the Raman microscope head and seamlessly connect it to its motorized sample platform. Optical or laser sensors are used to achieve real-time position feedback and coordinate matching. Control module: Based on PLC or industrial computer architecture, supports Modbus and TCP / IP communication protocols, and is deeply linked to the Raman microscopy detection system; Analysis module: CNN-MLP integrated algorithm is used to analyze the SERS fingerprint of exosomes.

7. The SERS-based exosome capture and label-free detection platform according to claim 6, characterized by: The analysis module adopts the CNN-MLP integrated algorithm and includes the following contents: (1) Dynamic baseline correction and regional intensity normalization preprocessing; (2) Convolutional neural network: 3 layers of convolution, filter size 16 / 32 / 64; (3) Multilayer Perceptron: 2 hidden layers, 64 neurons in each layer.

8. A method for automatic capture and label-free detection of exosomes, the method being based on the SERS-based exosome capture and label-free detection platform of claim 7, characterized by: The sample is precisely added to the designated position of the reagent kit through a high-precision micropipette pump, supporting multi-channel synchronous operation; it is then shaken and mixed by an adjustable frequency vortex oscillator; during the cleaning stage, a multi-axis robotic arm with an integrated nozzle and vacuum aspiration device completes automatic PBS cleaning, supporting multiple cycles; after the sample processing is completed, the reagent kit is automatically positioned under the Raman microscope platform through a high-precision transmission system, and real-time position calibration is achieved through optical or laser sensors; during the Raman acquisition process, the system randomly selects 5 different positions in the reagent reaction area for scanning, with an acquisition range of 10μm×10μm, a step spacing of 1μm, and an excitation current that can be adjusted to 5-10 mA to ensure the representativeness of the spectral data and acquisition stability, thereby achieving efficient and accurate Raman detection.

9. A terminal device, characterized in that: The terminal device includes: a processor, a memory, a communication interface and a bus; the processor, the memory and the communication interface are connected via the bus and communicate with each other; the memory stores executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the analysis of SERS fingerprint information by the analysis module in the SERS-based exosome capture and label-free detection platform according to claim 7.

Images