TWI827170B - Surface enhanced raman spectroscopy chip and the detection system - Google Patents

Abstract

The present invention relates to a surface-enhanced Raman spectroscopy chip, comprising a substrate and a plurality of wafers disposed on the surface of the substrate to form a wafer array, each wafer has functional groups modified on silver nanoparticles, and each wafer is modified by different functional groups. The surface-enhanced Raman spectroscopy chip can be used as a system for disease prediction or analysis by using breath, including: a breath collection device for collecting sample breath and passing the sample breath through the surface-enhanced Raman spectroscopy chip, a spectrometer detection device for detecting the surface-enhanced Raman spectroscopy chip, obtaining a Raman scattering spectrum image of the sample breath, and storing the Raman scattering spectrum image in a storage device, and a computer inputs the Raman scattering spectrum image of the sample breath into a learning and classification system to obtain a disease detection result.

Description

Surface-enhanced Raman spectroscopy chip and its detection system

The invention relates to a wafer and its detection system, in particular to a surface-enhanced Raman spectroscopy wafer and its detection system.

Raman Spectroscopy is a type of molecular vibration (Vibration) spectroscopy. Its principle is to use a fixed wavelength laser light source to excite the sample. When the excitation light interacts with the sample molecules, energy exchange occurs after the photons collide with the molecules. The photons Part of the energy is transferred to or obtained from the sample molecules, thereby changing the frequency of light. This change is called Raman shift. The amount of Raman shift will not change due to the laser wavelength, so Can be used to reflect molecular bonding and structure. However, the signal of Raman spectroscopy is very weak and susceptible to fluorescence interference, and its application range is subject to many limitations. Therefore, the emergence of surface enhanced Raman spectroscopy (SERS) has made the application of Raman spectroscopy technology more and more popular. The more extensive.

Surface-enhanced Raman spectroscopy is a phenomenon in which molecules adsorbed on rough metal surfaces resonate with the metal surface to enhance Raman scattering. Compared with Raman spectroscopy technology, surface-enhanced Raman spectroscopy has the characteristics of micro-measurement and surface specificity. It has been widely used in biomedicine in recent years. It is a non-invasive detection technology with high specificity and high sensitivity. , fast and other advantages, it can detect the structure of biomolecules such as proteins, nucleic acids, lipids and sugars in tissue cells at the molecular level, and then by comparing the Raman spectra between cancer tissues and normal tissues, tissue lesions can be reflected Characteristic spectrum and use it in early diagnosis of cancer.

Most of today's medical technologies focus on the detection of a single disease. However, diseases such as cancer are caused by multiple diseases. If only a single disease target is detected, patients need to undergo multiple examinations in a time-consuming manner, and the cause of the cancer cannot be accurately determined. Therefore, if a system that can detect multiple targets simultaneously can be developed, the accuracy of cancer diagnosis can be improved and the time spent on conducting multiple examinations can be reduced.

All publications and patent applications cited in this specification are hereby incorporated by reference, and each individual publication or patent application is expressly and individually indicated to be incorporated by reference for any and all purposes. In the event of any inconsistency between this document and any publication or patent application incorporated by reference, this document shall control.

The terms "includes," "has," and "includes" are used herein in an open, non-limiting sense. The terms "a" and "the" shall be understood to cover both the plural and the singular. The term "one or more" means "at least one" and thus may include a single feature or a mixture/combination of features.

This summary is intended to provide a simplified summary of the invention to provide the reader with a basic understanding of the invention. This summary is not an extensive overview of the invention and it is not intended to identify key or critical elements of the embodiments of the invention or to delineate the scope of the invention.

The advantages of surface-enhanced Raman spectroscopy are that it is sensitive, fast, and highly specific. It is suitable for use in medical detection. Surface-enhanced Raman spectroscopy is a non-invasive technology that does not damage the sample and can detect any state (solid, Liquid, gaseous) samples have good application prospects in diagnosing diseases.

In view of this, in order to solve the problem of being unable to detect multiple diseases at the same time, the present invention provides a surface-enhanced Raman spectroscopy chip.

The invention provides a surface-enhanced Raman spectroscopy chip, which includes: a substrate with nanometal particles distributed on the surface; and n chips (n≥1) are arranged on the surface of the substrate to form a chip array, each of which is disposed on the surface of the substrate. A wafer has functional groups modified on the nanometal particles. The functional groups modified by each wafer are different, and the functional groups are at least one or more of the following: 4-Mercaptopyridine, 4-aminobenzene 4-Aminothiophenol, 4-Mercaptophenylboronic acid, 4-bromothiophenol, 4-Mercaptobenzoic acid, 2-thiopa 2-Thiobarbituric acid, 2-Thiouracil, 4-Thiouracil, 4-Methylpyridine, 4-aminophenyl disulfide 4-Aminophenyl disulfide, 4-(Methylthio)-benzaldehyde, 2-Mercapto-4-methyl-pyrimidine hydrochloride pyrimidine hydrochloride), 2-Mercapto-1-methylimidazole, 2-Mercapto-thiazoline, 3-amino-5-mercapto-1,2,4- Triazole (3-Amino-5-mercapto-1,2,4-triazole), 3-mercapto-1,2,4-triazole (3-Mercapto-1,2,4-triazole), 5-mercapto- 1-Methyltetrazole (5-Mercapto-1-methyltetrazole), 2-Mercaptoimidazole (2-Mercaptoimidazole), 5,5'-Dithiobis(2-nitrobenzoic acid) (5,5'-Dithiobis -(2-nitrobenzoic acid, DTNB), 2-Mercapto-4-methyl-5-thiazoleacetic acid, 4,6-dihydroxy-2-mercaptopyrimidine (4,6-Dihydroxy-2-mercaptopyrimidine), 2-Mercapto-5-methylbenzimidazole, 1-Naphthalenthiol, 2-methoxybenzene 2-Methoxythiophenol, 2-Aminothiophenel, 2-Mercaptobenzothiazole, 2-Mercaptobenzoxazole, 2-mercapto-5 -Nitrobenzimidazole (2-Mercapto-5-nitro-benzimidazole), 5-Amino-1,3,4-thiadiazole-2-thiol (5-Amino-1,3,4-thiadiazole-2 -thiol), 4-Amino-6-hydroxy-2-mercaptopyrimidine monohydrate (4-Amino-6-hydroxy-2-mercaptopyrimidine monohydrate), 2-Mercaptoimidazole (2-Mercaptoimidazole), 2-mercaptopyrimidine (2 -Mercaptopyrimidine), 3-Methoxythiophenol, 4-Hydroxy-2-mercapto-6-methylpyrimidine or 2-mercapto-4 (3H)-Quazolinone (2-Mercapto-4(3H)-quinazolinone.

According to an embodiment of the present invention, the nanometal particles are silver nanoparticles.

According to an embodiment of the present invention, the shape of the silver nanoparticles is cube.

According to an embodiment of the present invention, the substrate is a silicon substrate, stainless steel, polyvinylidene fluoride, glass or plastic sheet.

The present invention also provides a system for disease prediction or analysis using gas, which includes: a gas collection device for collecting sample gas and passing the sample gas through the surface-enhanced Raman spectroscopy chip; a spectrometer detection device for detecting The surface-enhanced Raman spectroscopy chip performs detection, obtains a Raman scattering spectrum image of the sample gas, and stores the Raman scattering spectrum image in a temporary or non-transitory storage device; and a computer, A pre-trained learning classification system is loaded and executed through at least one storage unit, and the computer obtains a disease detection result by inputting the Raman scattering spectrum image of the sample gas into the neural network system.

According to an embodiment of the present invention, the neural network system uses the negative exhaled gas or positive exhaled gas of the specified disease to train the weight value of each layer using the training Raman scattering spectrum image reflected by the surface-enhanced Raman spectroscopy chip. .

According to an embodiment of the present invention, the computer includes: a storage database for storing the training Raman scattering spectrum image reflected by the negative exhaled gas or the positive exhaled gas on the surface-enhanced Raman spectroscopy chip.

According to an embodiment of the present invention, the learning classification system is a neural network-like system and is trained according to the following method: the gas collection device receives the negative exhaled gas or the positive exhaled gas; the spectrometer detection device is used to detect the negative exhaled gas or the positive exhaled gas; The surface-enhanced Raman spectroscopy chip is used for detection, and the training Raman scattering spectrum image of the negative exhaled gas or the positive exhaled gas is obtained, and the training Raman scattering spectrum image and the corresponding negative information or positive information are Stored in the temporary or non-transitory storage device; the computer obtains the training Raman scattering spectrum image and the corresponding negative information or positive information through the temporary or non-transitory storage device, and converts the training The Raman scattering spectrum image is input to the neural network system to obtain an output result, and the weight value of each layer of the neural network system is modified through back propagation of the error of the output result and the expected result.

According to an embodiment of the invention, the designated disease is cancer, hormonal imbalance, bacterial infection or viral infection.

According to an embodiment of the present invention, the designated disease is pneumonia.

The surface-enhanced Raman spectroscopy chip of the present invention has the following advantages compared with the prior art.

(1) The surface-enhanced Raman spectroscopy chip of the present invention contains multiple chips on the substrate, and each chip has different functional groups synthesized, so it can detect multiple analytes at the same time and improve detection efficiency.

(2) The surface-enhanced Raman spectroscopy chip can also be used in systems that use gases for disease prediction or analysis. The collected sample gas is analyzed through the chip to obtain a Raman scattering spectrum image, and then the computer is used to analyze the sample gas. The image is input to the learning classification system for comparison to obtain a disease detection result. Users only need to blow air into the system, the operation is simple and convenient, and the test results can be obtained without complicated procedures.

(3) The learning classification system in the computer contains a database of negative breath and positive breath for multiple diseases for comparison, such as cancer, hormonal imbalance, bacterial infection, viral infection or pneumonia, and can perform multiple tests at the same time, It reduces the time required to perform various tests and can also provide information on the development of the disease and improve the accuracy of diagnosis.

In accordance with common practice, the various features and components in the figures are not drawn to actual scale, but are drawn in a manner intended to best present the specific features and components relevant to the present invention. In addition, the same or similar element symbols are used to refer to similar elements and components between different drawings.

In order to make the description of the present invention more detailed and complete, the following provides an illustrative description of the implementation modes and specific embodiments of the present invention, but this is not the only form of implementing or using the specific embodiments of the present invention. In this specification and the appended patent claims, "a" and "the" may also be interpreted as plural unless the context indicates otherwise. In addition, in the scope of this specification and the appended patent application, unless otherwise specified, "disposed on something" can be regarded as directly or indirectly contacting the surface of something through attachment or other forms. The definition of the surface Judgment should be based on the meaning of the content before and after the description/paragraphs and the common knowledge in the field to which this description belongs.

The following content, combined with the drawings, presents more specific implementation examples or comparative examples of the present invention. The purpose is only to illustrate the content of the present invention more clearly, and is not intended to limit the scope of the present invention.

Please refer to Figure 1 and Figure 2 together to provide a schematic diagram 100 of a surface-enhanced Raman spectroscopy chip according to the present invention. The chip includes: a substrate 110 with nanometal particles 111 distributed on the surface; and a plurality of chips. A chip array 120 is formed on the surface of the substrate.

In the surface chip array of the substrate, each chip has a functional group modified on the nanometal particles. The functional groups modified on each chip are different and can be used to analyze different compounds. The functional group may be at least one or more of the following: 4-Mercaptopyridine, 4-Aminothiophenol, 4-Mercaptophenylboronic acid, 4-bromo 4-bromothiophenol, 4-Mercaptobenzoic acid, 2-Thiobarbituric acid, 2-Thiouracil, 4-thiouracil Uracil (4-Thouracil), 4-Methylpyridine (4-Methylpyridine), 4-Aminophenyl disulfide (4-Aminophenyl disulfide), 4-(Methylthio)-benzaldehyde (4-(Methylthio) -benzaldehyde), 2-Mercapto-4-methyl-pyrimidine hydrochloride, 2-Mercapto-1-methylimidazole, 2 -Mercapto-thiazoline, 3-Amino-5-mercapto-1,2,4-triazole, 3-mercapto- 1,2,4-triazole (3-Mercapto-1,2,4-triazole), 5-Mercapto-1-methyltetrazole (5-Mercapto-1-methyltetrazole), 2-Mercaptoimidazole ), 5,5'-Dithiobis-(2-nitrobenzoic acid, DTNB), 2-mercapto-4-methyl-5-thiazoleacetic acid (2-Mercapto-4-methyl-5-thiazoleacetic acid), 4,6-Dihydroxy-2-mercaptopyrimidine (4,6-Dihydroxy-2-mercaptopyrimidine), 2-mercapto-5-methylbenzimidazole ( 2-Mercapto-5-methylbenzimidazole), 1-Naphthalenthiol, 2-Methoxythiophenol, 2-Aminothiophenel, 2-mercaptobenzene 2-Mercaptobenzothiazole, 2-Mercaptobenzoxazole, 2-Mercapto-5-nitro-benzimidazole, 5-amino-1, 3,4-thiadiazole-2-thiol (5-Amino-1,3,4-thiadiazole-2-thiol), 4-amino-6-hydroxy-2-mercaptopyrimidine monohydrate (4-Amino- 6-hydroxy-2-mercaptopyrimidine monohydrate), 2-Mercaptoimidazole, 2-Mercaptopyrimidine, 3-Methoxythiophenol, 4-hydroxy-2- 4-Hydroxy-2-mercapto-6-methylpyrimidine or 2-Mercapto-4(3H)-quinazolinone, these functional groups The S contained therein can be used to bond with nanometal particles.

In a preferred embodiment, the metal nanoparticle 111 can be a gold nanoparticle, a silver nanoparticle, an aluminum nanoparticle, a titanium nanoparticle or a copper nanoparticle. Preferably, the nanometal particles 111 are silver nanoparticles. In a preferred embodiment, the diameter of the nanometal particles 111 is 10-250nm, for example: 15nm, 35nm, 55nm, 85nm, 105nm, 135nm, 150nm, 200nm, or 250nm. The nanometal particles 111 are distributed on the substrate 110 because when the molecules in the object to be measured are adsorbed on the nanometal particles 111, the electromagnetic effect (excitation of the polarity of the surface plasma) and the chemical effect (adsorption on the nanoparticles) The molecules of nano-metal particles are in contact with the nano-metal structure, and the charge transfer therebetween (also known as the first layer effect) results in the enhancement of the Raman spectrum scattering signal, overcoming the low sensitivity of traditional Raman detection. problem.

In a preferred embodiment, the shape of the silver nanoparticles 111 can be a cube, a sphere, an ellipsoid, a cuboid, a triangle, or a decahedron. Preferably, the shape is a cube, as shown in Figure 2. The method of making the substrates with nanometal particles 111 on the surface can be by conventional methods. These methods can be divided into physical and chemical preparation methods; physical preparation methods such as but not limited to vacuum evaporation (in In a vacuum environment, laser or electron beam impact or heating is used to melt the plated metal, evaporate the atoms, and then collide with the surface of the substrate to adhere to it), electrochemical deposition method (using a three-electrode electrochemical system to apply potential The metal to be plated will produce successive redox reactions on the working electrode). The chemical preparation method includes but is not limited to the sol-gel method (using inorganic metal salts or pure metal oxides as precursors, through hydrolysis, condensation, Cross-linked colloid is formed after polymerization), chemical replacement deposition method (i.e., electrodeless replacement method, using the redox potential difference between different metals to perform its own redox reaction), etc. In one embodiment of the present invention, the manufacturing method of the substrate with metallic silver particles is as follows: the substrate (glass or plastic sheet) is ultrasonically cleaned in ethanol and acetone respectively, dried and then placed in H ₂ SO ₄ / H ₂ O ₂ =3:1 (v/v), heat and clean; then clean the substrate with ultrapure water, blow it dry, and place it in a toluene solution of (3-mercaptopropyl)trimethoxysilane; Then clean the substrate with toluene, acetone and ethanol respectively and set aside for use. Dissolve silver nitrate in deionized water, slowly add NaOH (10%) to the silver nitrate solution, and then add ammonia solution drop by drop in the beaker until the precipitate disappears; then place the solution in an ice water bath to cool. Add glutaraldehyde solution to the cooled solution, react for 10 seconds, place it in a 90°C water bath, and place the substrate in the solution to react, so that silver nanoparticles grow on the substrate to form metallic silver particles. The substrate; then immediately take out the substrate and place it in pre-cooled ethylene glycol to stop growth; finally, perform ultrasonic waves for 20 seconds to remove non-specifically attached silver nanoparticles.

An exemplary manufacturing method for modifying metal particles containing different functional groups is as follows: soak the metal particle substrate in an ethanol solution containing different functional groups, filter and separate, rinse with ethanol to remove unreacted molecules, and then rinse in water Under a steam atmosphere, the solvent is slowly evaporated to obtain a wafer modified with a functional group.

It is known that different compounds can be used to modify the nanometal particles 111 . These characteristic functional groups can form strong chemical bonds with the nanometal particles 111 , making it easy for the functional groups to be adsorbed on the surface of the nanometal particles 111 . Since the intermolecular Van der Waals force of the functional groups can make them orderly arranged on the surface of the nanometal particles 111, the molecules can be hydrophilic and hydrophobic, so that the chemical substances contained in the substance to be tested can be enriched on the surface of the nanometal particles 111. , and the nanometal particles adsorbed by functional groups can also form a hot spot effect (a large electromagnetic field effect is generated between the gaps or connections between metal nanoparticles), which can generate relatively large energy and achieve higher sensitivity. SERS detection. It is known that the compounds that can be used to modify the nanometal particles 111 include phenyl ring thiol compounds or heterocyclic thiol compounds with N atoms, and there are various types; however, in the prior art, no compound like the inventor of this case has been found. After extensive screening, the selected specific types of functional groups can be used to distinguish different diseases represented by the samples, the probability of positive and/or negative disease, the severity of the disease, and the depth of the disease (degree of infiltration). or severity), etc.

In a preferred embodiment, the material of the substrate can be a silicon substrate, stainless steel, polyvinylidene fluoride, glass or plastic sheet. And the thickness of the substrate is preferably within 1 mm, such as: 0.1mm, 0.2mm, 0.3mm, 0.4mm, 0.5mm, 0.6mm, 0.7mm, 0.8mm, 0.9mm or 1mm.

Surface-enhanced Raman spectroscopy chip samples suitable for use in the present invention include but are not limited to gases, biological fluids, extracellular vesicles, and cell debris. In terms of convenience in sample acquisition, it is better to collect the sample in the form of gas. Because the breath exhaled by an individual contains more than 3,000 volatile organic compounds (VOCs), and patients with different diseases are significantly different from healthy people, and even the disease severity or severity of different diseases are significantly different. Depending on the depth of the disease, the composition of organic matter in the breath of individuals is also different; the chip array of the present invention has been used to test the chip modified with functional groups, which can be used to distinguish the composition of organic matter in the breath of individuals, so it can be used By detecting the VOCs content of the breath exhaled by the user, the probability of positive and/or negative disease, the severity level of the disease, the depth of the disease (degree of infiltration or severity), etc. can be obtained, thereby enabling identification in a short time. There are subjects who need additional confirmation of diagnosis.

The term "individual" as used herein refers to mammals and non-mammals that are at risk of possible disease. The aforementioned mammals include, but are not limited to: humans, non-human primates, sheep, dogs, murine rodents (such as mice, rats, etc.), guinea pigs, cats, rabbits, cows, and horses; the aforementioned non-mammals Including but not limited to: birds, amphibians and reptiles. Preferably, the individual is a human being.

Please refer to Figures 3 to 5 together. The present invention also provides a system 200 for disease prediction or analysis using gas. The system 200 includes: a gas collection device 210 (the gas collection device 210 is equipped with a surface-enhanced Raman Spectrum chip (not shown in the figure), a spectrometer detection device 230, and a computer 250. The user H blows into the gas collection device 210. After the gas collection device 210 collects the exhaled gas, it is collected in the system 200 Perform analysis.

The method performed by the gas disease prediction or analysis system 200 is as follows: 1. The gas collection device 210 collects the gas exhaled by the user H (S310), and causes the sample gas to pass through a surface-enhanced Raman spectroscopy chip 100 provided in the gas collection device 210 (S311); 2. Detect the surface-enhanced Raman spectrum chip 100 by the spectrometer detection device 230 to obtain the Raman scattering spectrum image of the sample gas (S312), and store the Raman scattering spectrum image in a Temporary or non-transitory storage device (not shown in the figure); 3. The computer 250 loads at least one storage unit of the Raman scattering spectrum image stored in the storage device and executes a pre-trained learning classification system by inputting the Raman scattering spectrum image of the sample gas. To the learning classification system (S313), after comparing the weight values of each layer with the first pre-trained model (S314), a negative or positive result of a disease can be obtained (S315), or the severity level of the disease and the severity of the disease can be obtained. Depth (degree or severity of infiltration).

The spectrometer detection device 230 described in this article detects the surface-enhanced Raman spectroscopy chip 100, which includes when the gas passes through the surface-enhanced Raman spectroscopy chip 100, the spectrometer detection device 230 emits a laser light. The laser light Projected on the detection object, when the laser light interacts with the target object of the detection object, energy exchange occurs after the photon collides with the target object molecules, changes the frequency of the light, and generates scattered light; the spectrometer detection device receives the scattered light , which is the Raman scattering spectrum image of the sample gas. In addition, in another embodiment, the data of the Raman scattering spectrum image is calculated using the PCA method in the SPSS software analysis tool. The steps are to calculate the skew variance matrix of each wavelength data, and then calculate the true value of the matrix. value and the original vector, use T test analysis to obtain the PCA score with the most significant difference, and finally establish a corresponding analysis scatter plot based on the selected principal components, and then obtain a surface-enhanced Raman spectroscopy analysis chart.

The storage devices described in this article are temporary or non-transitory storage devices, such as but not limited to well-known tape systems such as tapes or cassettes, including floppy disks, hard disks and other magnetic disks, compact optical disks-ROM ( Read Only Memory)/MO (Magneto Optical, magneto-optical disk)/MD (Mini Disc, small disk)/digital audio and video disc/compact disc-R and other optical disc systems, IC cards, memory cards, optical discs card system or mask ROM/EPROM (Erasable Programmable Read Only Memory, erasable programmable read-only memory)/EEPROM (Electrically Erasable Programmable Read Only Memory, electronically erasable programmable read-only memory) /Flash ROM and other semiconductor memory systems, etc. The Raman scattering spectrum image results stored in the storage device's temporary or non-transitory storage device can be transmitted to the computer 250 using wired or wireless transmission; wired transmission such as USB transmission, wireless transmission such as Bluetooth transmission, but in this article are not limited to this method.

Here, the computer 250 includes an input part 251, a weight allocation part 252, a learning classification system 253, an output part 254 and a storage database 255. The input part 251 is used to input the subject's relevant information (such as but not limited to name, gender or age) and the subject's Raman scattering spectrum image; the weight allocation part 252 is used to input the subject's information in the above input part. The input subject's Raman scattering spectrum image is assigned a weight of spectral intensity, and the spectrum after weight distribution can be further standardized and regularized; the learning classification system 253 inputs the spectral data after weight distribution in sequence. The input layer in the system repeatedly processes the weighted coefficients of the learning classification, and uses its learning to complete the model to infer the subject's disease classification, the severity of the disease, and the depth of the disease (degree of infiltration). or severity); an output unit 255 outputs the data of the disease inferred by the learning classification system 2536. The output unit 255 may also output at least one of the inferred disease name, disease severity level or disease depth of the subject. are displayed together on a display (not shown in the figure); and a storage database 256 stores a learning completion model learned through the above-mentioned learning classification system 253.

In a preferred embodiment, the computer further includes a storage database for storing the training Raman scattering spectrum pattern reflected by the negative exhaled gas or the positive exhaled gas on the surface-enhanced Raman spectroscopy chip. The image is used as a control group to compare the user's exhaled breath.

The learning classification system referred to in this article covers, for example, but is not limited to: booster tree classification (booster) system, gradient classifier tree classification (gradient classifier) system, strong gradient boosting machine classification system, weak gradient boosting machine classification system, regression tree classification System, random forest (random forest) classification system, decision tree classification system, neural network-like classification system, deep neural network classification system (DNN), recursive neural network (RNN) classification system, long short-term memory model classification (LSTM) ) system, multi-layer perceptual classification (MLP) system, ensemble learning classification (ensemble learning) system, weak learning classification system, strong learning classification system, support vector machines (support vector machines) classification system, supervised learning classification system or semi-supervised Learn classification systems, etc.

In a preferred implementation form, the learning classification system is a neural network-like classification system that performs deep learning. Deep learning uses neural networks composed of overlapping complex layers to learn high-order feature values based on input data. In addition, the learning completion model is also a kind of neural network, that is, a deep learning model. In an implementation form, it has a complex array of convolution (Convolution) layer and pooling (Pooling) layer, and is located in the subsequent section. The fully combined layer and Softmax layer. The complex group of stacking layers and pooling layers is to repeatedly perform stacking and pooling when inputting the Raman scattering spectrum image of the subject, and extract the feature quantity of the input image. The fully combined layer system combines the extracted feature quantities at one node. The softmax layer outputs at least one of the probability of each disease, the disease severity level, or the disease depth based on the output from the fully combined layer.

This learning completion model uses the data of a large number of subjects to learn weighted coefficient distribution in advance. Specifically, the parameters of the filter used in the stacked layer and the weighted parameters of the nodes in the fully combined layer are learned. Please refer to Figure 6 as well. In a preferred embodiment, the learning classification system is a neural network-like system, and is generated by using the negative exhaled air or positive exhaled air of the specified disease on the surface-enhanced Raman spectroscopy chip. The reflected training uses Raman scattering spectrum images to train the weight values of each layer, and is trained according to the following method: 1. The gas collection device receives the negative exhaled gas or the positive exhaled gas of the plurality of diseases (S410), and passes it through the surface-enhanced Raman spectroscopy chip (S411); 2. Use the spectrometer detection device to detect the surface-enhanced Raman spectroscopy chip, obtain the training Raman scattering spectrum image of the negative exhaled gas or the positive exhaled gas (S412), and use the training Raman The scattering spectrum image and the corresponding negative information or positive information are stored in the temporary or non-transitory storage device; 3. The computer obtains the training Raman scattering spectrum image and the corresponding negative information or positive information through the temporary or non-transitory storage device, and inputs the training Raman scattering spectrum image to the type of neural network The network system obtains an output result (S413), modifies the weight value of each layer of the neural network system through back propagation of the error of the output result and the expected result (S414), and integrates the modified weight values of each layer. The first pre-trained model is produced (S415).

When multiple pre-training models are completed, they are integrated into a learning model and stored in a database. The learning model in this embodiment is assumed to be used as a part (program module) of artificial intelligence software. The learning completion model of this embodiment is used in a computer equipped with a CPU and a memory. Specifically, the CPU of the computer operates in the following manner: In accordance with the instructions from the learning completion model memorized in the memory, the input to the input layer (the initial stacked layer) of the neural network is The data (Raman scattering spectrum image) are stacked and pooled to obtain the weighted feature quantity of the image, and the obtained feature quantity is calculated according to the weighted coefficient of the learning completed model, and the output layer (Softmax layer) ) Output results (having at least one of probability values of various diseases, disease severity levels or disease depth); in addition, although the probability value is output at the node of each disease or healthy person, when learning, "1" is applied to the node corresponding to the teaching data (ie, the disease of the subject), and "0" is applied to other nodes.

Although the configuration of the computer 250 in this embodiment has been described above, it may be a computer equipped with a CPU, RAM, ROM, hard disk, display, keyboard, mouse, communication interface, etc. In addition, a program having a module that implements each of the above functions is prestored in RAM or ROM, and the program is executed by the CPU. Such program products are also included in the scope of the present invention.

In a preferred embodiment, the designated disease is cancer (including but not limited to cancer characterized by solid tumors (e.g., pneumonia, prostate cancer, kidney cancer, liver cancer, pancreatic cancer, gastric cancer, breast cancer, lung cancer, head and neck cancer) carcinoma, thyroid cancer, glioblastoma, Kaposi's sarcoma, Castleman's disease, uterine smooth muscle malignant sarcoma, melanoma, etc.), hematological cancers (e.g., lymphoma, leukemia, e.g. acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), or multiple myeloma) and skin cancers, such as cutaneous T-cell lymphoma (CTCL) and cutaneous B-cell lymphoma. Exemplary CTCLs include Sessa Sezary syndrome and mycosis fungoides), hormonal disorders (including primary or secondary), bacterial infections (including but not limited to Klebsiella pneumoniae, Escherichia coli, cloaca) Bacterial infections such as Enterobacter Cloacae, Pseudomonas aeruginosa, Burkholderia cepacia and Acinetobacter baumannii), viral infections (including but not limited to Picornaviridae) (Picornaviridae), Orthomyxoviridae (Orthomyxoviridae), Flaviviridae (Flaviviridae), Coronaviridae (Coronaviridae), and combinations thereof; in some embodiments, the virus is enterovirus, influenza Influenza virus, flavivirus, coronavirus, and combinations thereof), or pneumonia caused by bacterial or viral infection.

The following is a description of one of the embodiments of the present invention; first, collect the gas from multiple pneumonia patients infected with the new coronavirus (positive individuals) and multiple uninfected normal individuals (negative individuals); pass the gas through The surface-enhanced Raman spectroscopy chip of the present invention uses a spectrometer detection device to detect the surface-enhanced Raman spectroscopy chip to obtain the training Raman scattering spectrum image of the negative exhaled gas or the positive exhaled gas. , the training Raman scattering spectrum image is input to the learning classification system to obtain an output result, and the error of the output result and the expected result is back-propagated to modify the weight value of each layer of the neural network system, and then stored In the storage database; when there is an individual who is unknown to be infected with the new coronavirus, the gas of the individual to be tested can be collected, the system of the present invention is used to obtain the Raman scattering spectrum image of the gas, and the image is input to the The learning classification system is used to obtain an output result of whether an individual is infected or not infected after calculation. The above is just an illustrative example. The disease of the individual can be collected according to the detection requirements. Positive individuals or negative individuals of different diseases (such as but not limited to normal individuals and diabetic patients, normal individuals and breast cancer patients, normal individuals and colorectal cancer patients, etc. ), even disease-positive individuals with different levels of disease (such as stage 0 cancer, stage 1 cancer, stage 2 cancer, etc.) can obtain training results of different stages, and thus can be used for subsequent prediction or analysis of different diseases.

To sum up, the surface-enhanced Raman spectroscopy chip of the present invention contains multiple chips on the substrate, and the functional groups synthesized by each chip are different, so it can detect multiple analytes at the same time and improve the detection efficiency; and in one In a better embodiment, the surface-enhanced Raman spectroscopy chip can also be applied to a system that uses gas for disease prediction or analysis. The user only needs to blow air into the system, and the operation is simple and convenient, without complicated processes. Test results are available.

The present invention has been described in detail above. However, the above descriptions are only preferred embodiments of the present invention. They should not be used to limit the scope of the present invention, that is, any equivalent changes made in accordance with the scope of the patent application of the present invention. and modifications should still fall within the scope of the patent of the present invention.

100:Surface enhanced Raman spectroscopy wafer 110:Substrate 120:Chip 111:Nano metal particles H: User 200:System 210:Gas collection device 230: Spectrometer detection device 250:Computer 251:Input part 252: Weight allocation department 253:Learning Classification Systems 254:Output Department 255:Storage database S310~S315: steps S410~S415: steps

Implementations of the present technology will now be described by way of example only with reference to the accompanying drawings, in which:

Figure 1 is a schematic diagram of a surface-enhanced Raman chip according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a silver nanoparticle substrate of a surface-enhanced Raman chip according to an embodiment of the present invention.

FIG. 3 is a schematic diagram of the detection method of the surface-enhanced Raman chip analysis system according to an embodiment of the present invention.

Figure 4 is a detection process of a surface-enhanced Raman chip analysis system according to an embodiment of the present invention.

Figure 5 is a schematic diagram of a system for disease prediction or analysis using gas according to an embodiment of the present invention.

Figure 6 shows the training process of a neural network system such as a surface-enhanced Raman chip analysis system according to an embodiment of the present invention.

It should be understood that aspects of the present invention are not limited to the arrangements, means and characteristics shown in the drawings.

without

100:Surface enhanced Raman spectroscopy wafer

110:Substrate

120:Chip

Claims (10)

A surface-enhanced Raman spectroscopy wafer containing: a substrate with nanometal particles distributed on its surface; and n wafers (n≥1) are arranged on the surface of the substrate to form a wafer array. Each wafer has a functional group modified on the nanometal particle. The functional group modified by each wafer is different, and the functional group is At least one or more of the following: 4-Mercaptopyridine, 4-Aminothiophenol, 4-Mercaptophenylboronic acid, 4-bromothiophenol -bromothiophenol), 4-Mercaptobenzoic acid, 2-Thiobarbituric acid, 2-Thiouracil, 4-Thiouracil Thiouracil), 4-Methylpyridine, 4-Aminophenyl disulfide, 4-(Methylthio)-benzaldehyde, 2 -Mercapto-4-methyl-pyrimidine hydrochloride, 2-Mercapto-1-methylimidazole, 2-mercaptothiazoline ( 2-Mercapto-thiazoline), 3-Amino-5-mercapto-1,2,4-triazole (3-Amino-5-mercapto-1,2,4-triazole), 3-mercapto-1,2,4 -Triazole (3-Mercapto-1,2,4-triazole), 5-Mercapto-1-methyltetrazole (5-Mercapto-1-methyltetrazole), 2-Mercaptoimidazole (2-Mercaptoimidazole), 5,5 '-Dithiobis-(2-nitrobenzoic acid) (5,5'-Dithiobis-(2-nitrobenzoic acid), DTNB), 2-mercapto-4-methyl-5-thiazoleacetic acid (2-Mercapto- 4-methyl-5-thiazoleacetic acid), 4,6-Dihydroxy-2-mercaptopyrimidine (4,6-Dihydroxy-2-mercaptopyrimidine), 2-mercapto-5-methylbenzimidazole (2-Mercapto-5 -methylbenzimidazole), 1-Naphthalenthiol, 2-Methoxythiophenol, 2-Aminothiophenel, 2-mercaptobenzothiazole Mercaptobenzothiazole), 2-mercaptobenzoxazole (2-Mercaptobenzoxazole), 2-mercapto-5-nitro-benzimidazole (2-Mercapto-5-nitro-benzimidazole), 5-amino-1,3,4-thi Diazole-2-thiol (5-Amino-1,3,4-thiadiazole-2-thiol), 4-Amino-6-hydroxy-2-mercaptopyrimidine monohydrate (4-Amino-6-hydroxy-2 -mercaptopyrimidine monohydrate), 2-Mercaptoimidazole, 2-Mercaptopyrimidine, 3-Methoxythiophenol, 4-hydroxy-2-mercapto-6-methyl 4-Hydroxy-2-mercapto-6-methylpyrimidine or 2-Mercapto-4(3H)-quinazolinone. The surface-enhanced Raman spectroscopy chip of claim 1, wherein the nanometal particles are silver nanoparticles. The surface-enhanced Raman spectroscopy chip of claim 2, wherein the shape of the silver nanoparticles is cubic. Such as the surface-enhanced Raman spectroscopy chip of claim 1, wherein the substrate is a silicon substrate, stainless steel, polyvinylidene fluoride, glass or plastic sheet. A system for disease prediction or analysis using gases, including: A gas collection device for collecting sample gas and passing the sample gas through the surface-enhanced Raman spectroscopy chip according to any one of claims 1 to 4; A spectrometer detection device used to detect the surface-enhanced Raman spectroscopy chip, obtain the Raman scattering spectrum image of the sample gas, and store the Raman scattering spectrum image in a temporary or non-temporary storage sexual storage device; and A computer loads and executes a pre-trained learning classification system through at least one storage unit, and the computer obtains a disease detection result by inputting the Raman scattering spectrum image of the sample gas into the learning classification system. The system of claim 5, wherein the learning classification system uses the negative exhaled gas or positive exhaled gas of the specified disease to train the weight value of each layer using the training Raman scattering spectrum image reflected by the surface-enhanced Raman spectroscopy chip. . For example, the system of claim 6, wherein the computer includes: a storage database for storing the training Raman scattering spectrum image reflected by the negative exhaled gas or the positive exhaled gas on the surface-enhanced Raman spectroscopy chip. . For example, the system of claim 7, wherein the learning classification system is a neural network system and is trained according to the following method: the gas collection device receives the negative exhaled gas or the positive exhaled gas; the spectrometer detection device is used to detect the negative exhaled gas or the positive exhaled gas. The surface-enhanced Raman spectroscopy chip is used for detection, and the training Raman scattering spectrum image of the negative exhaled gas or the positive exhaled gas is obtained, and the training Raman scattering spectrum image and the corresponding negative information or positive information are Stored in the temporary or non-transitory storage device; the computer obtains the training Raman scattering spectrum image and the corresponding negative information or positive information through the temporary or non-transitory storage device, and converts the training The Raman scattering spectrum image is input to the neural network system to obtain an output result, and the weight value of each layer of the neural network system is modified through back propagation of the error of the output result and the expected result. For example, the system of claim 8, wherein the specified disease is cancer, hormonal imbalance, bacterial infection or viral infection. For example, claim the system of item 8, wherein the designated disease is pneumonia.