Fluorescence immunochromatography test strip rapid reading and analyzing method based on deep learning and application of fluorescence immunochromatography test strip rapid reading and analyzing method

Abstract

The invention provides a fluorescence immunochromatography test strip rapid reading and analyzing method based on deep learning and application. The method comprises the following steps: acquiring fluorescence spectrum data of a to-be-detected sample on a fluorescence immunochromatography test strip; constructing a signal preprocessing model; constructing a signal feature extraction model; constructing a time sequence prediction model; constructing a detection result output model; the method comprises the following steps: inputting a collected reaction phase fluorescence intensity signal to a signal preprocessing model for time domain distinguishing labeling, performing signal feature vector extraction by a signal feature extraction model, performing reaction complete feature vector prediction by a time sequence prediction model, and finally obtaining a detection result of a to-be-detected object of a to-be-detected sample through a detection result output model. Through a deep learning method, a to-be-detected object can be predicted and detected according to a reaction signal without waiting for complete reaction on the fluorescence immunochromatography test strip, and the requirement for rapid detection is met.

Description

Technical Field

[0001] The present application relates to the field of fluorescent immunochromatography technology, and specifically to a method and application of rapid reading of fluorescent immunochromatography test strips based on deep learning. Background Art

[0002] Machine learning, especially deep learning, can handle complex nonlinear relationships. When there are complex and variable dependencies between input data, it can capture high-dimensional features in the data through learning. Machine learning has been widely used in multiple medical fields, especially in image analysis, disease prediction, genomics, and biological signal processing. In the diagnosis of cardiovascular diseases, deep learning is used to analyze electrocardiograms (ECGs), ultrasound images, and other related data to help doctors make faster and more accurate diagnoses. For example, models based on deep learning can automatically identify abnormal phenomena such as arrhythmias and myocardial infarctions in ECG signals, helping doctors to detect potential risks in advance and improve clinical response speed. In the field of in vitro diagnosis, machine learning can also help analyze data and enhance accuracy. The present invention innovatively uses a machine learning algorithm in the process of reading test strips to predict the growth trend of fluorescence intensity signals through time series data, thereby shortening the diagnosis time of myocardial infarction. Summary of the invention

[0003] The present application provides a method for rapid reading of fluorescent immunochromatographic test strips based on deep learning, which can solve the technical problem in the prior art that the fluorescent immunochromatographic test strips have a long sample addition reaction time, which is not conducive to rapid detection.

[0004] In a first aspect, the present application provides a method for rapid reading of fluorescent immunochromatographic test strips based on deep learning, comprising the following steps: Collect and obtain the fluorescence spectrum data of the sample to be tested on the fluorescent immunochromatographic test strip; Constructing a signal preprocessing model to distinguish and annotate the collected fluorescence spectrum data according to the reaction stages, and obtaining the annotated fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; Construct a signal feature extraction model to extract fluorescence intensity signal feature vectors from the annotated fluorescence spectrum data; Constructing a time series prediction model to perform time series prediction processing on the reaction feature vector to obtain the complete reaction feature vector; Constructing a test result output model for performing diagnosis based on the complete characteristic vector of the reaction and obtaining the diagnosis result of the object to be tested in the test sample; The collected fluorescence intensity signal of the reaction stage is input into the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model is used to extract the signal feature vector, and the time series prediction model is used to predict the complete feature vector of the reaction. Finally, the test result of the test object of the test sample is obtained through the test result output model.

[0005] In one embodiment of the present invention, the acquisition of the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip specifically includes the following steps: The sample to be tested is added to the fluorescent immunochromatographic test strip, and the fluorescence intensity signal at each position on the fluorescent immunochromatographic test strip is read at a preset frequency until the reaction is complete, and the fluorescence spectrum data is obtained.

[0006] In one embodiment of the present invention, the reaction not started signal refers to the fluorescence spectrum signal read before the sample to be tested migrates to the detection line of the fluorescent immunochromatographic test strip; the reaction complete signal refers to the signal read when the change rate of the fluorescence spectrum signal within a preset time period is lower than the change rate threshold; the reaction progress signal refers to the fluorescence spectrum signal read during the time between the reaction not started and the reaction complete.

[0007] In one embodiment of the present invention, the construction of the signal preprocessing model specifically includes the following steps: According to the reaction kinetic characteristics of the fluorescent immunochromatographic test paper, a reaction stage segmentation standard is established, wherein the reaction stage includes the reaction not started, the reaction is in progress and the reaction is completed; According to the established reaction stage segmentation standard, the acquired fluorescence spectrum data are distinguished and labeled for the reaction stages.

[0008] In one embodiment of the present invention, the construction of the signal feature extraction model specifically includes the following steps: Obtain multidimensional characteristic values ​​of fluorescence spectrum data of the reaction; The fluorescence spectrum data of the reaction is taken as input and the multi-dimensional feature values ​​are taken as output to train and obtain the signal feature extraction model.

[0009] In one embodiment of the present invention, the multidimensional characteristic values ​​include but are not limited to T line peak area, C line peak area, T line fluorescence intensity, C line fluorescence intensity, T / C line peak area ratio, T / C line fluorescence intensity ratio and background fluorescence intensity.

[0010] In one embodiment of the present invention, the construction of the time series prediction model specifically includes the following steps: Combining multiple fluorescence intensity signal features into a feature vector; Set the time window to obtain a fixed-length feature vector sequence; Slide the time window to capture the feature vector covering all stages of the reaction and obtain training samples; Input the training samples into the Transformer model to build a time series prediction model.

[0011] In one embodiment of the present invention, the construction of the detection result output model specifically includes the following steps: Setting the concentration threshold of the analyte corresponding to each diagnostic conclusion of the analyte in the blood sample; Obtaining the complete fluorescence intensity signal characteristic vector of the reaction of the analyte; According to the set diagnostic threshold of the concentration of the analyte, the complete characteristic vector of the reaction is detected and judged to obtain the detection result of the analyte in the sample to be tested.

[0012] In one embodiment of the present invention, the complete reaction feature vector is a complete reaction feature vector extracted by a signal feature extraction model or a complete reaction feature vector obtained by time series prediction using a time series prediction model.

[0013] In the second aspect, the present application provides an application of a deep learning-based rapid reading method of a fluorescent immunochromatographic test strip for detecting troponin I in the blood.

[0014] In a third aspect, the present application provides a deep learning-based rapid reading system for fluorescent immunochromatographic test strips, comprising: A fluorescence spectrum data acquisition module is used to collect and acquire the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip; A signal preprocessing model building module is communicatively connected with the fluorescence spectrum data acquisition module, and is used to build a signal preprocessing model, distinguish and annotate the collected fluorescence spectrum data according to the reaction stages, and acquire the annotated fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; A signal feature model building module is communicatively connected with the signal preprocessing model building module and is used to build a signal feature extraction model to extract a fluorescence intensity signal feature vector from the annotated fluorescence spectrum data; A time series prediction model building module is connected to the signal feature model building module for building a time series prediction model and performing time series prediction processing on the reaction feature vector to obtain a complete reaction feature vector; A result output model construction module, which is in communication connection with the time series prediction model construction module, is used to construct a detection result output model, perform diagnosis according to the complete reaction feature vector, and obtain the diagnosis result of the object to be tested in the sample to be tested; The detection result acquisition module is communicatively connected with the signal preprocessing model construction module, the signal feature model construction module, the timing prediction model construction module and the result output model construction module, and is used for inputting the collected fluorescence intensity signal of the reaction stage into the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model for signal feature vector extraction, and the timing prediction model for prediction of complete reaction feature vector, and finally obtaining the detection result of the object to be tested of the sample to be tested through the detection result output model.

[0015] The beneficial effects brought by the technical solution provided by the embodiment of the present application include at least: Through deep learning methods, there is no need to wait for the analyte to react completely on the fluorescent immunochromatographic test strip. The analyte can be predicted and tested based on the reaction signal, meeting the needs of rapid testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A method flow chart of a method for rapid reading of fluorescent immunochromatographic test strips based on deep learning provided in an embodiment of the present application; Figure 2 Another method flow chart of the method for rapid reading of fluorescent immunochromatographic test strips based on deep learning provided in an embodiment of the present application; Figure 3 A typical fluorescence signal spectrum of the fluorescence immunochromatography method provided in the embodiment of the present application; Figure 4 A schematic diagram of the network structure of the signal preprocessing model provided in the embodiment of the present application; Figure 5 A schematic diagram of the network structure of the signal feature extraction model provided in the embodiment of the present application; Figure 6 A schematic diagram of the network structure of the time series prediction model provided in the embodiment of the present application; Figure 7 A schematic diagram of the network structure of the detection result output model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0019] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0020] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0021] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0022] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0023] First, as Figure 1-Figure 2 As shown, the embodiment of the present application provides a method for rapid reading of fluorescent immunochromatographic test strips based on deep learning, which specifically includes the following steps: Step S1: collecting and obtaining fluorescence spectrum data of a sample to be tested on a fluorescent immunochromatographic test strip, wherein the sample to be tested is a liquid sample, such as blood; Step S2: constructing a signal preprocessing model to distinguish and annotate the collected fluorescence spectrum data according to the reaction stages, and obtain the annotated fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; Step S3: construct a signal feature extraction model for extracting fluorescence intensity signal feature vectors from the annotated fluorescence spectrum data, including extracting reaction progress feature vectors from the fluorescence spectrum data before the reaction starts and extracting reaction complete feature vectors from the fluorescence spectrum data after the reaction is complete; in this application, it is mainly used to extract reaction progress feature vectors from the fluorescence spectrum data after the reaction is complete; Step S4: constructing a time series prediction model for performing time series prediction processing on the reaction feature vector to obtain a complete reaction feature vector; Step S5: constructing a detection result output model for performing diagnosis according to the complete reaction feature vector and obtaining the diagnosis result of the object to be tested in the sample to be tested; Step S6: Input the collected fluorescence intensity signal of the reaction stage to the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model for signal feature vector extraction, and the time series prediction model for prediction of the complete reaction feature vector, and finally obtain the test result of the test object of the test sample through the test result output model.

[0024] Through deep learning methods, based on the reaction signals, predictive detection and analysis of the analytes in the test samples can be performed. There is no need to wait for the analytes to react completely on the fluorescent immunochromatography test strips to achieve predictive detection of the analytes, meeting the needs of rapid detection of analytes by fluorescent immunochromatography methods.

[0025] In one embodiment, the step S1: acquiring the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip specifically includes the following steps: By dropping the sample to be tested onto the fluorescent immunochromatographic test strip and reading the fluorescence intensity signal at each position on the fluorescent immunochromatographic test strip at a preset frequency, the reaction line can be accurately located, signal changes can be dynamically quantified, and multi-dimensional features can be extracted to provide high-resolution spatiotemporal data for the machine learning model, thereby achieving fast and high-precision early predictive detection until the reaction is complete, and obtaining fluorescence spectrum data of the entire reaction process.

[0026] In a specific embodiment, step S1 is implemented as follows: Each time, 75 μL of blood sample from a patient suspected of myocardial infarction was dripped onto a fluorescent immunochromatographic test strip, and then the test strip was immediately placed in a reader, and the fluorescence intensity spectrum of the test strip was read every 30 seconds. The reading was continued until the reaction was complete, that is, the fluorescence intensity of two adjacent time steps remained basically unchanged.

[0027] In one embodiment, in step S2: The reaction not started signal refers to the fluorescence spectrum signal read before the sample to be tested migrates to the detection line of the fluorescent immunochromatographic test strip; The reaction completion signal refers to the signal read when the fluorescence spectrum signal change rate within a preset time period is lower than the change rate threshold; specifically, the reaction completion signal refers to the signal read when the T / C line peak area ratio change rate within 30 seconds is lower than 10%; The reaction progress signal refers to the fluorescence spectrum signal read during the period between when the reaction has not started and when the reaction is complete; In one embodiment, the step S2: constructing a signal preprocessing model specifically includes the following steps: Step S21: According to the reaction kinetics characteristics of the fluorescent immunochromatographic test paper, a reaction stage segmentation standard is established, wherein the reaction stage includes the reaction not started, the reaction in progress and the reaction completed; wherein the reaction stage segmentation standard is: The liquid sample flows from the sample addition area to the absorbent pad and has not yet reached the test line (T line) and the quality control line (C line); The target antigen (such as cTnI) in the liquid sample specifically binds to the fluorescent-labeled antibody pre-embedded on the test strip to form an "antigen-antibody-fluorescent marker" complex and deposits on the test line (T line). At the same time, excess fluorescent-labeled antibodies continue to flow and bind to the quality control line (C line), causing the fluorescence spectrum signal intensity of the T line and C line to gradually increase over time; Reaction completion stage: The fluorescence spectrum signal intensity of the T line and the C line reaches a stable state; specifically, the change rate of the T / C line peak area ratio within 30 consecutive seconds is less than 10%.

[0028] Figure 3 (a) is the fluorescence intensity change diagram of the fluorescence immunochromatography method of a strong positive sample; Figure 3 (b) is the fluorescence intensity change diagram of the fluorescence immunochromatography method for weak positive samples; Figure 3 (c) is the fluorescence intensity change diagram of the fluorescent immunochromatography method for negative samples.

[0029] like Figure 3 In the fluorescence intensity spectrum data of (b), the blood sample has not migrated to the T line in the time period of 0min-1min, which means the reaction has not started; and the reaction is complete at 8min. The signal read from 0-1.5min is marked as the reaction has not started signal; the signal read from 1.5-7.5min is marked as the reaction progress signal; the signal read from 8min is marked as the reaction complete signal; Step S22: Based on the established reaction stage segmentation standard, a two-layer fully connected neural network, i.e., a signal preprocessing model, is constructed for signal classification, and the reaction stage of each signal in the acquired fluorescence spectrum data is distinguished and labeled. The network structure is as follows: Figure 4 As shown, it includes an input layer, a one-dimensional flattening layer, a hidden layer, a Softmax layer and an output layer. The input layer is used to input the acquired fluorescence spectrum data for data processing. The fluorescence spectrum data includes fluorescence intensity and position. The one-dimensional flattening layer is used to erase the position information in the data, arrange the fluorescence intensity in order of position, and obtain a one-dimensional vector of fluorescence intensity. The one-dimensional flattening layer is used to erase the position information in the data, arrange the fluorescence intensity in order of position, and obtain a one-dimensional vector of fluorescence intensity. The hidden layer is used to learn the mapping relationship between the one-dimensional vector and the reaction stage by adjusting the weight and bias; the Softmax layer is used to convert the 128-dimensional vector output by the hidden layer into a probability distribution; the output layer is used to distinguish and mark each reaction stage according to the probability distribution output by the Softmax layer.

[0030] In one embodiment, the step S3: constructing a signal feature extraction model specifically includes the following steps: Step S31: obtaining multidimensional characteristic values ​​of the fluorescence spectrum data of the reaction; wherein the multidimensional characteristic values ​​include but are not limited to T line peak area, C line peak area, T line fluorescence intensity, C line fluorescence intensity, T / C line peak area ratio, T / C line fluorescence intensity ratio and background fluorescence intensity; Step S32: Using the fluorescence spectrum data of the reaction as input and the multidimensional feature value as output, training is performed to obtain Figure 5 The signal feature extraction model shown includes an input layer, a one-dimensional flattening layer, a parallel fully connected feature extraction layer and a feature output layer. The input layer is used to input the fluorescence spectrum data of the reaction, which is two-dimensional data including position and fluorescence intensity. The one-dimensional flattening layer is used to flatten the two-dimensional fluorescence spectrum data into a one-dimensional vector. The parallel fully connected feature extraction layer includes three independent fully connected layers. The output feature of the first branch fully connected layer is the T line area, which quantifies the total amount of fluorescence signal of the detection line (T line) and directly reflects the concentration of the analyte. The output feature of the second branch fully connected layer is the C line area and the signal intensity of the quality control line (C line) to verify the validity of the test strip. The output feature of the third branch fully connected layer is the T / C ratio and the background intensity, which are used to standardize the signal (eliminate environmental interference) and background noise. Each layer dimension: input 128 dimensions → output 128 dimensions (marked with "128" in the figure); the feature output layer extracts four major features, including T line area, C line area, T / C ratio and background intensity.

[0031] In one embodiment, the step S4: constructing a time series prediction model specifically includes the following steps: Step S41: combining multiple fluorescence intensity signal features into a feature vector; first, the combination of multiple feature vectors (such as T / C line peak area ratio, fluorescence intensity, etc.) comprehensively captures the reaction dynamics and avoids single feature deviation; Step S42: setting a time window to obtain a feature vector sequence of fixed length; Step S43: Sliding the time window to capture the feature vectors covering the entire reaction stage and obtain training samples; the sliding time window mechanism converts the continuously sampled time series data into trainable sequence samples, which not only expands the amount of effective data but also retains the time dependence of the reaction kinetics; Step S44: Input the training sample into the Transformer model, adjust the Transformer model hyperparameters, and construct the following Figure 6 The time series prediction model shown in the figure can be quickly converged and achieve high prediction accuracy (mean square error 0.28 in the embodiment) through hyperparameter tuning.

[0032] Through multi-feature fusion, timing modeling and parameter optimization of reaction progress signals, Transformer can accurately predict the endpoint signal based on early reaction data.

[0033] In a specific embodiment, the step S4: constructing a time series prediction model is specifically implemented as follows: The 13 reaction feature vectors obtained on a single test strip are grouped together, and there are several groups of feature vectors. The time window is set to 4, and the time window starts from the first time step (t1) and slides to the 12th time step (t12). Figure 6 As shown. This operation is performed for each set of feature vectors, resulting in a large amount of training data. Transformer is called from the deep learning library and the network structure is defined: input dimension 4, hidden layer 128, number of attention heads 2, number of encoder layers 2. Absolute position encoding is used to enable Transformer to perceive the position of the time step in the input sequence. The hidden state of the last position of the input sequence is extracted as the representation of the entire sequence, and the final prediction result is generated through this state. Finally, with the time window as input and the t13 feature vector as output, the Transformer is trained for 1000 iterations, and the mean square error is about 0.28.

[0034] In one embodiment, the step S5: constructing a detection result output model specifically includes the following steps: Step S51: setting the concentration threshold of the analyte corresponding to each diagnostic conclusion of the analyte in the blood sample; specifically, the concentration is inversely calculated by the T / C peak area ratio of the complete reaction, and a blood sample with a concentration of <0.1 ng / mL is defined as negative, a blood sample with a concentration of 0.1-0.3 ng / mL is defined as weak positive, and a blood sample with a concentration of >0.3 ng / mL is defined as positive; Step S52: obtaining a complete reaction fluorescence intensity signal feature vector of the object to be tested; wherein the complete reaction feature vector is a complete reaction feature vector extracted by a signal feature extraction model or a complete reaction feature vector obtained by time series prediction using a time series prediction model; Step S53: The diagnosis result is used as a label, and the t13 feature vector output by the time series prediction model is used as input to train the detection result output model. Figure 7 The test result output model shown includes an input layer, a hidden layer and an output layer, wherein the input layer is used to input a complete reaction feature vector; the first layer of the hidden layer maps the 4-dimensional original features to a 128-dimensional space to extract nonlinear combination features; the second layer is used to deeply abstract features from the extracted nonlinear combination features to capture the interactive relationship between the T / C ratio and the background intensity; the output layer is used to output the test result of the test object according to the interactive relationship between the T / C ratio and the background intensity, when the concentration of the test object is lower than the detection threshold, it is negative; when the concentration of the test object is between the detection threshold and the critical value range, it is weakly positive; when the concentration of the test object is higher than the diagnostic threshold, it is positive.

[0035] In one embodiment, step S6: inputting the collected fluorescence intensity signal of the reaction process stage to the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model for signal feature vector extraction, and the time series prediction model for prediction of the complete reaction feature vector, and finally obtaining the test result of the analyte of the sample to be tested through the test result output model, is specifically implemented as follows: Finally, integrate the trained model, such as Figure 2 As shown, the time-continuous spectral data are sequentially input into the signal preprocessing model, and the signals that cannot or do not need to be predicted, such as the reaction not started signal, are eliminated. The reaction progress signal is converted into a feature vector composed of 4 features through the signal feature extraction model. The feature vectors of four consecutive time steps are input into the timing prediction model to obtain the t13 feature vector. The t13 feature vector is passed into the test result output model to obtain the test result of the object to be tested. The traditional 15-minute detection process is compressed to 3.5 minutes, while maintaining a diagnostic accuracy rate of more than 95%, to gain golden treatment time for critical diseases such as acute myocardial infarction. On the second aspect, this application provides an application of a deep learning-based fluorescent immunochromatographic test strip rapid reading method for detecting troponin I in the blood, which is conducive to shortening the in vitro diagnosis time of troponin I, so as to automatically identify the elevated concentration of troponin I in the test strip fluorescence intensity signal, help doctors to discover potential risks in advance, and improve clinical response speed. The final test accuracy was above 95%, and the diagnosis time was shortened to a minimum of 3.5 min.

[0036] In a third aspect, the present application provides a deep learning-based rapid reading system for fluorescent immunochromatographic test strips, comprising: A fluorescence spectrum data acquisition module is used to collect and acquire the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip; A signal preprocessing model building module is communicatively connected with the fluorescence spectrum data acquisition module, and is used to build a signal preprocessing model, distinguish and annotate the collected fluorescence spectrum data according to the reaction stages, and acquire the annotated fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; A signal feature model building module is communicatively connected with the signal preprocessing model building module and is used to build a signal feature extraction model to extract a fluorescence intensity signal feature vector from the annotated fluorescence spectrum data; A time series prediction model building module is connected to the signal feature model building module for building a time series prediction model and performing time series prediction processing on the reaction feature vector to obtain a complete reaction feature vector; A result output model construction module, which is in communication connection with the time series prediction model construction module, is used to construct a detection result output model, perform diagnosis according to the complete reaction feature vector, and obtain the diagnosis result of the object to be tested in the sample to be tested; The detection result acquisition module is communicatively connected with the signal preprocessing model construction module, the signal feature model construction module, the timing prediction model construction module and the result output model construction module, and is used for inputting the collected fluorescence intensity signal of the reaction stage into the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model for signal feature vector extraction, and the timing prediction model for prediction of complete reaction feature vector, and finally obtaining the detection result of the object to be tested of the sample to be tested through the detection result output model.

[0037] In a fourth aspect, an embodiment of the present application provides a deep learning-based rapid reading device for fluorescent immunochromatographic test strips for rapid in vitro diagnosis of troponin I. The deep learning-based rapid reading device for fluorescent immunochromatographic test strips for rapid in vitro diagnosis of troponin I can be a personal computer (PC), a laptop computer, a server, an embedded device, or other device with data processing capabilities.

[0038] The communication interface includes input / output (I / O) interface, physical interface and logical interface, etc., which are used to interconnect the devices inside the fluorescent immunochromatographic test strip rapid reading device based on deep learning for rapid in vitro diagnosis of troponin I, and the interface used to interconnect the fluorescent immunochromatographic test strip rapid reading device based on deep learning for rapid in vitro diagnosis of troponin I with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display, a keyboard, etc.

[0039] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0040] The processor may be a general-purpose processor, which may call the deep learning-based fluorescent immunochromatographic test strip rapid reading program for rapid in vitro diagnosis of troponin I stored in the memory, and execute the deep learning-based fluorescent immunochromatographic test strip rapid reading method provided in the embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the deep learning-based fluorescent immunochromatographic test strip rapid reading program for rapid in vitro diagnosis of troponin I is called may refer to the various embodiments of the deep learning-based fluorescent immunochromatographic test strip rapid reading method of the present application, which will not be repeated here.

[0041] In a fifth aspect, an embodiment of the present application also provides a readable storage medium.

[0042] The readable storage medium of the present application stores a deep learning-based rapid reading program for fluorescent immunochromatographic test strips for rapid in vitro diagnosis of troponin I. When the deep learning-based rapid reading program for fluorescent immunochromatographic test strips for rapid in vitro diagnosis of troponin I is executed by a processor, the steps of the deep learning-based rapid reading method for fluorescent immunochromatographic test strips as described above are implemented.

[0043] Among them, the method implemented when the deep learning-based fluorescent immunochromatographic test strip rapid reading program for rapid in vitro diagnosis of troponin I is executed can refer to the various embodiments of the deep learning-based fluorescent immunochromatographic test strip rapid reading method of the present application, and will not be repeated here.

[0044] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0045] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0046] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for rapid reading of fluorescent immunochromatographic test strips based on deep learning, characterized in that: The following steps are involved: Collect and obtain the fluorescence spectrum data of the sample to be tested on the fluorescent immunochromatographic test strip; Constructing a signal preprocessing model to distinguish and annotate the collected fluorescence spectrum data according to the reaction stages, and obtaining the annotated fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; Construct a signal feature extraction model to extract fluorescence intensity signal feature vectors from the annotated fluorescence spectrum data; Constructing a time series prediction model to perform time series prediction processing on the reaction feature vector to obtain the complete reaction feature vector; Constructing a test result output model for performing diagnosis based on the complete characteristic vector of the reaction and obtaining the diagnosis result of the object to be tested in the test sample; The collected fluorescence intensity signal of the reaction stage is input into the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model is used to extract the signal feature vector, and the time series prediction model is used to predict the complete feature vector of the reaction. Finally, the test result of the test object of the test sample is obtained through the test result output model.

2. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The method of collecting and obtaining the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip specifically includes the following steps: The sample to be tested is added to the fluorescent immunochromatographic test strip, and the fluorescence intensity signal at each position on the fluorescent immunochromatographic test strip is read at a preset frequency until the reaction is complete, and the fluorescence spectrum data is obtained.

3. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The reaction not started signal refers to the fluorescence spectrum signal read before the sample to be tested migrates to the detection line of the fluorescent immunochromatographic test strip; the reaction complete signal refers to the signal read when the change rate of the fluorescence spectrum signal within a preset time period is lower than the change rate threshold; the reaction progress signal refers to the fluorescence spectrum signal read during the time period between the reaction not started and the reaction complete.

4. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The construction of the signal preprocessing model specifically includes the following steps: According to the reaction kinetic characteristics of the fluorescent immunochromatographic test paper, a reaction stage segmentation standard is established, wherein the reaction stage includes the reaction not started, the reaction is in progress and the reaction is completed; According to the established reaction stage segmentation standard, the acquired fluorescence spectrum data are distinguished and labeled for the reaction stages.

5. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The construction of the signal feature extraction model specifically includes the following steps: Obtain multidimensional characteristic values ​​of fluorescence spectrum data of the reaction; The fluorescence spectrum data of the reaction is taken as input and the multi-dimensional feature values ​​are taken as output to train and obtain the signal feature extraction model.

6. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The construction of the time series prediction model specifically includes the following steps: Combining multiple fluorescence intensity signal features into a feature vector; Set the time window to obtain a fixed-length feature vector sequence; Slide the time window to capture the feature vector covering all stages of the reaction and obtain training samples; Input the training samples into the Transformer model to build a time series prediction model.

7. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 1, characterized in that: The construction of the detection result output model specifically includes the following steps: Setting the concentration threshold of the analyte corresponding to each diagnostic conclusion of the analyte in the blood sample; Obtaining the complete fluorescence intensity signal characteristic vector of the reaction of the analyte; According to the set diagnostic threshold of the concentration of the analyte, the complete characteristic vector of the reaction is detected and judged to obtain the detection result of the analyte in the sample to be tested.

8. The method for rapid reading of fluorescent immunochromatographic test strips based on deep learning according to claim 7, characterized in that: The complete reaction feature vector is a complete reaction feature vector extracted by a signal feature extraction model or a complete reaction feature vector obtained by time series prediction through a time series prediction model.

9. An application of the deep learning-based rapid reading method for fluorescent immunochromatographic test strips as described in any one of claims 1 to 8, characterized in that: Used to measure troponin I in the blood.

10. A rapid reading system for fluorescent immunochromatographic test strips based on deep learning, characterized in that: include: A fluorescence spectrum data acquisition module is used to collect and acquire the fluorescence spectrum data of the sample to be tested on the fluorescence immunochromatography test strip; A signal preprocessing model building module is connected to the fluorescence spectrum data acquisition module for building a signal preprocessing model, distinguishing and labeling the collected fluorescence spectrum data according to the reaction stages, and acquiring the labeled fluorescence spectrum data, wherein the reaction stages include the reaction not started, the reaction in progress, and the reaction completed; A signal feature model building module is communicatively connected with the signal preprocessing model building module and is used to build a signal feature extraction model to extract a fluorescence intensity signal feature vector from the annotated fluorescence spectrum data; A time series prediction model building module is connected to the signal feature model building module for building a time series prediction model and performing time series prediction processing on the reaction feature vector to obtain a complete reaction feature vector; A result output model construction module, which is in communication connection with the time series prediction model construction module, is used to construct a detection result output model, perform diagnosis according to the complete reaction feature vector, and obtain the diagnosis result of the object to be tested in the sample to be tested; The detection result acquisition module is communicatively connected with the signal preprocessing model construction module, the signal feature model construction module, the timing prediction model construction module and the result output model construction module, and is used for inputting the collected fluorescence intensity signal of the reaction stage into the signal preprocessing model for time domain differentiation and annotation, the signal feature extraction model for signal feature vector extraction, and the timing prediction model for prediction of complete reaction feature vector, and finally obtaining the detection result of the object to be tested of the sample to be tested through the detection result output model.

Images