Digital imaging and AI fused RPR shaking table detection method and system

Abstract

The invention relates to the technical field of medical detection, and particularly discloses a digital imaging and AI fused RPR shaking table detection method and system.The method comprises the steps that a serum sample is evenly shaken and mixed through an RPR shaking table, and the shaking frequency and amplitude are adjusted through a real-time monitoring and feedback system; acquiring a serum sample image processed by the RPR shaking table in real time through a high-resolution CMOS / CCD (Complementary Metal Oxide Semiconductor / Charge Coupled Device) digital camera or camera; preprocessing the collected serum sample image, including denoising, enhancement, segmentation and edge detection, and extracting key features in the serum sample; according to the intelligent RPR shaking table detection method and system combining digital imaging, AI identification and remote management, the detection efficiency can be remarkably improved, the misjudgment rate can be reduced, data storage and remote review are supported, and the wide market application prospect is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical detection, and particularly relates to an RPR shaker detection method and system integrating digital imaging and AI. Background Art

[0002] RPR (Rapid Plasma Reagin) detection is a serological detection method commonly used for the diagnosis of diseases such as syphilis. It evaluates the infection situation by judging the degree of antigen-antibody reaction in the serum. Traditional RPR detection methods mostly rely on manual visual observation of the changes in serum samples after reaction on a shaker and judge the reaction intensity by the naked eye. However, this method has several deficiencies, seriously affecting the accuracy and efficiency of detection.

[0003] First of all, manual visual judgment is easily affected by subjective factors, resulting in a relatively high misjudgment rate. Due to the differences in observation standards and experience among different operators, misjudgments and inconsistencies are likely to occur during batch detection. Secondly, the detection efficiency of traditional methods is relatively low. Especially during batch detection, it is necessary to manually observe and record the results one by one, which is time-consuming and labor-intensive and does not meet the needs of modern medical care. In addition, traditional RPR detection methods usually cannot be digitally archived, making it difficult to store, manage, and remotely review data, resulting in the inability to effectively trace historical data later and increasing the difficulty of data management. Finally, existing RPR detection methods lack intelligent analysis functions, cannot automatically distinguish different detection results such as negative, weakly positive, and positive, rely on manual judgment, are easily interfered by human errors, and cannot meet the requirements of high precision and high efficiency.

[0004] Although there are some digital detection devices on the market currently, the main problems of these devices have not been effectively solved. First of all, most of the existing devices only have the function of image magnification, cannot perform intelligent recognition and automatic analysis, still require manual intervention, and cannot achieve true automation and intelligence. Secondly, the imaging quality of the devices is affected by light and sample position, which may lead to unclear imaging or large errors, thus affecting the accuracy of the interpretation results. More critically, most of the existing devices cannot be seamlessly docked with the hospital information system (HIS / LIS), cannot support batch management and real-time sharing of data, and limit their popularization and application in clinical practice.

[0005] Therefore, it is necessary to propose an RPR shaker detection method and system integrating digital imaging and AI to solve the problems existing in the prior art that existing devices only have the function of image magnification, cannot perform intelligent recognition, and the imaging quality is limited, and light and sample position may affect the interpretation results.

[0006] The above information disclosed in this background technology is only used to enhance the understanding of the background technology of the present invention. Therefore, it may include prior art that is not known to those of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide an RPR shaker detection method and system that integrates digital imaging and AI to solve the problems presented in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solutions:

[0009] An RPR shaker detection method that integrates digital imaging and AI, including:

[0010] Use an RPR shaker to evenly shake and mix serum samples, and adjust the shaking frequency and amplitude through a real-time monitoring and feedback system;

[0011] Through a high-resolution CMOS / CCD digital camera or camera, real-time collect images of serum samples processed by the RPR shaker;

[0012] Preprocess the collected serum sample images, including denoising, enhancement, segmentation, and edge detection, and extract key features in the serum samples;

[0013] Based on a deep learning model, analyze the preprocessed serum sample images, infer whether the serum samples contain target substances, and generate analysis results;

[0014] Compare the analysis results with historical data and output a detection report that includes the concentration of target substances, reaction patterns, and conclusions on whether there are abnormalities in the serum samples;

[0015] Real-time feedback the detection report to the operation interface. If an abnormal reaction is detected, automatically prompt relevant measures, and self-diagnose and perform performance checks on the operating status of the RPR shaker regularly or according to preset conditions.

[0016] Preferably, the step of using an RPR shaker to evenly shake and mix serum samples and adjusting the shaking frequency and amplitude through a real-time monitoring and feedback system includes:

[0017] Start the RPR shaker, set the initial shaking frequency to 100 revolutions per minute and the amplitude to 10 degrees;

[0018] During the shaking process, use sensors to real-time collect serum sample status data, including shaking frequency, amplitude, and sample temperature;

[0019] During the shaking process, real-time monitor the particle precipitation situation in the serum samples. Once particle precipitation is detected, immediately adjust the shaking frequency and amplitude;

[0020] During the shaking process, remotely adjust the shaking frequency and amplitude according to the collected sample status data through a wireless connection protocol.

[0021] Preferably, the serum sample images after being processed by an RPR shaker are collected in real time through a high-resolution CMOS / CCD digital camera or a camera, including:

[0022] Initialize the high-resolution CMOS / CCD digital camera or the camera, and set the acquisition frequency to 30 frames per second;

[0023] Configure the adaptive focusing function of the camera, and automatically adjust the focal length according to the distance and light conditions of the serum sample;

[0024] Enable multi-spectral LED lighting, and select the optimal lighting spectrum according to the characteristics of the serum sample;

[0025] Collect serum sample images from multiple angles and synthesize a panoramic image;

[0026] During the acquisition process, monitor the position change of the serum sample in real time, and automatically adjust the camera focus to obtain serum sample images at different positions.

[0027] Preferably, preprocess the collected serum sample images, including denoising, enhancement, segmentation, and edge detection, and extract the key features in the serum sample, including:

[0028] Use the deep convolutional denoising autoencoder algorithm to denoise the collected serum sample images and eliminate the noise in the images;

[0029] Apply the adaptive histogram equalization algorithm to enhance the denoised images and optimize the contrast of the images;

[0030] Adaptive histogram equalization algorithm formula:

[0031]

[0032] In the formula, Ν(x,y) is the neighborhood of the target pixel point (x,y), h(i,j) is the histogram value of the pixels in the neighborhood, and N is the neighborhood size;

[0033] Combine the segmentation algorithm based on an adaptive threshold with the semantic segmentation algorithm to segment and detect the edges of the enhanced images, and extract the boundary between the serum components and the background.

[0034] Preferably, analyze the preprocessed serum sample images based on a deep learning model, infer whether the target substance is contained in the serum sample, and generate an analysis result, including:

[0035] Build a deep learning model, adopt the CNN+YOLO object detection architecture, and combine with the LSTM network to process temporal information to infer the dynamic changes of particles in serum sample images;

[0036] CNN output features:

[0037] F CNN = CNN(I)

[0038] YOLO object detection:

[0039] D = YOLO(F CNN )

[0040] LSTM network:

[0041]

[0042] In the formula, I is the input image, h t is the hidden state of the LSTM network;

[0043] The model is trained using a clinical-annotated RPR rocker reaction image dataset, and rare type sample image data is synthesized through data augmentation techniques;

[0044] Use the trained deep learning model to analyze the preprocessed serum sample images, and dynamically adjust and optimize the errors of the analysis results in combination with the reinforcement learning algorithm.

[0045] Preferably, comparing the analysis results with historical data and outputting a detection report including the concentration of the target substance, reaction pattern, and whether it is abnormal in the serum sample, including:

[0046] The report is generated by an AI model trained based on historical detection data, automatically analyzes the concentration change trends of various substances in the serum, and visually displays them in a combined manner of charts and images;

[0047] Upload the detection report to the cloud for storage results, provide a remote diagnosis function, upload data to the hospital LIS / HIS system in real time, and integrate it into the hospital laboratory automation system;

[0048] The AI model adopts federated learning technology to train and optimize the model among multiple medical institutions;

[0049] Local model update and global aggregation formula:

[0050]

[0051] In the formula, is the model parameter of the k-th client, w t+1 is the global model parameter.

[0052] Preferably, the RPR shaker performs self-diagnosis and performance inspection on its operating status regularly or according to preset conditions, including:

[0053] When potential faults or performance degradation are detected, an early warning is immediately sent to the operator for timely repair or replacement of components;

[0054] An anomaly detection algorithm based on deep learning is used to monitor and give early warnings on the performance of key components of the RPR shaker in real time;

[0055] Anomaly detection algorithm formula:

[0056]

[0057] In the formula, ξ z is a slack variable, s is the weight vector of the decision boundary, ρ is the bias term, and R is the total number of samples in the dataset.

[0058] The RPR shaker detection system integrating digital imaging and AI includes:

[0059] An RPR shaker for uniformly shaking and mixing serum samples and adjusting the shaking frequency and amplitude through a real-time monitoring and feedback system;

[0060] A high-resolution CMOS / CCD digital camera or camera for real-time acquisition of images of serum samples processed by the RPR shaker;

[0061] An image processing unit for preprocessing the acquired serum sample images and extracting key features in the serum samples;

[0062] A deep learning model for analyzing the preprocessed serum sample images, inferring whether the serum samples contain target substances, and generating analysis results;

[0063] A report generation unit for comparing the analysis results with historical data and outputting a detection report including the concentration of target substances, reaction patterns, and conclusions on whether there are abnormalities in the serum samples;

[0064] A self-diagnosis and performance inspection unit for performing self-diagnosis and performance inspection on the operating status of the RPR shaker regularly or according to preset conditions.

[0065] Preferably, the RPR shaker includes multiple independently controllable shaking units, and each shaking unit automatically adjusts the shaking frequency and amplitude according to the characteristics of the serum samples;

[0066] The real-time monitoring and feedback system includes multiple sensors for real-time acquisition of serum sample status data and introducing an adaptive control algorithm to adjust the shaking frequency and amplitude through real-time feedback data;

[0067] Adaptive control algorithm formula:

[0068]

[0069] Wherein, u(t) is the control signal, e(t) is the error, M(t) is the time-varying gain, and α is the compensation coefficient. is the integral of the error from time 0 to the current time t;

[0070] The high-resolution CMOS / CCD digital camera or camera includes an adaptive focusing function, automatically adjusts the focal length according to the distance and light conditions of the serum sample, and enables multi-spectral LED lighting to select the best lighting spectrum according to the characteristics of the serum sample;

[0071] The image processing unit uses the deep convolutional denoising autoencoder algorithm to denoise the acquired images, and applies the adaptive histogram equalization algorithm to enhance the denoised images;

[0072] A generative adversarial network is introduced to enhance the image quality. The formula of the generative adversarial network loss function is as follows:

[0073]

[0074] Wherein, D(a) is the output probability of the discriminator for the real data, G(r) is the output of the generator, Pdata(a) is the real data distribution, and P r (r) is the distribution of the latent space.

[0075] Preferably, the deep learning model adopts the CNN+YOLO architecture, combines the long short-term memory network with temporal information, infers the dynamic changes of the particles in the serum sample image, and dynamically adjusts and optimizes the error of the analysis result through the reinforcement learning algorithm;

[0076] A graph neural network is introduced to optimize the spatial relationship and dynamic changes between different regions in the serum sample image;

[0077] Graph neural network information propagation formula:

[0078]

[0079] Wherein, is the representation of node d in the l+1 layer, B(d) is the set of neighbor nodes of node d, and A (l) and b (l) are the weights and biases of the l-th layer, and σ is the activation function;

[0080] The report generation unit is based on the AI model trained from historical detection data, automatically analyzes the concentration change trend of various substances in the serum, and visually displays it in a combined manner of charts and images;

[0081] The self-diagnosis and performance inspection unit uses an anomaly detection algorithm based on deep learning to monitor and give early warnings about the performance of key components of the RPR shaker in real time.

[0082] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0083] Through high-resolution imaging, image preprocessing, deep learning model analysis, and a real-time feedback system, the present invention significantly improves the detection accuracy and efficiency, reduces the errors caused by manual intervention, automatically identifies the target substances in serum samples and their concentration changes, and provides high-quality detection reports. Moreover, it supports data storage, comparison of historical data, and remote diagnosis, enhancing the ability of data management and sharing. In addition, the present invention also has a self-diagnosis and performance inspection function to ensure the stability and reliability of the device during operation, meeting the requirements of modern medicine for high precision and high efficiency.

[0084] In summary, the present invention proposes an intelligent RPR shaker detection method and system combining digital imaging, AI recognition, and remote management, which can significantly improve the detection efficiency, reduce the misjudgment rate, support data storage and remote reexamination, and has broad market application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is a flowchart of the RPR shaker detection method integrating digital imaging and AI of the present invention;

[0086] Figure 2 It is a framework diagram of the RPR shaker detection system integrating digital imaging and AI of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0087] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0088] Embodiment 1:

[0089] Please refer to Figure 1 As shown, the RPR shaker detection method integrating digital imaging and AI includes:

[0090] Using the RPR shaker to uniformly shake and mix the serum sample, and adjusting the shaking frequency and amplitude through a real-time monitoring and feedback system;

[0091] Start the RPR shaker, and set the initial shaking frequency to 100 revolutions per minute and the amplitude to 10 degrees;

[0092] During the shaking process, the status data of the serum sample is collected in real time through sensors, including the shaking frequency, amplitude, and sample temperature;

[0093] During the shaking process, the particle precipitation in the serum sample is monitored in real time. Once particle precipitation is detected, the shaking frequency and amplitude are immediately adjusted;

[0094] During the shaking process, according to the collected sample status data, the shaking frequency and amplitude are remotely adjusted through a wireless connection protocol.

[0095] Furthermore, through real-time monitoring and feedback adjustment of the shaking frequency and amplitude, this RPR shaker device can ensure the uniform mixing of the serum sample during the shaking process, effectively avoiding particle precipitation. At the same time, the device has the functions of real-time data collection and remote adjustment, dynamically optimizing the shaking parameters according to the sample status, improving the detection efficiency and intelligent level, and solving the problems of manual operation and low efficiency in traditional methods.

[0096] Through a high-resolution CMOS / CCD digital camera or camera, the image of the serum sample after being processed by the RPR shaker is collected in real time;

[0097] Initialize the high-resolution CMOS / CCD digital camera or camera with a resolution ≥ 12 million pixels and set the acquisition frequency to 30 frames per second;

[0098] Configure the adaptive focusing function of the camera to automatically adjust the focal length according to the distance and light conditions of the serum sample;

[0099] Enable multi-spectral LED lighting and select the best lighting spectrum according to the characteristics of the serum sample;

[0100] Collect serum sample images from multiple angles and synthesize a panoramic image;

[0101] During the acquisition process, the position change of the serum sample is monitored in real time, and the camera focus is automatically adjusted to obtain serum sample images at different positions.

[0102] Furthermore, this image acquisition method realizes the high-precision real-time image acquisition of the serum sample through technologies such as high-resolution CMOS / CCD digital cameras, intelligent focusing, multi-spectral LED lighting, and polarization light to eliminate reflections. It supports multi-angle shooting and panoramic image synthesis, improving the comprehensiveness and accuracy of image acquisition, and contributing to subsequent intelligent analysis and result interpretation.

[0103] Preprocess the collected serum sample images, including denoising, enhancement, segmentation, and edge detection, to extract the key features in the serum sample;

[0104] The deep convolutional denoising autoencoder algorithm is used to denoise the collected serum sample images and eliminate the noise in the images;

[0105] The adaptive histogram equalization algorithm is applied to enhance the denoised images and optimize the contrast of the images;

[0106] Combining the segmentation algorithm based on adaptive threshold and the semantic segmentation algorithm, the enhanced images are segmented and edge detected to extract the boundary between the serum components and the background.

[0107] Furthermore, through the deep convolutional denoising autoencoder, histogram equalization, adaptive threshold segmentation and semantic segmentation algorithms, this image preprocessing process provides high-quality and detailed image data for subsequent analysis, thus improving the accuracy and reliability of serum sample analysis.

[0108] Based on the deep learning model, the preprocessed serum sample images are analyzed to infer whether the target substance is contained in the serum samples and generate analysis results;

[0109] A deep learning model is constructed, adopting the CNN+YOLO object detection architecture and combining with the LSTM network to process the temporal information and infer the dynamic changes of the particles in the serum sample images;

[0110] The model training uses the clinically annotated RPR shaker reaction image dataset (including negative, weakly positive, positive, and image data with multiple concentration gradients), and synthesizes rare type of sample image data through data augmentation techniques;

[0111] The trained deep learning model is used to analyze the preprocessed serum sample images, and the error of the analysis results is dynamically adjusted and optimized by combining with the reinforcement learning algorithm.

[0112] Classification criteria:

[0113] Interpretation is based on the size, shape, and distribution density of the agglutinated particles;

[0114] The results are automatically classified as negative (-), weakly positive (±), positive (+), strongly positive (++);

[0115] Automatically compare with the standard samples to reduce the misjudgment rate (target accuracy rate > 98%).

[0116] Furthermore, through the clinically annotated RPR shaker reaction image dataset and data augmentation techniques, the model can accurately process various types of sample data and be automatically classified as negative, weakly positive, positive or strongly positive. Combining with the reinforcement learning algorithm, the system dynamically optimizes the analysis results, reduces the misjudgment rate, and ensures that the target accuracy rate exceeds 98%, greatly improving the accuracy and efficiency of serum sample analysis.

[0117] Compare the analysis results with historical data and output a test report containing the concentration of the target substance, reaction mode, and whether it is abnormal in the serum sample;

[0118] The report is generated by an AI model trained based on historical test data, automatically analyzing the concentration change trends of various substances in the serum, and visually displaying them in a combined way of charts and images;

[0119] Upload the test report to the cloud for storage results and provide remote diagnosis functions, support Wi-Fi / Bluetooth / USB interfaces, upload data to the hospital LIS / HIS system in real time, and integrate it into the hospital laboratory automation system through the API open interface;

[0120] The AI model adopts federated learning technology to train and optimize the model among multiple medical institutions.

[0121] Furthermore, by automatically analyzing serum samples and combining historical data to generate a detailed test report, providing substance concentration, reaction mode, and abnormal conclusions, and visually displaying the test results. Using AI models and federated learning technology, sharing data and optimizing models among multiple medical institutions to improve diagnostic accuracy. At the same time, supporting cloud storage and remote diagnosis functions, realizing real-time data upload and integration with the hospital LIS / HIS system, promoting the efficient circulation and intelligent management of medical data.

[0122] Feed back the test report to the operation interface in real time. If an abnormal reaction is found, automatically prompt relevant measures, and conduct self-diagnosis and performance inspection on the operation status of the RPR shaker regularly or according to preset conditions;

[0123] When potential faults or performance degradation are found, immediately issue a warning to the operator and repair or replace components in a timely manner;

[0124] Adopt an anomaly detection algorithm based on deep learning to monitor and warn the performance of key components of the RPR shaker in real time.

[0125] Furthermore, by feeding back the test report in real time and automatically prompting relevant measures for abnormal reactions, the reaction speed and accuracy in the operation process are improved. Combining regular self-diagnosis with an anomaly detection algorithm based on deep learning, monitoring the performance of key components of the RPR shaker in real time, and issuing warnings in a timely manner when potential faults or performance degradation occur, helping the operator to repair or replace components in a timely manner, ensuring the stability and reliability of the equipment.

[0126] Example 2:

[0127] Please refer to Figure 2 As shown, the RPR shaker detection system integrating digital imaging and AI includes:

[0128] The RPR shaker is used to uniformly shake and mix serum samples, and adjust the shaking frequency and amplitude through a real-time monitoring and feedback system;

[0129] The RPR shaker includes multiple independently controllable shaking units, and each shaking unit automatically adjusts the shaking frequency and amplitude according to the characteristics of the serum sample;

[0130] The base of the RPR shaker is driven by a brushless motor to ensure the stability of long-term operation, and the oscillation frequency and amplitude can be adjusted (range: 50 - 150 rpm) to optimize the antigen-antibody mixing effect;

[0131] The real-time monitoring and feedback system includes multiple sensors for real-time collecting the status data of serum samples, and introduces an adaptive control algorithm to adjust the shaking frequency and amplitude through real-time feedback data;

[0132] A high-resolution CMOS / CCD digital camera or camera is used to real-time collect the images of serum samples processed by the RPR shaker;

[0133] The high-resolution CMOS / CCD digital camera or camera includes an adaptive focusing function, automatically adjusts the focal length according to the distance and light conditions of the serum sample, and enables multi-spectral LED lighting to select the best lighting spectrum according to the characteristics of the serum sample;

[0134] The image processing unit is used to preprocess the collected images of serum samples and extract the key features in the serum samples;

[0135] The image processing unit uses the deep convolutional denoising autoencoder algorithm to denoise the collected images, and applies the adaptive histogram equalization algorithm to enhance the denoised images;

[0136] The deep learning model is used to analyze the preprocessed images of serum samples, infer whether the target substance is contained in the serum samples, and generate analysis results;

[0137] The deep learning model adopts the CNN + YOLO architecture, combines the long short-term memory network with time series information, infers the dynamic changes of particles in the serum sample images, and dynamically adjusts and optimizes the error of the analysis results through the reinforcement learning algorithm;

[0138] Introduce the graph neural network to optimize the spatial relationship and dynamic changes between different regions in the serum sample images;

[0139] The report generation unit is used to compare the analysis results with historical data and output a detection report including the concentration of the target substance in the serum sample, the reaction mode, and the conclusion of whether it is abnormal;

[0140] The report generation unit uses an AI model trained based on historical detection data to automatically analyze the concentration change trends of various substances in the serum and visually display them in a combined way of charts and images;

[0141] The self-diagnosis and performance inspection unit is used to perform self-diagnosis and performance inspection on the operating status of the RPR shaker regularly or according to preset conditions;

[0142] The self-diagnosis and performance inspection unit uses an anomaly detection algorithm based on deep learning to monitor and give early warnings on the performance of key components of the RPR shaker in real time;

[0143] The user interaction interface uses a 7-inch touch screen to display the detection progress and analysis results and supports remote operation (synchronous monitoring on the PC side / mobile side).

[0144]

[0145] Application example: Medical staff use the RPR shaker to detect patients' serum samples

[0146] In a clinical laboratory of a hospital, medical staff need to use the RPR shaker to detect a patient's serum sample to determine whether it contains specific target substances, such as a certain pathogen or biomarker.

[0147] I. Application process

[0148] Medical staff first prepare the patient's serum sample and place it on the RPR shaker.

[0149] Start the RPR shaker through the operation interface, set the initial shaking frequency to 100 revolutions per minute and the amplitude to 10 degrees. The shaker starts to shake the serum sample evenly to ensure that the components in the sample are fully mixed.

[0150] During the shaking process, the real-time monitoring and feedback system collects the status data of the serum sample in real time through sensors, including the shaking frequency, amplitude and sample temperature.

[0151] Medical staff can observe these data through the operation interface. Once abnormal situations such as particle precipitation are found, the system immediately automatically adjusts the shaking frequency and amplitude to ensure the uniform mixing of the sample.

[0152] When the serum sample has been shaken and mixed for a period of time, medical staff start a high-resolution CMOS / CCD digital camera or camera to collect the image of the serum sample processed by the RPR shaker in real time. The camera or camera is configured with an adaptive focusing function, multi-spectral LED lighting and polarization light reflection elimination function to ensure that the collected image quality is clear and accurate.

[0153] The collected serum sample images are transmitted to the image processing unit for preprocessing. The preprocessing includes steps such as denoising, enhancement, segmentation, and edge detection to extract the key features in the serum samples.

[0154] After preprocessing, the preprocessed serum sample images are analyzed based on a deep learning model (a combination of CNN + YOLO architectures and an LSTM network). The model infers whether the target substance is contained in the serum sample and generates an analysis result.

[0155] The analysis result is compared with historical data to generate a test report containing the concentration of the target substance in the serum sample, the reaction pattern, and the conclusion of whether it is abnormal.

[0156] The test report is fed back to the operation interface in real time, and medical staff can intuitively view and analyze the report. If an abnormal reaction is detected, the system will also automatically prompt relevant measures for medical staff to take further actions in a timely manner.

[0157] During the entire detection process, the RPR shaker will also perform self-diagnosis and performance checks regularly or according to preset conditions. An anomaly detection algorithm based on deep learning is used to monitor and give early warnings about the performance of the key components of the shaker in real time. Once potential faults or performance degradation are detected, the system immediately issues a warning to medical staff for timely repair or replacement of components.

[0158] II. Advantages and Experiences

[0159] Improve detection accuracy and efficiency: Through high-resolution imaging, image preprocessing, deep learning model analysis, and real-time feedback system, the RPR shaker significantly improves detection accuracy and efficiency, enabling medical staff to obtain accurate test results faster.

[0160] Reduce manual intervention errors: Automatically identify the target substance and its concentration changes in serum samples, reducing manual intervention errors and improving the accuracy and reliability of detection.

[0161] Support data storage and remote review: Provide high-quality test reports and support data storage, historical data comparison, and remote diagnosis, enhancing the ability of data management and sharing. Medical staff can conveniently view and analyze historical data for remote review and consultation.

[0162] Self-diagnosis and performance check function: Ensure the stability and reliability of the equipment during operation, reducing maintenance costs. Medical staff can use the RPR shaker for detection work with more confidence.

[0163] Example 3:

[0164] An embodiment of the present invention also provides a computer-readable storage medium, on which a program of the RPR rocker detection system integrating digital imaging and AI as described in any one of the above is stored. When the program is executed by a processor, it implements each process of the above RPR rocker detection system embodiment and can achieve the same technical effects. To avoid repetition, it will not be elaborated here. Among them, the computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0165] In the description of this specification, the description referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0166] In the drawings of the disclosed embodiments of the present invention, only the structures related to the disclosed embodiments of the present invention are involved. Other structures can refer to the general design. Without conflict, the same embodiment and different embodiments of the present invention can be combined with each other.

[0167] The flowcharts shown in the drawings are only illustrative examples, and do not necessarily include all the contents and operations / steps, nor do they necessarily need to be executed in the described order. For example, some operations / steps can also be decomposed, combined, or partially merged, so the actual execution order may change according to the actual situation.

[0168] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. RPR rocker detection method integrating digital imaging and AI, characterized in that Including: Use an RPR shaker to uniformly shake and mix the serum sample, and adjust the shaking frequency and amplitude through a real-time monitoring and feedback system; Through a high-resolution CMOS / CCD digital camera or camera, real-time collect the image of the serum sample after being processed by the RPR shaker; Preprocess the collected serum sample image, including denoising, enhancement, segmentation and edge detection, and extract the key features in the serum sample; Based on a deep learning model, analyze the preprocessed serum sample image, infer whether the target substance is contained in the serum sample, and generate an analysis result; Compare the analysis result with historical data, and output a detection report including the concentration of the target substance, reaction mode and whether it is abnormal in the serum sample; Real-time feedback the detection report to the operation interface. If an abnormal reaction is found, automatically prompt relevant measures, and self-diagnose and perform performance checks on the operating status of the RPR shaker regularly or according to preset conditions.

2. The RPR rocker detection method for digital imaging and AI fusion according to claim 1, wherein: The use of an RPR shaker to uniformly shake and mix the serum sample, and adjust the shaking frequency and amplitude through a real-time monitoring and feedback system, includes: Start the RPR shaker, set the initial shaking frequency to 100 revolutions per minute and the amplitude to 10 degrees; During the shaking process, use a sensor to real-time collect the state data of the serum sample, including shaking frequency, amplitude, and sample temperature; During the shaking process, real-time monitor the particle precipitation in the serum sample. Once particle precipitation is found, immediately adjust the shaking frequency and amplitude; During the shaking process, according to the collected sample state data, remotely adjust the shaking frequency and amplitude through a wireless connection protocol.

3. The RPR rocker detection method for digital imaging and AI fusion according to claim 2, characterized in that: The real-time collection of the serum sample image after being processed by the RPR shaker through a high-resolution CMOS / CCD digital camera or camera, includes: Initialize the high-resolution CMOS / CCD digital camera or camera, and set the collection frequency to 30 frames per second; Configure the adaptive focusing function of the camera, and automatically adjust the focal length according to the distance and light conditions of the serum sample; Enable multi-spectral LED lighting, and select the best lighting spectrum according to the characteristics of the serum sample; Collect serum sample images from multiple angles and synthesize a panoramic image; During the collection process, real-time monitor the position change of the serum sample, and automatically adjust the camera focus to obtain serum sample images at different positions.

4. The RPR rocker detection method integrating digital imaging and AI according to claim 3, characterized in that: The preprocessing of the collected serum sample image, including denoising, enhancement, segmentation and edge detection, and extraction of the key features in the serum sample, includes: Use a deep convolutional denoising autoencoder algorithm to denoise the collected serum sample image and eliminate the noise in the image; Apply an adaptive histogram equalization algorithm to enhance the denoised image and optimize the contrast of the image; Adaptive histogram equalization algorithm formula: In the formula, Ν(x,y) is the neighborhood of the target pixel point (x,y), h(i,j) is the histogram value of the pixels in the neighborhood, and N is the neighborhood size; Combine an adaptive threshold-based segmentation algorithm with a semantic segmentation algorithm to segment and detect the edges of the enhanced image, and extract the boundary between the serum components and the background.

5. The RPR rocker detection method for digital imaging and AI fusion according to claim 4, wherein: The preprocessed serum sample images are analyzed based on a deep learning model to infer whether the target substance is contained in the serum sample and generate an analysis result, including: Construct a deep learning model, adopt a CNN+YOLO object detection architecture, and combine with an LSTM network to process temporal information to infer the dynamic changes of particles in the serum sample image; CNN output features: F CNN = CNN(I) YOLO object detection: D = YOLO(F CNN ) LSTM network: where I is the input image and h t is the hidden state of the LSTM network; The model is trained using a clinically annotated RPR shaker reaction image dataset, and rare type sample image data is synthesized through data augmentation techniques; Use the trained deep learning model to analyze the preprocessed serum sample images, and dynamically adjust and optimize the error of the analysis result by combining with a reinforcement learning algorithm.

6. The RPR rocking bed detection method integrating digital imaging and AI according to claim 5, characterized in that: The analysis result is compared with historical data, and a detection report including the concentration of the target substance, reaction pattern, and whether it is abnormal in the serum sample is output, including: The report is generated based on an AI model trained from historical detection data, automatically analyzes the concentration change trend of various substances in the serum, and visually displays it in a combined way of charts and images; Upload the detection report to the cloud for storage results, and provide a remote diagnosis function, upload data to the hospital LIS / HIS system in real time, and integrate it into the hospital laboratory automation system; The AI model adopts federated learning technology to perform model training and optimization among multiple medical institutions; Local model update and global aggregation formula: In the formula, is the model parameter of the k-th client, and w t+1 is the global model parameter.

7. The RPR rocking bed detection method integrating digital imaging and AI according to claim 6, characterized in that: Self-diagnosis and performance inspection of the RPR shaker operation status are performed regularly or according to preset conditions, including: When potential faults or performance degradation are found, immediately issue a warning to the operator for timely repair or replacement of components; Adopt an anomaly detection algorithm based on deep learning to monitor and warn the performance of key components of the RPR shaker in real time; Anomaly detection algorithm formula: where ξ z is the slack variable, s is the weight vector of the decision boundary, ρ is the bias term, and R is the total number of samples in the dataset.

8. The RPR rocker detection system integrating digital imaging and AI is characterized in that, Including: An RPR shaker for uniformly shaking and mixing serum samples, and adjusting the shaking frequency and amplitude through a real-time monitoring and feedback system; A high-resolution CMOS / CCD digital camera or camera for real-time collection of serum sample images processed by the RPR shaker; An image processing unit for preprocessing the collected serum sample images and extracting key features in the serum samples; A deep learning model for analyzing the preprocessed serum sample images to infer whether the target substance is contained in the serum sample and generate an analysis result; A report generation unit for comparing the analysis result with historical data and outputting a detection report including the concentration of the target substance, reaction pattern, and whether it is abnormal in the serum sample; A self-diagnosis and performance inspection unit for self-diagnosis and performance inspection of the RPR shaker operation status regularly or according to preset conditions.

9. The RPR rocker detection system integrating digital imaging and AI according to claim 8, characterized in that: The RPR shaker includes multiple independently controllable shaking units, and each shaking unit automatically adjusts the shaking frequency and amplitude according to the characteristics of the serum sample; The real-time monitoring and feedback system includes multiple sensors for real-time collection of serum sample status data, and introduces an adaptive control algorithm to adjust the shaking frequency and amplitude through real-time feedback data; Adaptive control algorithm formula: where \(u(t)\) is the control signal, \(e(t)\) is the error, \(M(t)\) is the time-varying gain, \(\alpha\) is the compensation coefficient, is the integral of the error from time 0 to the current time \(t\); The high-resolution CMOS / CCD digital camera or camera includes an adaptive focusing function, automatically adjusts the focal length according to the distance and light conditions of the serum sample, and enables multi-spectral LED illumination to select the optimal illumination spectrum according to the characteristics of the serum sample; The image processing unit uses the deep convolutional denoising autoencoder algorithm to denoise the acquired images, and applies the adaptive histogram equalization algorithm to enhance the denoised images; A generative adversarial network is introduced to enhance the image quality, and the formula of the generative adversarial network loss function is as follows: Where D(a) is the output probability of the discriminator for real data, G(r) is the output of the generator, Pdata(a) is the real data distribution, and P r (r) is the distribution of the latent space.

10. The RPR rocking bed detection system integrating digital imaging and AI according to claim 9, characterized in that: The deep learning model adopts the CNN+YOLO architecture, combines the long short-term memory network with temporal information, infers the dynamic changes of particles in the serum sample image, and dynamically adjusts and optimizes the error of the analysis result through the reinforcement learning algorithm; A graph neural network is introduced to optimize the spatial relationship and dynamic changes between different regions in the serum sample image; Graph neural network information propagation formula: where is the representation of node d at the (l + 1)-th layer, B(d) is the set of neighbor nodes of node d, A (l) and b (l) are the weights and biases at the l-th layer, and σ is the activation function; The report generation unit is based on the AI model trained from historical detection data, automatically analyzes the concentration change trend of various substances in the serum, and visually displays it in a way that combines charts and images; The self-diagnosis and performance inspection unit uses an anomaly detection algorithm based on deep learning to monitor and give early warnings about the performance of the key components of the RPR shaker in real time.

Images