Instrument of automatic immunofluorescence vaginal secretion detector

Abstract

The invention discloses a vagina micro-ecology automatic detection system and method based on immunofluorescence microscopic scanning and AI analysis. The system comprises a fluorescence imaging module, a motion control module, an image processing module and an AI intelligent analysis module. The motion control module is used for bearing a vaginal secretion slide sample and controlling the vaginal secretion slide sample to move in a three-dimensional space, and is matched with the fluorescence imaging module to carry out full-automatic scanning; the image processing module is used for enhancing and de-noising the collected original fluorescence image so as to improve the image quality; the AI analysis module performs multi-target recognition and quantitative analysis on the processed image based on a deep learning algorithm, and counts the number and proportion of various bacteria and cells; and finally, automatically generating a micro-ecology assessment report according to clinical rules. The full-flow automatic and intelligent detection of the vaginal secretion sample is realized, the problems of strong subjectivity and low efficiency of manual microscopic examination are effectively solved, and higher sensitivity, specificity and repeatability in vaginitis diagnosis are realized.

Description

Technical Field

[0001] This invention belongs to the field of vaginal secretion detection technology, specifically an automated immunofluorescence vaginal secretion detector. Background Technology

[0002] In the field of vaginal secretion testing, traditional methods mainly rely on direct observation under a microscope, which has many limitations. First, traditional observation methods can only provide morphological information of the sample under ordinary light, and cannot deeply analyze the characteristics of the sample under fluorescence excitation, limiting the dimensions and depth of detection. This leads to the omission of some potentially important information, affecting the accuracy of microbial identification and diagnosis.

[0003] Secondly, due to a lack of high-performance computing power, the equipment is inefficient when processing large amounts of fluorescence image data and complex algorithm calculations, resulting in long detection times and failing to meet the needs of rapid clinical diagnosis. Simultaneously, the equipment's operating environment is unstable and susceptible to external interference, leading to system crashes or data loss, reducing its safety and stability, and hindering the smooth progress of testing.

[0004] Furthermore, due to the lack of effective denoising, enhancement, and correction algorithms, the acquired fluorescence images are of low quality, suffering from noise interference, insufficient contrast, and geometric distortion, increasing the difficulty of subsequent analysis and reducing the accuracy of microbial identification. In addition, feature extraction methods mainly rely on manual methods, failing to fully explore the deep-seated features of microorganisms, resulting in insufficient accuracy and comprehensiveness in feature extraction.

[0005] Furthermore, existing methods for generating test reports also have shortcomings. Report content often relies on manual compilation, which is not only inefficient but also prone to inconsistencies and non-standardization due to human error. Simultaneously, the reports lack comprehensive information on patient symptoms and medical history, failing to provide doctors with complete diagnostic evidence and impacting the accuracy of diagnosis and the rationality of treatment plans. Summary of the Invention

[0006] In view of the above situation and to overcome the defects of the prior art, the present invention provides an automated immunofluorescence vaginal secretion detector to at least partially solve the above technical problems.

[0007] The technical solution adopted in this invention is as follows:

[0008] This invention proposes an automated immunofluorescence vaginal secretion analyzer, comprising:

[0009] Fluorescence Module: A fluorescence module is added to the original system. The fluorescence module can capture images of fluorescent slides to obtain image information of the sample under fluorescence excitation for subsequent image analysis. The fluorescence module uses a specific wavelength excitation light source to excite fluorescent substances in the sample to emit fluorescence. It is also equipped with a high-sensitivity image acquisition device to capture weak fluorescence signals and ensure that the captured fluorescence images have sufficient clarity and contrast, providing accurate basic data for subsequent microbial identification and diagnosis.

[0010] Built-in micro server: The original system's supporting computer and software algorithms are ported to a Linux-based micro server, and the server is built into the device. The micro server has high-performance computing power and a stable operating environment, which can quickly process large amounts of image data and complex algorithm calculations, improve the integration and portability of the device, and enhance the security and stability of the device, avoiding interference from external factors to the operation of the server.

[0011] Touchscreen operation interface: Access the server and provide an operation interface through another touchscreen. The touchscreen uses a high-resolution display and has good human-computer interaction performance. Operators can easily perform various operations on the equipment through the touchscreen, such as starting detection, setting parameters, and viewing detection results. The touchscreen is connected to the built-in micro server through a high-speed data interface to ensure that operation commands can be transmitted to the server in a timely and accurate manner and the server's processing results are displayed in real time.

[0012] The new algorithm system can acquire clearer fluorescence images and analyze them to distinguish epithelial cells, bacteria, fungi, trichomonas microorganisms. The algorithm system consists of multiple sub-algorithm modules, including image preprocessing algorithm, feature extraction algorithm, and classification and recognition algorithm. Through a series of processing and analysis of fluorescence images, it can accurately identify and classify different microorganisms.

[0013] Reporting System: The reporting system can output a clinical test report based on the above algorithm results. The reporting system can automatically obtain the analysis results from the algorithm system and generate a detailed clinical test report according to the preset templates and rules. The report includes sample information, test results, and diagnostic suggestions to facilitate doctors' diagnostic and treatment decisions.

[0014] A fluorescence imaging module is used to excite fluorescence signals from vaginal secretion samples mounted on a glass slide and capture fluorescence images;

[0015] A motion control module is used to carry and drive the glass slide to move precisely in three-dimensional space, so as to realize the fully automatic scanning of the glass slide by the fluorescence imaging module;

[0016] The image processing module is communicatively connected to the fluorescence imaging module and is used to enhance and denoise the captured raw fluorescence image to improve the image signal-to-noise ratio and clarity.

[0017] The AI ​​analysis module is communicatively connected to the image processing module. The AI ​​intelligent analysis module includes at least one processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it is used to perform multi-target recognition and quantitative analysis on the preprocessed image and generate a microecological assessment report according to clinical rules.

[0018] In one embodiment of the present invention, the fluorescence imaging module includes an LED excitation light source, a filter group of a specific wavelength, and a scientific-grade CCD or CMOS camera; the motion control module includes a high-precision XYZ three-axis electrically controlled translation stage and a stepper motor, with a positioning accuracy better than 5 micrometers;

[0019] The image processing module performs image preprocessing operations including a denoising algorithm, an enhancement algorithm, and a correction algorithm. The denoising algorithm uses a wavelet transform-based method to effectively remove noise interference in the fluorescence image and improve the signal-to-noise ratio. The enhancement algorithm uses histogram equalization and contrast stretching to enhance the contrast and brightness of the image and highlight useful information in the image. The correction algorithm is used to correct geometric distortion and color deviation in the image to ensure the accuracy and consistency of the image.

[0020] The AI ​​analysis module calls multiple trained deep learning models and image processing algorithms to perform collaborative analysis on the preprocessed fluorescence image. The collaborative analysis includes the identification and classification of targets in the image.

[0021] In one embodiment of the present invention, the plurality of trained deep learning models and image processing algorithms include:

[0022] The image quality assessment unit is configured to run a deep learning model based on a convolutional neural network to classify the fluorescence image into four quality levels: excellent, good, medium, and poor, so as to output image quality information.

[0023] The quantitative analysis unit is configured to receive filtered fluorescence images and, for epithelial cells and bacteria, execute image processing algorithms based on threshold segmentation and morphological priors to output connected components and precise quantity information of the target objects.

[0024] The target detection unit is configured to receive the filtered fluorescence image and run a deep learning model based on a convolutional neural network to output bounding box information containing leukocytes, leukocytes containing toxic granules, clue cells, basal cells, Candida spores and hyphae, and Trichomonas vaginalis.

[0025] In one embodiment of the present invention, the AI ​​analysis module further includes:

[0026] The statistical analysis unit, based on the output results of the quantitative analysis unit and the target detection unit, counts the number and relative proportion of epithelial cells, leukocytes, leukocytes containing toxic granules, Candida, clue cells, basal cells, and Trichomonas vaginalis.

[0027] The report generation unit is configured to receive the output of the statistical analysis unit, automatically calculate vaginal cleanliness, flora density, pathogen positivity, Nugent score and Donders score according to clinical rules, and generate a structured analysis report.

[0028] The image quality assessment unit executes a convolutional neural network based on the ResNet architecture;

[0029] The target detection unit executes a convolutional neural network based on the YOLO architecture.

[0030] In one embodiment of the invention, the quantitative analysis unit is configured to perform quantitative analysis of epithelial cells according to the following steps:

[0031] Gaussian blur is applied to the fluorescence image to reduce noise, resulting in a preprocessed image.

[0032] Threshold segmentation is applied to the preprocessed image to separate cell clusters from the background, background markers are generated, and an adaptive threshold is applied to the cell cluster region to obtain candidate nuclear channels and cytoplasmic channels.

[0033] Connectivity analysis was performed on the candidate cell nuclear channels to remove impurities with too small an area. The best-fit ellipse was calculated, and the ellipticity index was calculated based on the ellipse parameters. Candidate nuclei with excessive deviation were removed based on the threshold, and the remaining candidate nuclei were used as cell nuclear markers.

[0034] The gradient operator is used to calculate the gradients in the X and Y directions of the binary image of the cytoplasmic channels to obtain a gradient magnitude image;

[0035] Using the background markers, cell nucleus markers, and gradient magnitude images, the watershed algorithm is applied to obtain the connected components of a single cell and output the epithelial cell count information.

[0036] The quantitative analysis unit is configured to perform quantitative analysis of bacteria according to the following steps:

[0037] Gaussian blur is applied to the fluorescence image to reduce noise, resulting in a preprocessed image.

[0038] Threshold segmentation is applied to the preprocessed image to separate the foreground bacteria from the background, generating a binary image;

[0039] Connectivity analysis is performed on the binary image. Based on preset area and shape priors, the connected components are classified into bacilli, cocci, and impurities to obtain the number and relative proportion of bacteria.

[0040] In one embodiment of the invention, the statistical analysis unit is configured to count the number of cells / bacteria according to the following steps:

[0041] Acquire image quality information and cell / bacterial quantity information;

[0042] Calculate the global weighted average quantity E:

[0043]

[0044] Where i is the i-th preprocessed image, N is the total number of images contained in the slide, Ci is the number of targets detected in the i-th image, Qi is the quality category of the i-th image, and W(Qi) is the weight corresponding to the quality category.

[0045] The weighted average is compared with the clinical standard number of class thresholds to determine the number of target cells or bacteria.

[0046] The statistical analysis unit is also configured to statistically analyze the proportions of pathogenic microorganisms and cells according to the following steps:

[0047] Obtain bounding box information from the target detection unit and count the number of white blood cells, white blood cells containing toxic particles, and clue cells detected in each image;

[0048] For Candida spores, hyphae, and Trichomonas vaginalis, the confidence level of each bounding box of the target pathogen is compared with a preset threshold. When the confidence level exceeds the threshold, the pathogen is recorded as positive.

[0049] Calculate the proportion of white blood cells containing toxic particles (T):

[0050]

[0051] Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ti is the number of toxic granular white blood cells in the i-th image, and li is the number of white blood cells in the i-th image.

[0052] The number of epithelial cells is obtained from the quantitative analysis unit, and the proportion of basal cells J is calculated.

[0053]

[0054] Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ji is the number of basal cells in the i-th image, and ei is the number of epithelial cells in the i-th image;

[0055] The report generation unit is configured to obtain bacterial quantity level information, pathogenic microorganism positive information and cell ratio information from the statistical analysis unit, and automatically calculate at least one of vaginal cleanliness, flora density, Nugent score, Donders score and pathogen positive judgment according to clinical rules.

[0056] In one embodiment of the present invention, the excitation light source in the fluorescence module is a laser diode of a specific wavelength, and the image acquisition device is a high-resolution CCD or CMOS sensor, which can capture the subtle features of the fluorescence image and improve the image clarity and resolution.

[0057] In one embodiment of the present invention, the server runs a customized Linux operating system, which is optimized to efficiently run software algorithms, process image data, and ensure system stability and security. The server is also equipped with data backup and recovery functions to prevent data loss.

[0058] In one embodiment of the present invention, the touch screen is connected to the built-in micro server via a USB 3.0 or Ethernet interface to ensure that operation commands and detection results can be transmitted quickly and accurately. The touch screen's user interface is simple and intuitive, providing a user-friendly experience and making it easy for operators to get started quickly.

[0059] In one embodiment of the present invention, the image preprocessing algorithm in the novel algorithm system includes a denoising algorithm, an enhancement algorithm, and a correction algorithm. The denoising algorithm uses a wavelet transform-based method to effectively remove noise interference in the fluorescence image and improve the signal-to-noise ratio of the image. The enhancement algorithm uses histogram equalization and contrast stretching methods to enhance the contrast and brightness of the image and highlight useful information in the image. The correction algorithm is used to correct the geometric distortion and color deviation of the image to ensure the accuracy and consistency of the image.

[0060] In one embodiment of the present invention, the feature extraction algorithm in the novel algorithm system adopts a deep learning method to construct a deep convolutional neural network (CNN) model, which is trained with a large amount of fluorescence image data to learn the morphological and texture features of different microorganisms. During the training process, data augmentation techniques, such as rotation, flipping, and scaling, are used to expand the training dataset and improve the generalization ability of the model. After training, the CNN model can automatically extract effective features from the fluorescence images for subsequent classification and recognition.

[0061] In one embodiment of the present invention, the fluorescence focusing method in the novel algorithm system innovatively adopts the following steps: first focusing with white light, then calculating the focus point of the fluorescence using the phase difference between white light and fluorescence, and finally scanning with fluorescence.

[0062] White light focusing: The sample is illuminated by a white light source, the white light image is acquired by an image acquisition device, and the white light focusing position is quickly determined by an autofocus algorithm (such as phase detection focusing or contrast detection focusing). White light focusing has the characteristics of high brightness and short exposure time, and will not quench the sample by high-energy excitation light. It can quickly and accurately determine the focusing position.

[0063] Phase difference calculation: By analyzing the phase difference between white light and fluorescence, the correct focus point for fluorescence is calculated. During the device development process, numerous calibration experiments are conducted beforehand to obtain the conversion relationship between fluorescence and white light, establishing a mathematical model between phase difference and focus position. Based on this model, the focus position for fluorescence can be accurately calculated from the focus position for white light.

[0064] Fluorescence scanning: The focus position is adjusted to the calculated fluorescence focus position, the sample is illuminated with a fluorescence light source, and a fluorescence image is scanned. This avoids the problem of lower light intensity and darker areas in the focused area caused by taking many pictures while the focus process is constantly lit when using fluorescence focus. This is beneficial for clinical observation and reduces misjudgment.

[0065] The beneficial effects of the technical solution of this invention are as follows:

[0066] This invention adds a fluorescence module to the original system, enabling the acquisition of fluorescent slide images to obtain sample image information under fluorescence excitation. This provides a foundation for subsequent image analysis, allowing detection to move beyond traditional observation methods and enabling in-depth analysis of samples from the perspective of fluorescence characteristics. This significantly expands the detection dimensions, helps discover more potential information, and creates conditions for more accurate microbial identification and diagnosis. Using a specific wavelength excitation light source, fluorescent substances in the sample can be excited to emit fluorescence. Targeted excitation maximizes the excitation of target fluorescent substances, reduces non-target interference, and improves the specificity of the fluorescence signal. This allows the acquired fluorescence images to more accurately reflect the fluorescence characteristics of specific microorganisms in the sample, providing a reliable basis for accurate microbial identification.

[0067] This invention portes the original system's accompanying computer and software algorithms to a Linux-based microserver, embedding the server internally within the device. The microserver possesses high-performance computing capabilities, enabling rapid processing of large amounts of image data and complex algorithm calculations. In immunofluorescence secretion detection, which involves the analysis and processing of large amounts of fluorescence image data, high-performance computing significantly shortens processing time and improves detection efficiency. A stable operating environment ensures long-term stable server operation, preventing system crashes or data loss due to external interference, enhancing the device's security and stability, and ensuring the smooth progress of detection work.

[0068] This invention utilizes an algorithm system to acquire clearer fluorescence images, providing high-quality data for subsequent analysis. Clear fluorescence images more accurately reveal the morphology and characteristics of microorganisms in the sample, reducing misjudgments caused by image blur and improving the accuracy of microbial identification. The algorithm system can analyze fluorescence images to distinguish between epithelial cells, bacteria, fungi, and trichomonas microorganisms. Accurate differentiation of different microorganisms provides doctors with more detailed test results, helping them to more accurately diagnose conditions and develop targeted treatment plans. The algorithm system consists of multiple sub-algorithm modules, including image preprocessing algorithms, feature extraction algorithms, and classification and recognition algorithms. Modularization makes the algorithm system more flexible and scalable; each sub-algorithm module can be optimized for specific tasks, improving the overall performance of the algorithm system. Simultaneously, modularity facilitates algorithm maintenance and upgrades; when a sub-algorithm module needs improvement, it can be modified individually without affecting the normal operation of other modules.

[0069] This invention employs image preprocessing algorithms, including denoising, enhancement, and correction algorithms. The denoising algorithm, based on wavelet transform, effectively removes noise interference from fluorescence images, improving the signal-to-noise ratio and highlighting useful information. The enhancement algorithm uses histogram equalization and contrast stretching to enhance image contrast and brightness, further emphasizing useful information for subsequent analysis. The correction algorithm corrects geometric distortion and color deviation, ensuring image accuracy and consistency, providing a reliable guarantee for accurate microbial identification. The feature extraction algorithm utilizes deep learning, constructing a deep convolutional neural network (CNN) model and training it with a large amount of fluorescence image data to learn the morphological and textural features of different microorganisms. Deep learning has powerful feature learning capabilities, automatically extracting effective features from large amounts of data. Compared to traditional manual feature extraction methods, it can better uncover deeper features of microorganisms, improving the accuracy and comprehensiveness of feature extraction. Data augmentation techniques, such as rotation, flipping, and scaling, are used during training to expand the training dataset, improving the model's generalization ability and enabling it to achieve better recognition results in different scenarios.

[0070] This invention provides a reporting system that can output a clinical laboratory report based on the aforementioned algorithm results. It automatically retrieves analysis results from the algorithm system and generates a detailed clinical laboratory report according to preset templates and rules. The report includes sample information, test results, and diagnostic recommendations, facilitating doctors' diagnostic and treatment decisions. The automatic report generation function improves work efficiency, reduces the time and workload of manual report writing, and ensures the standardization and consistency of the reports. The reporting system also allows doctors to manually input other relevant information, such as patient symptoms and medical history, and integrates this information into the report. This makes the report richer and more comprehensive, providing doctors with more complete diagnostic evidence, helping them to more accurately judge the condition and formulate more reasonable treatment plans.

[0071] This invention illuminates the sample with a white light source, acquires a white light image using an image acquisition device, and employs an autofocus algorithm to quickly determine the white light focus position. White light focusing features high brightness and short exposure time, preventing the sample from being quenched by high-energy excitation light, and enabling rapid and accurate focus position determination. Performing white light focusing before fluorescence focusing avoids damage to the sample caused by directly using fluorescence focusing, while also quickly determining the focus position, providing a foundation for subsequent fluorescence focusing. By analyzing the phase difference between white light and fluorescence, the correct focus point for fluorescence is calculated. Extensive calibration experiments are conducted beforehand to obtain the conversion relationship between fluorescence and white light, establishing a mathematical model between phase difference and focus position. Based on this model, the fluorescence focus position can be accurately calculated from the white light focus position. The phase difference calculation method improves the accuracy of fluorescence focusing, avoids errors inherent in directly using fluorescence focusing, and ensures clear images are obtained during fluorescence scanning.

[0072] Adjust the focus position to the calculated fluorescence focus position, illuminate the sample with a fluorescence light source, and perform fluorescence image scanning. This avoids the problem of lower light intensity and darker areas in the focused area caused by the need to take many pictures under constant light during the focusing process when using fluorescence focusing directly. This is beneficial for clinical observation and reduces misdiagnosis.

[0073] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0074] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0075] Figure 1 This is a schematic diagram of the structure of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention;

[0076] Figure 2This is a front view of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention;

[0077] Figure 3 This is a right view of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention;

[0078] Figure 4 This is a top view of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention;

[0079] Figure 5 This is a left view of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention;

[0080] Figure 6 This is a diagram of the internal modules of the automated immunofluorescence vaginal secretion detector proposed in an embodiment of the present invention. Detailed Implementation

[0081] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0082] The instrument of the automated immunofluorescence vaginal secretion detector according to an embodiment of the present invention will now be described with reference to the accompanying drawings.

[0083] like Figures 1 to 6 As shown, this embodiment of the invention provides an automated immunofluorescence vaginal secretion detector, comprising: a fluorescence module: a fluorescence module is added to the original system, which can capture images of fluorescent slides to obtain image information of the sample under fluorescence excitation for subsequent image analysis. The fluorescence module uses a specific wavelength excitation light source to excite fluorescent substances in the sample to emit fluorescence, and is equipped with a high-sensitivity image acquisition device to capture weak fluorescence signals, ensuring that the captured fluorescence images have sufficient clarity and contrast, providing accurate basic data for subsequent microbial identification and diagnosis;

[0084] Built-in micro server: The original system's supporting computer and software algorithms are ported to a Linux-based micro server, and the server is built into the device. The micro server has high-performance computing power and a stable operating environment, which can quickly process large amounts of image data and complex algorithm calculations, improve the integration and portability of the device, and enhance the security and stability of the device, avoiding interference from external factors to the operation of the server.

[0085] Touchscreen operation interface: Access the server and provide an operation interface through another touchscreen. The touchscreen uses a high-resolution display and has good human-computer interaction performance. Operators can easily perform various operations on the equipment through the touchscreen, such as starting detection, setting parameters, and viewing detection results. The touchscreen is connected to the built-in micro server through a high-speed data interface to ensure that operation commands can be transmitted to the server in a timely and accurate manner and the server's processing results are displayed in real time.

[0086] The new algorithm system can acquire clearer fluorescence images and analyze them to distinguish epithelial cells, bacteria, fungi, trichomonas microorganisms. The algorithm system consists of multiple sub-algorithm modules, including image preprocessing algorithm, feature extraction algorithm, and classification and recognition algorithm. Through a series of processing and analysis of fluorescence images, it can accurately identify and classify different microorganisms.

[0087] Reporting System: The reporting system can output a clinical test report based on the above algorithm results. The reporting system can automatically obtain the analysis results from the algorithm system and generate a detailed clinical test report according to the preset templates and rules. The report includes sample information, test results, and diagnostic suggestions to facilitate doctors' diagnostic and treatment decisions.

[0088] In one possible implementation, the excitation light source in the fluorescence module is a laser diode of a specific wavelength, and the image acquisition device is a high-resolution CCD or CMOS sensor, which can capture the subtle features of the fluorescence image and improve the image clarity and resolution.

[0089] In one possible implementation, the server runs a customized Linux operating system that is optimized to efficiently run software algorithms, process image data, and ensure system stability and security. The server is also equipped with data backup and recovery functions to prevent data loss.

[0090] In one possible implementation, the touchscreen is connected to the built-in microserver via a USB 3.0 or Ethernet interface to ensure that operation commands and detection results can be transmitted quickly and accurately. The touchscreen's user interface is simple and intuitive, providing a user-friendly experience and making it easy for operators to get started quickly.

[0091] In one possible implementation, the image preprocessing algorithm in the novel algorithm system includes a denoising algorithm, an enhancement algorithm, and a correction algorithm. The denoising algorithm uses a wavelet transform-based method to effectively remove noise interference in the fluorescence image and improve the signal-to-noise ratio of the image. The enhancement algorithm uses histogram equalization and contrast stretching methods to enhance the contrast and brightness of the image and highlight useful information in the image. The correction algorithm is used to correct the geometric distortion and color deviation of the image to ensure the accuracy and consistency of the image.

[0092] In one possible implementation, the feature extraction algorithm in the novel algorithm system employs deep learning to construct a deep convolutional neural network (CNN) model. This model is trained using a large amount of fluorescence image data to learn the morphological and textural features of different microorganisms. During training, data augmentation techniques, such as rotation, flipping, and scaling, are used to expand the training dataset and improve the model's generalization ability. The trained CNN model can automatically extract effective features from fluorescence images for subsequent classification and recognition.

[0093] In one possible implementation, the fluorescence focusing method in the novel algorithm system innovatively employs a method of first focusing with white light, then calculating the appropriate focus point for the fluorescence using the phase difference between the white light and the fluorescence, and finally scanning with the fluorescence. The specific steps are as follows:

[0094] White light focusing: The sample is illuminated by a white light source, the white light image is acquired by an image acquisition device, and the white light focusing position is quickly determined by an autofocus algorithm (such as phase detection focusing or contrast detection focusing). White light focusing has the characteristics of high brightness and short exposure time, and will not quench the sample by high-energy excitation light. It can quickly and accurately determine the focusing position.

[0095] Phase difference calculation: By analyzing the phase difference between white light and fluorescence, the correct focus point for fluorescence is calculated. During the device development process, numerous calibration experiments are conducted beforehand to obtain the conversion relationship between fluorescence and white light, establishing a mathematical model between phase difference and focus position. Based on this model, the focus position for fluorescence can be accurately calculated from the focus position for white light.

[0096] Fluorescence scanning: The focus position is adjusted to the calculated fluorescence focus position, the sample is illuminated with a fluorescence light source, and a fluorescence image is scanned. This avoids the problem of lower light intensity and darker areas in the focused area caused by taking many pictures while the focus process is constantly lit when using fluorescence focus. This is beneficial for clinical observation and reduces misjudgment.

[0097] In specific applications, this invention adds a fluorescence module to the original system. This module uses a specific wavelength excitation light source to excite fluorescent substances in the sample, and a high-sensitivity image acquisition device (such as a high-resolution CCD or CMOS sensor) captures the weak fluorescence signal, thereby obtaining image information of the sample under fluorescence excitation. This module is fundamental to the entire detector's acquisition of key detection data. The specific wavelength excitation light source can specifically excite fluorescent substances in the sample, and the high-sensitivity image acquisition device ensures that even very weak fluorescence signals can be accurately captured, providing high-quality raw data for subsequent image analysis and helping to improve the accuracy of microbial identification and diagnosis.

[0098] The original system's accompanying computer and software algorithms were ported to a Linux-based microserver and integrated into the device. The server runs a customized Linux operating system, optimized for efficient execution of software algorithms and processing of image data. It also features data backup and recovery capabilities to ensure system stability and security. The built-in microserver integrates the device, improving its portability. Its high-performance computing capabilities and stable operating environment enable the server to quickly process large amounts of image data and complex algorithm calculations, preventing external interference and ensuring smooth detection and data security.

[0099] The server is accessed and an operating interface is provided via another touchscreen. The touchscreen features a high-resolution display with excellent human-computer interaction. The touchscreen connects to the built-in microserver via USB 3.0 or Ethernet. The user interface is simple and intuitive, allowing operators to easily perform various operations on the device, such as starting detection, setting parameters, and viewing test results. The touchscreen interface provides operators with a convenient and intuitive operating method, making the device operation simpler and easier to understand. A high-speed data interface ensures that operation commands are transmitted to the server promptly and accurately, and the server's processing results are displayed in real time, improving the device's usability and work efficiency.

[0100] The reporting system outputs a clinical laboratory report based on algorithmic results. It automatically retrieves analysis results from the algorithm and generates a detailed report according to preset templates and rules, including sample information, test results, and diagnostic recommendations. The system also allows doctors to manually input other relevant information, such as patient symptoms and medical history, and integrates this information into the report. The report supports multiple output formats, including printed paper reports, electronic reports, and submission to the Hospital Information System (HIS). The system presents test results to doctors in an intuitive and detailed manner, facilitating diagnostic and treatment decisions. The automatic report generation improves work efficiency, while the support for manual input of other relevant information and multiple output formats meets the needs of different doctors, making it convenient for doctors to review and archive reports.

[0101] The method employs a technique where white light is first used for focusing, then the phase difference between white light and fluorescence is used to calculate the correct focus point for fluorescence, followed by fluorescence scanning. The specific steps are as follows: White light focusing: The sample is illuminated with white light, an image is acquired using an image acquisition device, and an autofocus algorithm is used to quickly determine the white light focus position; Phase difference calculation: By analyzing the phase difference between white light and fluorescence, the correct focus point for fluorescence is calculated; Fluorescence scanning: The focus position is adjusted to the calculated fluorescence focus position, the sample is illuminated with a fluorescent light source, and a fluorescence image is scanned. This fluorescence focusing method avoids the problem of lower light intensity and darker areas in the focused region caused by the need for continuous illumination and numerous image captures during the focusing process when using direct fluorescence focusing. This is beneficial for clinical observation, reduces misjudgment, and improves the quality of fluorescence images and the accuracy of detection.

[0102] When the operator starts the detection via the touchscreen interface, the equipment begins to operate. First, a specific wavelength laser diode in the fluorescence module acts as an excitation source, illuminating the sample and exciting the fluorescent substances in the sample to emit fluorescence. Simultaneously, a high-resolution CCD or CMOS sensor acts as an image acquisition device, capturing the weak fluorescence signal and acquiring image information of the sample under fluorescence excitation.

[0103] In the image acquisition and algorithm analysis process, the innovative fluorescence focusing method plays a crucial role. First, the sample is illuminated with white light, and an image acquisition device captures the white light image. An autofocus algorithm then quickly determines the white light focus position. Next, by analyzing the phase difference between the white light and fluorescence, the correct focus point for the fluorescence is calculated. Finally, the focus position is adjusted to the calculated fluorescence focus position, and the sample is illuminated with a fluorescent light source. A fluorescence image scan is then performed to ensure a clear and accurate fluorescence image is obtained.

[0104] Based on algorithmic analysis, the reporting system provides detailed descriptions of the quantity and morphological characteristics of different microorganisms, and offers preliminary diagnostic suggestions in conjunction with clinical diagnostic criteria. Simultaneously, the system allows doctors to manually input other relevant information, such as patient symptoms and medical history, which is then integrated into the report. The report can be output in multiple formats, including printed paper reports, electronic reports generated and uploaded to the Hospital Information System (HIS), facilitating doctor access and archiving.

[0105] In one possible implementation, a fluorescence imaging module is used to excite a fluorescence signal on a vaginal secretion sample carried on a glass slide and capture a fluorescence image;

[0106] The motion control module is used to carry and drive the glass slide to move precisely in three-dimensional space, so as to realize the fully automatic scanning of the glass slide by the fluorescence imaging module;

[0107] The image processing module, which communicates with the fluorescence imaging module, is used to enhance and denoise the captured raw fluorescence image to improve the image signal-to-noise ratio and clarity.

[0108] The AI ​​analysis module communicates with the image processing module. The AI ​​intelligent analysis module includes at least one processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it is used to perform multi-target recognition and quantitative analysis on the preprocessed image and generate a microecological assessment report according to clinical rules.

[0109] The fluorescence imaging module includes an LED excitation light source, a filter group for specific wavelengths, and a scientific-grade CCD or CMOS camera; the motion control module includes a high-precision XYZ three-axis electrically controlled translation stage and a stepper motor, with a positioning accuracy better than 5 micrometers.

[0110] The image preprocessing operations performed by the image processing module include denoising, enhancement, and correction algorithms. The denoising algorithm uses a wavelet transform-based method to effectively remove noise interference in fluorescence images and improve the signal-to-noise ratio. The enhancement algorithm uses histogram equalization and contrast stretching to enhance the contrast and brightness of the image and highlight useful information. The correction algorithm is used to correct geometric distortion and color deviation in the image to ensure the accuracy and consistency of the image.

[0111] The AI ​​analysis module calls multiple trained deep learning models and image processing algorithms to perform collaborative analysis on the preprocessed fluorescence images. The collaborative analysis includes the identification and classification of targets in the images.

[0112] In one possible implementation, multiple trained deep learning models and image processing algorithms include: an image quality assessment unit configured to run a deep learning model based on a convolutional neural network to classify fluorescence images into four quality levels: excellent, good, medium, and poor, in order to output image quality information;

[0113] The quantitative analysis unit is configured to receive filtered fluorescence images and, for epithelial cells and bacteria, execute image processing algorithms based on threshold segmentation and morphological priors to output connected components and precise quantity information of the target objects.

[0114] The target detection unit is configured to receive the filtered fluorescence image and run a deep learning model based on a convolutional neural network to output bounding box information containing leukocytes, leukocytes containing toxic granules, clue cells, basal cells, Candida spores and hyphae, and Trichomonas vaginalis.

[0115] In one possible implementation, the AI ​​analysis module further includes: a statistical analysis unit, which, based on the output results of the quantitative analysis unit and the target detection unit, counts the number and relative proportion of epithelial cells, leukocytes, leukocytes containing toxic granules, Candida, clue cells, basal cells, and Trichomonas vaginalis.

[0116] The report generation unit is configured to receive the output of the statistical analysis unit, automatically calculate vaginal cleanliness, flora density, pathogen positivity, Nugent score and Donders score according to clinical rules, and generate a structured analysis report.

[0117] The image quality assessment unit executes a convolutional neural network based on the ResNet architecture;

[0118] The object detection unit executes a convolutional neural network based on the YOLO architecture.

[0119] In one possible implementation, the quantitative analysis unit is configured to perform quantitative analysis of epithelial cells by the following steps: applying Gaussian blur to the fluorescence image to reduce noise, thereby obtaining a preprocessed image;

[0120] Threshold segmentation is applied to the preprocessed image to separate cell clusters from the background and generate background markers. An adaptive threshold is applied to the cell cluster region to obtain candidate nuclear channels and cytoplasmic channels.

[0121] Connectivity analysis was performed on candidate nuclear channels to remove impurities with too small an area. The best-fit ellipse was calculated, and the ellipticity index was calculated based on the ellipse parameters. Candidate nuclei with excessive deviation were removed based on the threshold, and the remaining candidate nuclei were used as nuclear markers.

[0122] The gradient operator is used to calculate the gradients in the X and Y directions of the binary map of the cytoplasmic channels to obtain the gradient magnitude image;

[0123] Using background markers, cell nucleus markers, and gradient magnitude images, the watershed algorithm is applied to obtain the connected components of a single cell and output the epithelial cell count information.

[0124] The quantitative analysis unit is configured to perform quantitative analysis of bacteria according to the following steps:

[0125] Gaussian blur is applied to the fluorescence image to reduce noise, resulting in a preprocessed image.

[0126] Threshold segmentation is applied to the preprocessed image to separate the foreground bacteria from the background, generating a binary image.

[0127] Connectivity analysis is performed on binary images. Based on preset area and shape priors, connected components are classified into bacilli, cocci, and impurities to obtain the number and relative proportion of bacteria.

[0128] In one possible implementation, the statistical analysis unit is configured to count the number of cells / bacteria according to the following steps:

[0129] Acquire image quality information and cell / bacterial quantity information;

[0130] Calculate the global weighted average quantity E:

[0131]

[0132] Where i is the i-th preprocessed image, N is the total number of images contained in the slide, Ci is the number of targets detected in the i-th image, Qi is the quality category of the i-th image, and W(Qi) is the weight corresponding to the quality category.

[0133] The weighted average is compared with the clinical standard number of class thresholds to determine the number of target cells or bacteria.

[0134] The statistical analysis unit is also configured to perform statistical analysis on the proportions of pathogenic microorganisms and cells in the following steps:

[0135] Obtain bounding box information from the target detection unit and count the number of white blood cells, white blood cells containing toxic particles, and clue cells detected in each image;

[0136] For Candida spores, hyphae, and Trichomonas vaginalis, the confidence level of each bounding box of the target pathogen is compared with a preset threshold. When the confidence level exceeds the threshold, the pathogen is recorded as positive.

[0137] Calculate the proportion of white blood cells containing toxic particles (T):

[0138]

[0139] Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ti is the number of toxic granular white blood cells in the i-th image, and li is the number of white blood cells in the i-th image.

[0140] The number of epithelial cells was obtained from the quantitative analysis unit, and the proportion of basal cells J was calculated.

[0141]

[0142] Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ji is the number of basal cells in the i-th image, and ei is the number of epithelial cells in the i-th image;

[0143] The report generation unit is configured to obtain bacterial count information, pathogenic microorganism positive information, and cell ratio information from the statistical analysis unit, and automatically calculate at least one of vaginal cleanliness, flora density, Nugent score, Donders score, and pathogen positive judgment according to clinical rules.

[0144] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0145] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An automated immunofluorescence vaginal secretion analyzer, characterized in that, include: Fluorescence Module: A fluorescence module is added to the original system. The fluorescence module can capture images of fluorescent slides to obtain image information of the sample under fluorescence excitation for subsequent image analysis. The fluorescence module uses a specific wavelength excitation light source to excite fluorescent substances in the sample to emit fluorescence. It is also equipped with a high-sensitivity image acquisition device to capture weak fluorescence signals and ensure that the captured fluorescence images have sufficient clarity and contrast, providing accurate basic data for subsequent microbial identification and diagnosis. Built-in micro server: The original system's supporting computer and software algorithms are ported to a Linux-based micro server, and the server is built into the device. The micro server has high-performance computing power and a stable operating environment, which can quickly process large amounts of image data and complex algorithm calculations, improve the integration and portability of the device, and enhance the security and stability of the device, avoiding interference from external factors to the operation of the server. Touchscreen operation interface: Access the server and provide an operation interface through another touchscreen. The touchscreen uses a high-resolution display and has good human-computer interaction performance. Operators can easily perform various operations on the equipment through the touchscreen, such as starting detection, setting parameters, and viewing detection results. The touchscreen is connected to the built-in micro server through a high-speed data interface to ensure that operation commands can be transmitted to the server in a timely and accurate manner and the server's processing results are displayed in real time. The new algorithm system can acquire clearer fluorescence images and analyze them to distinguish epithelial cells, bacteria, fungi, trichomonas microorganisms. The algorithm system consists of multiple sub-algorithm modules, including image preprocessing algorithm, feature extraction algorithm, and classification and recognition algorithm. Through a series of processing and analysis of fluorescence images, it can accurately identify and classify different microorganisms. Reporting System: The reporting system can output a clinical test report based on the above algorithm results. The reporting system can automatically obtain the analysis results from the algorithm system and generate a detailed clinical test report according to the preset templates and rules. The report includes sample information, test results, and diagnostic suggestions to facilitate doctors' diagnostic and treatment decisions. The excitation light source in the fluorescence module uses a laser diode of a specific wavelength, and the image acquisition device uses a high-resolution CCD or CMOS sensor, which can capture the subtle features of the fluorescence image and improve the image clarity and resolution. The server runs a customized Linux operating system, which is optimized to efficiently run software algorithms, process image data, and ensure system stability and security. The server is also equipped with data backup and recovery functions to prevent data loss.

2. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, Also includes: A fluorescence imaging module is used to excite fluorescence signals from vaginal secretion samples mounted on a glass slide and capture fluorescence images; A motion control module is used to carry and drive the glass slide to move precisely in three-dimensional space, so as to realize the fully automatic scanning of the glass slide by the fluorescence imaging module; The image processing module is communicatively connected to the fluorescence imaging module and is used to enhance and denoise the captured raw fluorescence image to improve the image signal-to-noise ratio and clarity. The AI ​​analysis module is communicatively connected to the image processing module. The AI ​​intelligent analysis module includes at least one processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it is used to perform multi-target recognition and quantitative analysis on the preprocessed image and generate a microecological assessment report according to clinical rules.

3. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The fluorescence imaging module includes an LED excitation light source, a filter group of specific wavelengths, and a scientific-grade CCD or CMOS camera; the motion control module includes a high-precision XYZ three-axis electrically controlled translation stage and a stepper motor, with a positioning accuracy better than 5 micrometers. The image processing module performs image preprocessing operations including a denoising algorithm, an enhancement algorithm, and a correction algorithm. The denoising algorithm uses a wavelet transform-based method to effectively remove noise interference in the fluorescence image and improve the signal-to-noise ratio. The enhancement algorithm uses histogram equalization and contrast stretching to enhance the contrast and brightness of the image and highlight useful information in the image. The correction algorithm is used to correct geometric distortion and color deviation in the image to ensure the accuracy and consistency of the image. The AI ​​analysis module calls multiple trained deep learning models and image processing algorithms to perform collaborative analysis on the preprocessed fluorescence image. The collaborative analysis includes the identification and classification of targets in the image.

4. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The trained deep learning models and image processing algorithms include: The image quality assessment unit is configured to run a deep learning model based on a convolutional neural network to classify the fluorescence image into four quality levels: excellent, good, medium, and poor, so as to output image quality information. The quantitative analysis unit is configured to receive filtered fluorescence images and, for epithelial cells and bacteria, execute image processing algorithms based on threshold segmentation and morphological priors to output connected components and precise quantity information of the target objects. The target detection unit is configured to receive the filtered fluorescence image and run a deep learning model based on a convolutional neural network to output bounding box information containing leukocytes, leukocytes containing toxic granules, clue cells, basal cells, Candida spores and hyphae, and Trichomonas vaginalis.

5. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The AI ​​analysis module also includes: The statistical analysis unit, based on the output results of the quantitative analysis unit and the target detection unit, counts the number and relative proportion of epithelial cells, leukocytes, leukocytes containing toxic granules, Candida, clue cells, basal cells, and Trichomonas vaginalis. The report generation unit is configured to receive the output of the statistical analysis unit, automatically calculate vaginal cleanliness, flora density, pathogen positivity, Nugent score and Donders score according to clinical rules, and generate a structured analysis report. The image quality assessment unit executes a convolutional neural network based on the ResNet architecture; The target detection unit executes a convolutional neural network based on the YOLO architecture.

6. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The quantitative analysis unit is configured to perform quantitative analysis of epithelial cells according to the following steps: Gaussian blur is applied to the fluorescence image to reduce noise, resulting in a preprocessed image. Threshold segmentation is applied to the preprocessed image to separate cell clusters from the background, background markers are generated, and an adaptive threshold is applied to the cell cluster region to obtain candidate nuclear channels and cytoplasmic channels. Connectivity analysis was performed on the candidate cell nuclear channels to remove impurities with too small an area. The best-fit ellipse was calculated, and the ellipticity index was calculated based on the ellipse parameters. Candidate nuclei with excessive deviation were removed based on the threshold, and the remaining candidate nuclei were used as cell nuclear markers. The gradient operator is used to calculate the gradients in the X and Y directions of the binary image of the cytoplasmic channels to obtain a gradient magnitude image; Using the background markers, cell nucleus markers, and gradient magnitude images, the watershed algorithm is applied to obtain the connected components of a single cell and output the epithelial cell count information. The quantitative analysis unit is configured to perform quantitative analysis of bacteria according to the following steps: Gaussian blur is applied to the fluorescence image to reduce noise, resulting in a preprocessed image. Threshold segmentation is applied to the preprocessed image to separate the foreground bacteria from the background, generating a binary image; Connectivity analysis is performed on the binary image. Based on preset area and shape priors, the connected components are classified into bacilli, cocci, and impurities to obtain the number and relative proportion of bacteria.

7. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The statistical analysis unit is configured to count the number of cells / bacteria according to the following steps: Acquire image quality information and cell / bacterial quantity information; Calculate the global weighted average quantity E: Where i is the i-th preprocessed image, N is the total number of images contained in the slide, Ci is the number of targets detected in the i-th image, Qi is the quality category of the i-th image, and W(Qi) is the weight corresponding to the quality category. The weighted average is compared with the clinical standard number of class thresholds to determine the number of target cells or bacteria. The statistical analysis unit is also configured to statistically analyze the proportions of pathogenic microorganisms and cells according to the following steps: Obtain bounding box information from the target detection unit and count the number of white blood cells, white blood cells containing toxic particles, and clue cells detected in each image; For Candida spores, hyphae, and Trichomonas vaginalis, the confidence level of each bounding box of the target pathogen is compared with a preset threshold. When the confidence level exceeds the threshold, the pathogen is recorded as positive. Calculate the proportion of white blood cells containing toxic particles (T): Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ti is the number of toxic granular white blood cells in the i-th image, and li is the number of white blood cells in the i-th image. The number of epithelial cells is obtained from the quantitative analysis unit, and the proportion of basal cells J is calculated. Where i represents the i-th fluorescence image, N is the total number of images contained in the slide, ji is the number of basal cells in the i-th image, and ei is the number of epithelial cells in the i-th image; The report generation unit is configured to obtain bacterial quantity level information, pathogenic microorganism positive information and cell ratio information from the statistical analysis unit, and automatically calculate at least one of vaginal cleanliness, flora density, Nugent score, Donders score and pathogen positive judgment according to clinical rules.

8. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The touchscreen connects to the built-in microserver via USB 3.0 or Ethernet interface, ensuring fast and accurate transmission of operation commands and detection results. The touchscreen's user interface is simple and intuitive, providing a user-friendly experience and allowing operators to quickly get started. The image preprocessing algorithms in the new algorithm system include denoising, enhancement, and correction algorithms. The denoising algorithm uses a wavelet transform-based method, which can effectively remove noise interference in fluorescence images and improve the signal-to-noise ratio of the image. The enhancement algorithm uses histogram equalization and contrast stretching to enhance the contrast and brightness of the image and highlight useful information in the image; the correction algorithm is used to correct the geometric distortion and color deviation of the image to ensure the accuracy and consistency of the image. The feature extraction algorithm in the new algorithm system adopts the deep learning method, constructs a deep convolutional neural network (CNN) model, and trains it with a large amount of fluorescence image data to learn the morphological and texture features of different microorganisms. During training, data augmentation techniques, such as rotation, flipping, and scaling, are used to expand the training dataset and improve the model's generalization ability. After training, the CNN model can automatically extract effective features from fluorescence images for subsequent classification and recognition.

9. The instrument of the automated immunofluorescence vaginal secretion detector according to claim 1, characterized in that, The novel fluorescence focusing method in the new algorithm system innovatively employs the following steps: first, focusing with white light; then, calculating the appropriate focus point for the fluorescence using the phase difference between the white light and the fluorescence; and finally, scanning with fluorescence. White light focusing: The sample is illuminated by a white light source, the white light image is acquired by an image acquisition device, and the white light focusing position is quickly determined by an autofocus algorithm (such as phase detection focusing or contrast detection focusing). White light focusing has the characteristics of high brightness and short exposure time, and will not quench the sample by high-energy excitation light. It can quickly and accurately determine the focusing position. Phase difference calculation: By analyzing the phase difference between white light and fluorescence, the correct focus point for fluorescence is calculated. During the device development process, numerous calibration experiments are conducted beforehand to obtain the conversion relationship between fluorescence and white light, establishing a mathematical model between phase difference and focus position. Based on this model, the focus position for fluorescence can be accurately calculated from the focus position for white light. Fluorescence scanning: The focus position is adjusted to the calculated fluorescence focus position, the sample is illuminated with a fluorescence light source, and a fluorescence image is scanned. This avoids the problem of lower light intensity and darker areas in the focused area caused by taking many pictures while the focus process is constantly lit when using fluorescence focus. This is beneficial for clinical observation and reduces misjudgment.

Images