Method and system for identifying oligonucleotide gel electrophoresis image

Abstract

The invention discloses a recognition method and system for an oligonucleotide gel electrophoresis image, and the method comprises the steps: obtaining a to-be-recognized image which is an oligonucleotide gel electrophoresis image uploaded by a user based on a web browser; extracting a lane area of the to-be-recognized image by using a multi-feature fusion recognition algorithm, and performing strip detection on the lane area to obtain a strip boundary; generating stripe feature parameters based on the stripe boundaries, and performing typing on the to-be-identified image according to the stripe feature parameters by using a preset typing rule to obtain a typing conclusion; and generating a structured diagnosis report based on the typing conclusion, and displaying the diagnosis report through a web browser. According to the method, the oligonucleotide gel electrophoresis image is identified and analyzed, so that the automation of a diagnostic report generation process is realized.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, specifically to a method and system for recognizing oligoclonal gel electrophoresis images. Background Technology

[0002] Oligoclonal gel electrophoresis is a key diagnostic tool for central nervous system diseases. Its core principle is to analyze the differences in protein band distribution between cerebrospinal fluid (CSF) and serum samples to determine the presence of intrathecal immunoglobulins in the central nervous system, providing crucial evidence for the diagnosis of diseases such as multiple sclerosis. With the increasing throughput of clinical testing, higher demands are being placed on the efficiency, accuracy, and standardization of oligoclonal gel electrophoresis image analysis.

[0003] Currently, most oligoclonal gel electrophoresis image analysis technologies rely on stand-alone software, which has significant technical limitations. Firstly, the deployment and maintenance of stand-alone software are complex, unable to support remote collaboration and data sharing across multiple institutions. Furthermore, the user interface is cumbersome and requires a high level of expertise, necessitating specialized training for ordinary laboratory personnel. Secondly, existing technologies lack standardized automatic typing functions, relying primarily on manual interpretation and typing of lanes and bands in electrophoresis images. This not only results in low analysis efficiency, with single-sample analyses taking 5-10 minutes, failing to meet high-throughput testing needs, but also leads to poor consistency and accuracy due to the subjectivity of manual interpretation, severely impacting the reliability of clinical diagnosis. In addition, existing systems offer limited result presentation, lacking visualization support and structured clinical interpretation, hindering rapid result interpretation by laboratory personnel and providing valuable reference for clinicians.

[0004] Therefore, there is an urgent need for a technical solution that can achieve automatic, efficient and accurate identification of oligoclonal gel electrophoresis images to solve the problems of reliance on manual interpretation, poor result consistency and insufficient visualization in existing technologies. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for recognizing oligoclonal gel electrophoresis images, aiming to identify and analyze oligoclonal gel electrophoresis images and automate the diagnostic report generation process.

[0006] This invention discloses a method for recognizing oligoclonal gel electrophoresis images, comprising: Acquire the image to be identified, which is an oligoclonal gel electrophoresis image uploaded by the user via a web browser; The swimlane region of the image to be identified is extracted using a multi-feature fusion recognition algorithm, and strip detection is performed on the swimlane region to obtain the strip boundary; Based on the strip boundary, strip feature parameters are generated, and the image to be identified is classified according to the strip feature parameters using preset classification rules to obtain classification conclusions; A structured diagnostic report is generated based on the classification results and displayed through a web browser.

[0007] Preferably, after acquiring the image to be identified, the method further includes: The format of the image to be recognized is verified. If the image to be recognized is in a preset format, a verification result indicating successful format verification is generated. The preset formats include PNG, JPG, JPEG and TIFF formats. The system performs security verification on the image to be recognized. If the image passes the security verification, a verification result indicating successful security verification is generated. The security verification includes file integrity verification, session security verification, and permission verification. If the verification result is that the format verification is successful and the security verification is successful, the image to be recognized will be stored in the image database. If the verification result is a format verification failure or a security verification failure, a prompt message will be returned to the user through the web browser, prompting the user to re-upload the image to be recognized.

[0008] Preferably, the swimlane region of the image to be identified is extracted using a multi-feature fusion recognition algorithm, including: The image to be identified is converted to the HSV color space to obtain an HSV image; The HSV image is binarized based on the first hue threshold range and the second hue threshold range to obtain a mask image. The mask image includes at least one swimlane. The first hue threshold range is used to mark the purple area, and the second hue threshold range is used to mark the white interval. Perform morphological closing operations on the mask image to fill the holes in the mask image and generate a swimlane mask; Calculate the vertical projection of the lane mask in the vertical direction, and locate the lane position list based on the vertical projection; The cerebrospinal fluid lanes and serum lanes are marked based on the lane location list, and the cerebrospinal fluid lanes and serum lanes are arranged alternately. The swim lane region is generated based on cerebrospinal fluid swim lanes and serum swim lanes.

[0009] Preferably, the HSV image is binarized based on a first hue threshold range and a second hue threshold range to obtain a mask image, including: A purple region mask is created based on a first preset hue range, which is [120,40,40]-[160,255,255]. A white area mask is created based on a second preset hue range, which is [0,0,200]-[180,50,255]. A mask image is generated based on the purple and white region masks.

[0010] Preferably, strip detection is performed on the lane area to obtain the strip boundaries, including: Obtain the swimlane image corresponding to the swimlane region; The swimlane image is converted to grayscale to obtain a swimlane grayscale image, and the average brightness value of the swimlane grayscale image is calculated. A dark threshold is set based on the average brightness value, and a composite dark mask image corresponding to the HSV image is created based on the dark threshold. The composite dark mask image includes a dark purple area corresponding to the third hue threshold range and a black area corresponding to the fourth hue threshold range. Morphological enhancement processing is performed on the composite dark mask image, and the horizontal projection of the enhanced composite dark mask image is calculated to obtain the projection data; The peak detection algorithm is used to detect the strip positions of the projected data, and strip boundaries are generated based on the strip positions.

[0011] Preferably, morphological enhancement processing is performed on the composite dark mask image, including: A horizontal kernel is used to perform a closing operation on the composite dark mask image to enhance horizontal features, and a vertical kernel is used to perform an opening operation on the composite dark mask image to suppress vertical noise. The peak detection algorithm is used to detect the strip positions of the projected data, and strip boundaries are generated based on the strip positions, including: The projection data is filtered to smooth it. The peak points of the smoothed projection data are identified using a peak detection algorithm, and the strip positions are determined from the peak points based on a preset peak height threshold. Delineate strip boundaries in swimlane images based on strip location.

[0012] Preferably, generating strip feature parameters based on strip boundaries includes: Based on the lane location list, the cerebrospinal fluid bands and serum bands corresponding to the band boundaries are marked in the lane images, where the cerebrospinal fluid bands are located on the cerebrospinal fluid lanes and the serum bands are located on the serum lanes. Count the number of cerebrospinal fluid bands and serum bands; Obtain information on the position and brightness of cerebrospinal fluid and serum bands; Strip feature parameters are generated based on information on the number of cerebrospinal fluid strips, the number of serum strips, the location of the strips, and their brightness.

[0013] Preferably, the image to be identified is classified according to the strip feature parameters using a preset classification rule, including: If the number of cerebrospinal fluid bands is less than a preset threshold and the number of serum bands is less than a preset threshold, the image to be identified is marked as a type I category. Type I category indicates negative, and the preset threshold is 2. If the number of cerebrospinal fluid bands is greater than or equal to a preset threshold, and the number of serum bands is 0, then the image to be identified is marked as type II, and type II represents the specificity of cerebrospinal fluid. If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to the preset number thresholds, and the number of cerebrospinal fluid bands is greater than the number of serum bands, then the image to be identified is marked as a type III category. The type III category indicates that the number of cerebrospinal fluid bands is greater than the number of serum bands. If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to the preset number threshold, and the number of cerebrospinal fluid bands is equal to the number of serum bands, and the bands are not located at the edge of the lane, then the image to be identified is marked as a type IV category. Type IV category indicates that the number of cerebrospinal fluid bands is equal to the number of serum bands. If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to a preset threshold, and the number of cerebrospinal fluid bands is equal to the number of serum bands, and the bands are located at the edge of the lane, then the image to be identified is labeled as a type 5 category, and the type 5 category represents monoclonalness.

[0014] Preferably, the diagnostic report includes the image to be identified, the analysis result image, and the classification conclusion. The analysis result image uses different colors to mark the lane regions, band boundaries, and band characteristic parameters.

[0015] This invention discloses an oligoclonal gel electrophoresis image recognition system, applied to the aforementioned oligoclonal gel electrophoresis image recognition method. The oligoclonal gel electrophoresis image recognition system includes: The image acquisition module is configured to acquire an image to be identified, which is an oligoclonal gel electrophoresis image uploaded by the user based on a web browser; The strip detection module is configured to: extract the swimlane region of the image to be identified using a multi-feature fusion recognition algorithm, and perform strip detection on the swimlane region to obtain the strip boundary; The image classification module is configured to: generate strip feature parameters based on strip boundaries, and classify the image to be identified according to the strip feature parameters using preset classification rules to obtain classification conclusions; The report generation module is configured to generate a structured diagnostic report based on the classification conclusions and display the diagnostic report through a web browser.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. It adopts a lightweight architecture for uploading and displaying via a web browser, eliminating the need for users to install a local client and lowering the barrier to entry for using the system.

[0017] 2. A multi-feature fusion recognition algorithm is adopted, which combines the purple lane and white interval features of the HSV color space, the brightness distribution of grayscale images and morphological processing to achieve accurate extraction of the lane region and automated detection of the strip boundaries. This effectively avoids the problems of missed or false detection of strips caused by subjective experience differences in manual interpretation.

[0018] 3. By quantifying the quantity and positional relationship of cerebrospinal fluid and serum bands, objective typing of electrophoresis images was achieved, eliminating the subjective bias of manual typing. Attached Figure Description

[0019] Figure 1 A schematic flowchart illustrating the method for recognizing oligoclonal gel electrophoresis images provided by the present invention; Figure 2 This is a schematic diagram of the structure of the oligoclonal gel electrophoresis image recognition system provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The present invention will now be described in further detail with reference to the accompanying drawings.

[0022] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for recognizing oligoclonal gel electrophoresis images, including the following steps.

[0023] S1. Obtain the image to be recognized.

[0024] In this embodiment of the invention, the user selects a locally stored oligoclonal gel electrophoresis image file through a web browser's visual upload interface. The system receives the file data stream via the HTTP protocol and performs format verification and security checks on the image.

[0025] Specifically, the image to be recognized is first validated for its format. If the image is in a preset format, a validation result indicating successful format validation is generated. Preset formats include PNG, JPG, JPEG, and TIFF. Next, a security check is performed on the image. If the image passes the security check, a validation result indicating successful security validation is generated. Security checks include file integrity verification, session security verification, and permission verification. If both format and security checks are successful, the image is stored in the image database. If either format or security checks fail, a prompt message is returned to the user via the web browser, prompting the user to re-upload the image.

[0026] For example, when a user uploads an image named "patient_20240512_elec.png", the system first reads the PNG format identifier in the file header to confirm that it meets the preset PNG format requirements. Then, it initiates a security verification process, using MD5 hash comparison to verify that no data loss or tampering occurred during file transmission. Simultaneously, it checks the validity of the current user session's token and whether the user has the necessary permissions for electrophoresis image analysis. If all verifications pass, the system stores the image in a designated partition of the distributed image database and generates a unique image index ID associated with the patient's test request. If verification fails, such as when the user uploads a BMP image, the system will display a red warning box in the web browser, showing the specific error message "Only PNG, JPG, JPEG, and TIFF formats are supported. Please upload again," and retain the upload history for the user to quickly correct the error.

[0027] S2. Use a multi-feature fusion recognition algorithm to extract the swimlane region of the image to be recognized, and perform strip detection on the swimlane region to obtain the strip boundary.

[0028] In this embodiment of the invention, the image to be identified is an RGB image, which is then converted to the HSV color space to obtain an HSV image. The HSV color space is a representation space based on the decoupling of color and brightness. It separates and processes color information (Hue), saturation, and value, which is beneficial for accurately identifying target features under complex background interference.

[0029] Next, the HSV image is binarized based on a first hue threshold range and a second hue threshold range to obtain a mask image. The mask image includes at least one swimlane, with the first hue threshold range used to mark purple regions and the second hue threshold range used to mark white intervals. During the generation of the mask image, a purple region mask is created based on a first preset hue range ([120,40,40]-[160,255,255]), and a white region mask is created based on a second preset hue range ([0,0,200]-[180,50,255]). The mask image is then generated based on the purple and white region masks.

[0030] Furthermore, a morphological closing operation is performed on the mask image to fill the holes in the mask image, generating a swimlane mask. The morphological closing operation is a nonlinear filtering operation that first dilates and then erodes, used to connect broken regions and smooth boundaries, effectively eliminating internal breaks or edge burrs in the swimlanes caused by uneven illumination or development deviations. The vertical projection of the swimlane mask is calculated, and a list of swimlane positions is located based on this projection. Cerebrospinal fluid (CSF) and serum swimlanes are then marked based on this list, arranged alternately. Swimlane regions are generated based on these CSF and serum swimlanes.

[0031] For example, in this embodiment of the invention, the image to be identified is converted from the RGB color space to the HSV color space to obtain an HSV image. Targeting the characteristic of purple lanes and white intervals in the oligoclonal gel electrophoresis image, a purple region mask is created based on a first preset hue range [120,40,40]-[160,255,255], and a white region mask is created based on a second preset hue range [0,0,200]-[180,50,255]. The two masks are then logically ORed to generate a mask image containing the lane outlines.

[0032] Morphological closing operations are then performed on the mask image; for example, 5×5 rectangular structuring elements are used to fill tiny holes in the mask to generate a continuous swimlane mask. The cumulative pixel value of the swimlane mask in the vertical direction is calculated as the vertical projection. The swimlane position list is located by detecting the peaks and troughs of the projection curve, where peaks correspond to swimlane regions and troughs correspond to interval regions. According to the standard settings of clinical electrophoresis experiments, odd-numbered positions in the labeling list are cerebrospinal fluid swimlanes, and even-numbered positions are serum swimlanes, thus obtaining the final swimlane regions.

[0033] In this embodiment of the invention, the swimlane region includes multiple stripes, requiring precise segmentation and feature extraction of these stripes. Specifically, the swimlane image corresponding to the swimlane region is obtained, and it is converted to grayscale to obtain a swimlane grayscale image. The average brightness value of all pixels in the grayscale image is then calculated.

[0034] A dark threshold is set based on the average brightness value, and a composite dark mask image corresponding to the HSV image is created based on the dark threshold. The composite dark mask image includes a dark purple area corresponding to the third hue threshold range and a black area corresponding to the fourth hue threshold range.

[0035] For example, using 60% of the average brightness value as the dark threshold, a composite dark mask image is created by combining the features of deep purple (hue range [130,50,30]-[150,255,200]) and black (hue range [0,0,0]-[180,255,50]) in the HSV image.

[0036] Furthermore, morphological enhancement processing is performed on the composite dark mask image, including: performing a closing operation on the composite dark mask image using a horizontal kernel to enhance horizontal features, and performing an opening operation on the composite dark mask image using a vertical kernel to suppress vertical noise. The horizontal projection of the enhanced composite dark mask image is then calculated to obtain projection data. A peak detection algorithm is used to detect the strip positions in the projection data, and strip boundaries are generated based on these positions. Specifically, the projection data is filtered to smooth it; the peak points of the smoothed projection data are identified using a peak detection algorithm, and strip positions are determined from these peak points according to a preset peak height threshold; and strip boundaries are delineated in the swimlane image based on the strip positions.

[0037] For example, during the morphological enhancement of the mask image, a 3×1 horizontal kernel is first used to perform a closing operation on the image to enhance the horizontal continuity of the stripes; then, a 1×3 vertical kernel is used to perform an opening operation on the image to remove vertical noise points. The cumulative pixel value in the horizontal direction of the enhanced mask image is calculated as projection data, and Savitzky-Golay filtering is applied to the projection data for smoothing to reduce noise interference. Then, a peak detection algorithm is used to identify the peak points of the smoothed projection data, and peaks with heights exceeding a preset threshold are selected as strip positions. Finally, based on the strip positions, a rectangular box containing the upper and lower boundaries of the stripes is drawn in the swimlane image to obtain the stripe boundaries.

[0038] For example, for an electrophoresis image containing 6 lanes (3 cerebrospinal fluid lanes and 3 serum lanes), the system detects 6 consecutive peaks through vertical projection, labeling positions 1, 3, and 5 as cerebrospinal fluid lanes and positions 2, 4, and 6 as serum lanes. During the band detection stage, the average brightness value of the grayscale image of a certain cerebrospinal fluid lane is 120, and the dark threshold is set to 84. Three dark purple band regions are captured using a composite dark mask. After morphological enhancement and peak detection, the boundaries of the three bands are finally determined to be the [12,35], [42,68], and [75,96] pixel row ranges of the lane image.

[0039] In this way, the multi-feature fusion recognition algorithm achieves precise positioning of the lane region and automated detection of the strip boundaries. This avoids the boundary shift problem caused by visual fatigue in traditional manual annotation and solves the risk of missed detection in complex backgrounds (such as discolored spots in the gel background or dark areas with uneven development) under single-feature recognition. For example, when the electrophoresis image has local overexposure that causes blurring of some lane edges, the hue threshold screening of the HSV color space can still stably identify the core area of ​​the purple lane. Combined with morphological closing operation, it further fills in the small breaks at the edges to ensure the integrity of the lane region. As for the strip breakage problem that may occur in strip detection, the horizontal kernel closing operation can effectively connect the broken parts at the same horizontal position, while the vertical kernel opening operation can filter out the vertical pseudo-strips caused by gel impurities. Finally, the strip boundary error obtained by peak detection is small, which meets the requirements of clinical diagnosis for strip position accuracy.

[0040] S3. Generate strip feature parameters based on strip boundaries, and use preset classification rules to classify the image to be identified to obtain classification conclusions.

[0041] In this embodiment of the invention, based on the lane position list, cerebrospinal fluid (CSF) bands and serum bands corresponding to the band boundaries are marked in the lane image, wherein the CSF bands are located on the CSF lanes and the serum bands are located on the serum lanes. The number of CSF bands and serum bands are counted; the band position and brightness information of the CSF and serum bands are obtained; and band feature parameters are generated based on the number of CSF bands, the number of serum bands, the band position, and the brightness information.

[0042] In this way, the band characteristic parameters not only cover the comparison of the number of bands in cerebrospinal fluid and serum, but also include the relative position coordinates and average brightness value of each band in the lane, providing a multi-dimensional quantitative basis for subsequent typing.

[0043] Furthermore, the image to be identified is classified according to the band feature parameters using preset classification rules. Specifically, if the number of cerebrospinal fluid (CSF) bands is less than a preset threshold, and the number of serum bands is also less than a preset threshold, the image is classified as Type I, which indicates negative. If the number of CSF bands is greater than or equal to the preset threshold, and the number of serum bands is 0, the image is classified as Type II, which indicates CSF specificity. If both the number of CSF bands and the number of serum bands are greater than or equal to the preset threshold, and the number of CSF bands is greater than the number of serum bands, the image is classified as Type III, which indicates that the number of CSF bands is greater than the number of serum bands. If both the number of CSF bands and the number of serum bands are greater than or equal to the preset threshold, and the number of CSF bands is equal to the number of serum bands, and the bands are not located at the edge of the lane, the image is classified as Type IV, which indicates that the number of CSF bands is equal to the number of serum bands. If the number of cerebrospinal fluid (CSF) bands and the number of serum bands are both greater than or equal to a preset threshold, and the number of CSF bands equals the number of serum bands, and the bands are located at the edge of the lane, then the image to be identified is labeled as a type 5 category, where type 5 represents monoclonalness. The preset threshold can be 2.

[0044] It should be noted that the preset quantity threshold and preset classification rules are constructed by combining clinical practice guidelines and multicenter experimental data, covering common oligoclonal banding classification standards. This method enables automated and standardized classification of electrophoresis images, avoiding inconsistencies caused by subjective judgment during manual classification. Furthermore, the preset quantity threshold can be adjusted according to specific circumstances; for example, it can be 3 or 4.

[0045] For example, if a patient's cerebrospinal fluid (CSF) electrophoresis lane shows 3 bands and their serum lane shows 0 bands, the system classifies it as type II, indicating the presence of CSF-specific oligoclonal bands. This result is highly consistent with the typical electrophoretic findings in multiple sclerosis (MS) in clinical practice. If another patient has 4 bands in their CSF lane and 3 bands in their serum lane, with the number of CSF bands exceeding the number of serum bands, the system labels them as type III, suggesting possible abnormal blood-brain barrier permeability accompanied by local immunoglobulin synthesis.

[0046] S4. Generate a structured diagnostic report based on the classification results and display the diagnostic report through a web browser.

[0047] In this embodiment of the invention, the diagnostic report includes the image to be identified, the analysis result image, and the classification conclusion. The analysis result image uses different colors to mark the lane regions, band boundaries, and band feature parameters.

[0048] Specifically, the classification conclusions, band feature parameters (including the number, location coordinates, and brightness values ​​of cerebrospinal fluid and serum bands), and lane region image data obtained in step S3 are integrated into a standardized recognition result set. Simultaneously, an interactive report interface is generated based on a front-end visualization component. The left side of the interface displays a comparison view of the original image to be identified and the processed lane mask image and band boundary annotation image. The right side presents the band feature parameters in table form, and the classification conclusions and corresponding clinical prompts are highlighted using different colored labels.

[0049] For example, if the classification conclusion is type II, the interface will automatically pop up a clinical suggestion box that says "Cerebrospinal fluid specific oligoclonal bands suggest further evaluation of central nervous system immune lesions in conjunction with MRI examination."

[0050] Furthermore, this invention supports exporting the recognition results in PDF format. The exported file contains complete image annotations, parameter lists, and classification logic explanations, meeting the needs of clinical archiving and academic research. For batch-uploaded electrophoresis images, the system can also generate summary statistical reports, displaying the classification distribution of different patients and statistical indicators such as the mean and standard deviation of band characteristics, providing data support for clinical research.

[0051] like Figure 2 As shown, this embodiment of the invention also provides an oligoclonal gel electrophoresis image recognition system, applied to the aforementioned oligoclonal gel electrophoresis image recognition method. The oligoclonal gel electrophoresis image recognition system includes an image acquisition module 201, a band detection module 202, an image typing module 203, and a report generation module 204.

[0052] The image acquisition module 201 is configured to acquire an image to be identified, which is an oligoclonal gel electrophoresis image uploaded by the user via a web browser. The band detection module 202 is configured to extract the lane regions of the image to be identified using a multi-feature fusion recognition algorithm, and perform band detection on the lane regions to obtain the band boundaries. The image classification module 203 is configured to generate band feature parameters based on the band boundaries, and classify the image to be identified according to the band feature parameters using preset classification rules to obtain a classification conclusion. The report generation module 204 is configured to generate a structured diagnostic report based on the classification conclusion and display the diagnostic report through a web browser.

[0053] As can be seen from the above technical solutions, this invention discloses a method and system for recognizing oligoclonal gel electrophoresis images. The method for recognizing oligoclonal gel electrophoresis images includes: acquiring an image to be recognized, which is an oligoclonal gel electrophoresis image uploaded by a user via a web browser; extracting the lane regions of the image to be recognized using a multi-feature fusion recognition algorithm, and performing band detection on the lane regions to obtain band boundaries; generating band feature parameters based on the band boundaries, and classifying the image to be recognized according to the band feature parameters using preset classification rules to obtain classification conclusions; generating a structured diagnostic report based on the classification conclusions, and displaying the diagnostic report through a web browser. This invention automates the diagnostic report generation process by recognizing and analyzing oligoclonal gel electrophoresis images.

[0054] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for recognizing oligoclonal gel electrophoresis images, characterized in that, include: Acquire the image to be identified, which is an oligoclonal gel electrophoresis image uploaded by the user via a web browser; The swimlane region of the image to be identified is extracted using a multi-feature fusion recognition algorithm, and strip detection is performed on the swimlane region to obtain the strip boundary; Based on the strip boundaries, strip feature parameters are generated, and the image to be identified is classified according to the strip feature parameters using preset classification rules to obtain classification conclusions. A structured diagnostic report is generated based on the classification results, and the diagnostic report is displayed through a web browser.

2. The method for recognizing oligoclonal gel electrophoresis images according to claim 1, characterized in that, After acquiring the image to be recognized, the process also includes: The format of the image to be identified is verified. If the image to be identified is in a preset format, a verification result indicating successful format verification is generated. The preset formats include PNG, JPG, JPEG, and TIFF formats. The image to be identified is subjected to security verification. If the image to be identified passes the security verification, a verification result indicating successful security verification is generated. The security verification includes file integrity verification, session security verification, and permission verification. If the verification result is that the format verification is successful and the security verification is successful, then the image to be identified is stored in the image database; If the verification result is a format verification failure or a security verification failure, a prompt message is returned to the user through the web browser, which prompts the user to re-upload the image to be recognized.

3. The method for recognizing oligoclonal gel electrophoresis images according to claim 1, characterized in that, The step of extracting the swimlane region of the image to be identified using a multi-feature fusion recognition algorithm includes: The image to be identified is converted to the HSV color space to obtain an HSV image; The HSV image is binarized based on a first hue threshold range and a second hue threshold range to obtain a mask image. The mask image includes at least one swimlane. The first hue threshold range is used to mark purple areas, and the second hue threshold range is used to mark white intervals. A morphological closing operation is performed on the mask image to fill the holes in the mask image, generating a swimlane mask; Calculate the vertical projection of the lane mask in the vertical direction, and locate the lane position list based on the vertical projection; The cerebrospinal fluid lanes and serum lanes are marked based on the lane location list, and the cerebrospinal fluid lanes and serum lanes are arranged alternately. The lane region is generated based on the cerebrospinal fluid lane and the serum lane.

4. The method for recognizing oligoclonal gel electrophoresis images according to claim 3, characterized in that, The binarization of the HSV image based on a first hue threshold range and a second hue threshold range to obtain a mask image includes: A purple region mask is created based on a first preset hue range, where the first preset hue range is [120,40,40]-[160,255,255]. A white area mask is created based on a second preset hue range, where the second preset hue range is [0,0,200]-[180,50,255]. The mask image is generated based on the purple region mask and the white region mask.

5. The method for recognizing oligoclonal gel electrophoresis images according to claim 3, characterized in that, The step of performing strip detection on the lane region to obtain the strip boundaries includes: Obtain the swim lane image corresponding to the swim lane region; The swim lane image is converted to grayscale to obtain a swim lane grayscale image, and the average brightness value of the swim lane grayscale image is calculated. A dark threshold is set based on the average brightness value, and a composite dark mask image corresponding to the HSV image is created according to the dark threshold. The composite dark mask image includes a dark purple region corresponding to the third hue threshold range and a black region corresponding to the fourth hue threshold range. The composite dark mask image is subjected to morphological enhancement processing, and the horizontal projection of the enhanced composite dark mask image is calculated to obtain projection data; The peak detection algorithm is used to detect the strip positions of the projected data, and the strip boundaries are generated based on the strip positions.

6. The method for recognizing oligoclonal gel electrophoresis images according to claim 5, characterized in that, The morphological enhancement processing of the composite dark mask image includes: The composite dark mask image is closed using a horizontal kernel to enhance horizontal features, and opened using a vertical kernel to suppress vertical noise. The step of detecting the strip positions of the projected data using a peak detection algorithm and generating the strip boundaries based on the strip positions includes: The projection data is filtered to smooth it. The peak points of the smoothed projection data are identified using a peak detection algorithm, and the strip positions are determined from the peak points according to a preset peak height threshold. The strip boundaries are defined in the lane image based on the strip positions.

7. The method for recognizing oligoclonal gel electrophoresis images according to claim 5, characterized in that, The generation of strip feature parameters based on the strip boundary includes: Based on the lane location list, cerebrospinal fluid bands and serum bands corresponding to the band boundaries are marked in the lane image, wherein the cerebrospinal fluid bands are located on the cerebrospinal fluid lanes and the serum bands are located on the serum lanes; Count the number of cerebrospinal fluid bands and serum bands; Obtain the band position and brightness information of the cerebrospinal fluid band and the serum band; The band feature parameters are generated based on the number of cerebrospinal fluid bands, the number of serum bands, the band positions, and the brightness information.

8. The method for recognizing oligoclonal gel electrophoresis images according to claim 7, characterized in that, The step of classifying the image to be identified using a preset classification rule based on the band feature parameters includes: If the number of cerebrospinal fluid bands is less than a preset threshold and the number of serum bands is less than a preset threshold, then the image to be identified is marked as a type I category, which represents negative, and the preset threshold is 2. If the number of cerebrospinal fluid bands is greater than or equal to the preset number threshold, and the number of serum bands is 0, then the image to be identified is marked as type II, and type II represents cerebrospinal fluid specificity. If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to the preset number threshold, and the number of cerebrospinal fluid bands is greater than the number of serum bands, then the image to be identified is marked as a type III category, where the type III category indicates that the number of cerebrospinal fluid bands is greater than the number of serum bands. If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to the preset number threshold, and the number of cerebrospinal fluid bands is equal to the number of serum bands, and the band position is not at the edge of the lane, then the image to be identified is marked as a type IV category, and the type IV category indicates that the number of cerebrospinal fluid bands is equal to the number of serum bands; If the number of cerebrospinal fluid bands and the number of serum bands are both greater than or equal to the preset number threshold, and the number of cerebrospinal fluid bands is equal to the number of serum bands, and the bands are located at the edge of the lane, then the image to be identified is marked as a type 5 category, and the type 5 category represents monoclonalness.

9. The method for recognizing oligoclonal gel electrophoresis images according to claim 1, characterized in that, The diagnostic report includes the image to be identified, the analysis result image, and the classification conclusion. The analysis result image uses different colors to mark the lane region, the band boundary, and the band feature parameters.

10. A system for recognizing oligoclonal gel electrophoresis images, applied to the method for recognizing oligoclonal gel electrophoresis images as described in any one of claims 1-9, characterized in that, include: The image acquisition module is configured to acquire an image to be identified, wherein the image to be identified is an oligoclonal gel electrophoresis image uploaded by the user based on a web browser; The strip detection module is configured to: extract the swimlane region of the image to be identified using a multi-feature fusion recognition algorithm, and perform strip detection on the swimlane region to obtain the strip boundary; The image classification module is configured to: generate strip feature parameters based on the strip boundaries, and classify the image to be identified according to the strip feature parameters using preset classification rules to obtain a classification conclusion; The report generation module is configured to generate a structured diagnostic report based on the classification conclusions and display the diagnostic report through a web browser.

Images