Meibomian gland laser treatment device

Abstract

The invention belongs to the technical field of medical treatment, and relates to a meibomian gland laser treatment device, which comprises a treatment handle provided with a laser module, an infrared camera module and a display module; the control terminal is electrically connected with the laser module, the infrared camera module and the display module, and the control terminal is used for obtaining the marked meibomian gland image and then sending the marked meibomian gland image to the display module. According to the meibomian gland treatment system, the temperature value of the meibomian gland treatment area and the meibomian gland image can be obtained in real time in the treatment process, the meibomian gland abnormal area in the meibomian gland image is marked and displayed in real time through the display module, medical staff can observe the meibomian gland form in real time and position the meibomian gland abnormal area, and the meibomian gland treatment effect is improved. According to the meibomian gland therapeutic apparatus, blocked meibomian ester is molten in a targeted manner, so that abnormal meibomian glands are treated, the situation that treatment is not in place or excessive treatment is caused by undifferentiated treatment is avoided, and the treatment effect on the xerophthalmia is improved.

Description

Technical Field

[0001] The invention belongs to the field of medical technology and relates to a meibomian gland laser treatment device. Background Art

[0002] Meibomian gland dysfunction is a chronic, diffuse meibomian gland disease characterized by obstruction of the terminal ducts of the meibomian glands and / or abnormal quality or quantity of meibomian secretion. It can cause abnormal meibomian gland discharge function and blockage of the meibomian gland terminals. It is difficult to clean and easy to recur. It is a key pathogenic factor of dry eye. About 70% of dry eye patients are caused by meibomian gland dysfunction. Meibomian gland dysfunction requires a comprehensive imaging diagnosis of the ocular surface before treatment, including tear film-related examinations, meibomian gland morphology examinations, tear secretion measurements, etc. Meibomian gland morphology is closely related to meibomian gland function. Abnormal morphology of the meibomian glands includes absence, atrophy, vesicle dilation, tortuosity, etc., which will further affect the normal secretion of meibomian, thereby leading to a decrease in tear film stability.

[0003] Currently, photoelectric heating treatment equipment is often used in the treatment of meibomian gland dysfunction. Photothermal therapy is used to treat the entire eyelid on a large scale, converting light energy into heat energy to melt the meibomian fat, which is then expelled under pressure, thereby effectively unblocking the meibomian glands and restoring their function.

[0004] However, since the above-mentioned treatment process directly acts on the entire eyelid, the morphology of the meibomian glands cannot be observed during the treatment process, which may easily cause medical personnel to be unable to locate the area where the meibomian glands are abnormal, resulting in inadequate or excessive treatment of the meibomian glands, resulting in poor treatment effect of dry eye symptoms. Summary of the invention

[0005] The purpose of the present invention is to provide a meibomian gland laser treatment device, which can observe the morphology of the meibomian glands in real time, locate the abnormal areas of the meibomian glands, and melt the blocked meibomian fat in a targeted manner during the treatment of the meibomian glands, thereby avoiding inadequate treatment or excessive treatment.

[0006] To achieve the above purpose, the specific technical solution of a meibomian gland laser treatment device provided by the present invention is as follows:

[0007] A meibomian gland laser treatment device, comprising:

[0008] A treatment handle is provided with a laser module, an infrared camera module and a display module, wherein the laser module is used to output laser light, the infrared camera module is used to capture image information of the meibomian glands in real time, and the display module is used to display the image of the meibomian glands in real time;

[0009] The control terminal is electrically connected to the laser module, the infrared camera module and the display module respectively. The control terminal is used to receive the image information of the meibomian gland, convert the image information of the meibomian gland into a meibomian gland image, pre-process the meibomian gland image to obtain a pre-processed meibomian gland image, mark the area of ​​the meibomian gland abnormality in the pre-processed meibomian gland image to obtain a marked meibomian gland image, and then send the marked meibomian gland image to the display module.

[0010] The present invention is also characterized in that:

[0011] The control terminals include:

[0012] A conversion module, used for converting the received meibomian gland image information into a meibomian gland image;

[0013] A preprocessing module, used for performing denoising, enhancement and smoothing on the meibomian gland image in sequence to obtain a preprocessed meibomian gland image;

[0014] An image marking module is used to mark the area where the meibomian gland is abnormal in the preprocessed meibomian gland image to obtain a marked meibomian gland image;

[0015] The sending module is used to send the marked meibomian gland image to the display module.

[0016] The image labeling module includes:

[0017] An input unit for receiving a preprocessed meibomian gland image;

[0018] A marking model is used to mark the abnormal area of ​​the meibomian gland in the preprocessed meibomian gland image to obtain a marked meibomian gland image;

[0019] The output unit is used to output the marked meibomian gland image to the sending module.

[0020] The tagging model includes:

[0021] An acquisition unit is used to acquire multiple meibomian gland images, and mark the meibomian gland range and each meibomian gland in each meibomian gland image as a training sample;

[0022] A segmentation unit is used to segment each training sample to obtain the number, length, width, area ratio, curvature and loss ratio of the meibomian glands in each training sample;

[0023] A labeling model unit is constructed to convert the number, length, width, area ratio, curvature and loss ratio of the meibomian glands of each training sample into a feature vector and its corresponding classification label. All feature vectors and their corresponding classification labels are input into the SVM model for training to obtain a test model, which is then corrected to obtain a labeling model.

[0024] The segmentation units include:

[0025] The MobileNet network structure is used to build a Mobile-U-Net algorithm based on the meibomian gland area image and each meibomian gland marked on each training sample, and to segment the meibomian gland area and meibomian gland in the image.

[0026] The preprocessing modules include:

[0027] De-noising unit, used to convert the meibomian gland image into a grayscale image, use Gaussian blur to reduce noise, apply Canny edge detection to quickly identify the eyelashes of the meibomian gland, apply binary threshold to detect highlight reflective points, and use dilation and erosion operations to remove unnecessary image features;

[0028] An enhancement processing unit, used for improving the contrast of the denoised meibomian gland image using CLAHE;

[0029] The smoothing processing unit is used to sharpen the enhanced meibomian gland image using a Laplace operator, reduce image noise, and smooth the image.

[0030] A manual control switch is provided on the treatment handle, the manual control switch is electrically connected to the laser module, and the manual control switch is used to control the start and stop of the laser module.

[0031] An infrared temperature measurement module and an alarm module are provided on the treatment handle. The infrared temperature measurement module is electrically connected to the control terminal and the display module respectively, and the alarm module is electrically connected to the control terminal. The infrared temperature measurement module is used to detect the temperature information of the meibomian gland treatment area in real time and send it to the control terminal. The control terminal converts the temperature information of the meibomian gland treatment area into the temperature value of the meibomian gland treatment area, and sends it to the display module and the alarm module respectively. The display module displays the temperature value of the meibomian gland treatment area in real time. The alarm module is used to compare the temperature value of the meibomian gland treatment area with a preset temperature value. When the temperature value of the meibomian gland treatment area is greater than the preset temperature value, an alarm is triggered.

[0032] The meibomian gland laser treatment device of the present invention has the following advantages:

[0033] First, through the cooperation of the infrared camera module, the infrared temperature measurement module, the display module and the control terminal, the temperature value of the meibomian gland treatment area and the meibomian gland image can be obtained in real time during the treatment process, and the area with abnormal meibomian glands in the meibomian gland image can be marked and displayed in real time through the display module, so that medical staff can observe the morphology of the meibomian glands in real time, locate the area with abnormal meibomian glands, and melt the blocked meibomian fat in a targeted manner, thereby treating the abnormal meibomian glands, avoiding inadequate treatment or excessive treatment caused by indifferent treatment, and improving the treatment effect of dry eye symptoms.

[0034] Second, through the cooperation of the infrared temperature measurement module, the alarm module and the control terminal, the temperature of the meibomian gland treatment area can be kept constant at 42.5°C, avoiding temperature-induced damage or insufficient treatment, etc., flexibly matching the individual conditions of different patients, accurately treating the meibomian glands, and heating the meibomian glands more focused without damaging normal eye tissues.

[0035] Third, by setting the laser wavelength to 1726nm, the selective photothermal effect of 1726nm laser on lipids can be used to efficiently target and liquefy meibomian gland meibum, relieve blockage, and restore a healthy microenvironment of the ocular surface. At the same time, melanin has no absorption of 1726nm laser, and there is no risk of hair removal on the eyelashes and eyebrows around the eyes. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0037] Figure 2 It is a schematic diagram of the structure of the treatment handle in the present invention;

[0038] Figure 3 An image of the meibomian gland taken by the infrared camera module of the present invention;

[0039] Figure 4 is the initial meibomian gland image after preprocessing of the present invention;

[0040] Figure 5 The pre-processed training samples are manually labeled according to the present invention;

[0041] Figure 6 The cropped meibomian gland region image of the present invention;

[0042] Figure 7 is a labeled image of the meibomian gland of the present invention;

[0043] Figure 8 It is a schematic diagram of the Mobile-U-Net network structure of the present invention.

[0044] Reference numerals:

[0045] 1. Treatment handle; 2. Laser module; 3. Infrared camera module; 4. Infrared temperature measurement module; 5. Display module. DETAILED DESCRIPTION

[0046] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. Among them, in the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, such as A and / or B, which can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" refers to two or more than two. The following terms "first" and "second" are used for descriptive purposes only and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0047] like Figure 1 , Figure 2As shown, the present invention provides a meibomian gland laser treatment device, including a treatment handle and a control terminal. The treatment handle is provided with a laser module, an infrared camera module, an infrared temperature measurement module and a display module. The laser module, the infrared camera module and the infrared temperature measurement module are respectively arranged at the head of the treatment handle, and the display module is arranged at the back of the treatment handle. The laser module is used to output laser, the infrared temperature measurement module is used to detect the temperature information of the meibomian gland treatment area in real time, the infrared camera module is used to capture the image information of the meibomian gland in real time, and the display module is used to display the temperature value and the meibomian gland image of the meibomian gland treatment area in real time. The control terminal is electrically connected to the laser module, the infrared camera module and the display module respectively, and the control terminal is used to receive the temperature information and the image information of the meibomian gland treatment area, convert the temperature information of the meibomian gland treatment area into the temperature value of the meibomian gland treatment area, convert the image information of the meibomian gland into the meibomian gland image, and then convert the meibomian gland into the meibomian gland image. After the gland image is preprocessed, a preprocessed meibomian gland image is obtained, and the area with abnormal meibomian glands in the preprocessed meibomian gland image is marked to obtain a marked meibomian gland image, and then the temperature value of the meibomian gland treatment area and the marked meibomian gland image are sent to the display module, so that the display module can display the temperature value of the meibomian gland treatment area and the marked meibomian gland image in real time, and control the laser module to output laser. Through the cooperation of the infrared camera module, the infrared temperature measurement module, the display module and the control terminal, the area with abnormal meibomian glands in the meibomian gland image can be marked, and the marked meibomian gland image and the temperature value of the meibomian gland treatment area can be displayed in real time, so that medical personnel can observe the morphology of the meibomian glands in real time, locate the abnormal position for treatment, and melt the blocked meibomian glands in a targeted manner, thereby treating the abnormal meibomian glands, avoiding the situation of inadequate treatment or excessive treatment caused by indifferent treatment, and improving the treatment effect of dry eye symptoms.

[0048] like Figure 1 As shown, the control terminal includes a conversion module, a preprocessing module, an image marking module and a sending module. The conversion module is used to convert the received temperature information of the meibomian gland treatment area into the temperature value of the meibomian gland treatment area, and the image information of the meibomian gland into the meibomian gland image. The preprocessing module is used to perform denoising, enhancement and smoothing on the meibomian gland image in sequence to obtain a preprocessed meibomian gland image. The image marking module is used to mark the area with abnormal meibomian glands in the preprocessed meibomian gland image to obtain a marked meibomian gland image. The sending module is used to send the temperature value of the meibomian gland treatment area and the marked meibomian gland image to the display module.

[0049] like Figure 3 , Figure 4As shown, the preprocessing module includes a denoising processing unit, an enhancement processing unit, a smoothing processing unit and an overlay unit. The denoising processing unit is used to convert the meibomian gland image into a grayscale image, use Gaussian blur to reduce noise, apply Canny edge detection to quickly identify the eyelashes of the meibomian gland, apply a binary threshold to detect highlight reflective points, and use expansion and corrosion operations to remove unnecessary image features. The enhancement processing unit is used to use CLAHE to improve the contrast of the meibomian gland image after denoising. The smoothing processing unit is used to use the Laplace operator to sharpen the meibomian gland image after enhancement, reduce image noise, and smooth the image. The overlay unit is used to overlay the smoothed image with the original image to obtain a preprocessed meibomian gland image.

[0050] Among them, the image marking module includes an input unit, a marking model and an output unit. The input unit is used to receive the preprocessed meibomian gland image, the marking model is used to mark the area of ​​abnormal meibomian glands in the preprocessed meibomian gland image to obtain a marked meibomian gland image, and the output unit is used to output the marked meibomian gland image to the sending module.

[0051] Among them, the labeling model includes an acquisition unit, a segmentation unit and a labeling model construction unit. The acquisition unit is used to acquire multiple meibomian gland images, and mark the meibomian gland range and each meibomian gland in each meibomian gland image as a training sample. The segmentation unit is used to segment each training sample to obtain the number, length, width, area ratio, curvature and loss ratio of the meibomian glands in each training sample. The labeling model construction unit is used to convert the number, length, width, area ratio, curvature and loss ratio of the meibomian glands of each training sample into a feature vector and its corresponding classification label. All feature vectors and their corresponding classification labels are input into the SVM model for training to obtain a test model, and then the test model is corrected to obtain a labeling model.

[0052] like Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 As shown in the figure, after the acquisition unit acquires multiple meibomian gland images, each meibomian gland image is first subjected to denoising, enhancement and smoothing to improve the recognition of the meibomian glands and ensure the accuracy and effect of subsequent processing. At the same time, when marking the meibomian gland range and each meibomian gland in each meibomian gland image, the meibomian gland range and each meibomian gland are marked manually using Labelme software as shown in the figure. Figure 5 (The closed outer broken line is the range of the meibomian gland, and each inner broken line is a meibomian gland) According to the marking results, the cropped original image generates two binary images respectively. The meibomian gland area image is as follows Figure 6 Labeled image of the meibomian glands Figure 7 .

[0053] The segmentation unit includes a MobileNet network structure, which is used to build a Mobile-U-Net algorithm based on the meibomian gland area image and each meibomian gland marked on each training sample, and segment the meibomian gland area and meibomian gland in the image. The MobileNet network structure is a lightweight convolutional neural network. Because the network has a smaller number of parameters, the network is more lightweight, and the separable convolution enables the network to extract features from channels, it has a stronger ability to extract features than traditional convolution. Specifically:

[0054] 1. Mobile-U-Net network: The backbone network uses the MobileNetV3 lightweight network, and the task network is the commonly used Unet network. The network structure is as follows: Figure 8 As shown in the figure, first, the depthwise_conv block module in MobileNetV3 is transplanted to the encoding part of the Unet network; secondly, depthwise_convblock modules of different depths are used to extract features according to the position of the Unet encoding part. The depthwise_conv block module is the basic structure of Mobile-Net, and the convolution is decomposed by using the depthwise convolution and 1×1 convolution methods.

[0055] 2. Loss function: Use a mixed loss function L, which is the pixel-level cross entropy loss L CE and Dice coefficient loss L DC sum

[0056] L=L CE +L DC

[0057]

[0058]

[0059] y c,i and p c,i They represent the manually labeled segmentation results of category c and the i-th pixel respectively, M is the number of categories, and N is the total number of pixels in each batch.

[0060] 3. Evaluation indicators: intersection over union (IOU), accuracy (Ac), and similarity coefficient (F1). The intersection over union (IOU) is the ratio of the intersection and union of the manual annotation and automatic segmentation results. The accuracy (Accuracy) is the ratio of the number of correctly predicted samples to the total number of predicted samples. F1 is the weighted harmonic average of Precision and Recall.

[0061] Among them, the number, length, width, area ratio, curvature and loss ratio of the meibomian glands are obtained by extracting parameters of the meibomian glands in the segmented meibomian gland area image and the marked image of the meibomian glands, and analyzing the morphological parameters of the meibomian glands, namely the number, length, width, area ratio, curvature and loss ratio of the meibomian glands. The number of meibomian glands is determined by marking the connected domains, the calculated length and width of the meibomian glands are measured by analyzing the difference in the boundary coordinates of the meibomian glands, the area ratio of the meibomian glands is evaluated by the ratio of the area of ​​the meibomian glands to the area of ​​the entire meibomian gland region, the curvature of the meibomian glands is evaluated by fitting a curve to the center line of the meibomian glands and calculating the curvature, and the loss ratio of the meibomian glands is evaluated by comparing the area ratio of the complete glands and the missing glands.

[0062] When correcting the test model, part of the training samples are input into the test model to correct the test model, and the test model is corrected by comparing the meibomian gland image marked by the test model with the manually marked meibomian gland image.

[0063] Among them, the number of training samples is greater than 300, and the number of training samples used to correct the test model is greater than 100.

[0064] like Figure 1 As shown, the control terminal also includes a parameter control module, which is electrically connected to the laser module. The parameter control module is used to set the laser parameters of the laser module and control the laser module to output lasers with corresponding parameters.

[0065] Among them, the control parameters of the laser module are: laser wavelength 1700nm~1750nm, laser pulse width 0ms~50ms, laser repetition frequency 0Hz~10Hz, laser spot diameter 0.5mm~3mm. Preferably, the laser wavelength is 1726nm. The 1726nm laser can selectively photothermally act on lipids, efficiently and targetedly liquefy meibomian gland meibum, relieve blockage, and restore the healthy microenvironment of the ocular surface. Melanin has no absorption of 1726nm laser, and there is no risk of hair loss on eyelashes and eyebrows around the eyes.

[0066] like Figure 2 As shown, a manual control switch is provided on the treatment handle, the manual control switch is electrically connected to the laser module, and the manual control switch is used to control the start and stop of the laser module.

[0067] like Figure 2As shown, the laser module includes a semiconductor laser and a laser driver. The semiconductor laser is used to emit laser. The laser driver is electrically connected to the manual control switch, the parameter control module and the semiconductor laser respectively. On the one hand, the laser driver is used to receive the start and stop instructions sent by the manual control switch to control the start and stop of the semiconductor laser. On the other hand, it is used to receive the parameter control instructions sent from the parameter control module to control the parameters of the semiconductor laser emitting laser, that is, the laser driver receives the control instructions sent by the parameter control module, controls the semiconductor laser through the laser driver, and controls the output pulse width, repetition frequency and power of the semiconductor laser according to the preset current parameters. The semiconductor laser is a multi-single tube structure and is used to output therapeutic Therapeutic laser, preferably, a red light indicator light source is provided in the semiconductor laser for indicating the optical path and displaying the spot shape, size and position of the therapeutic light beam. The treatment handle realizes low-loss beam transmission with the semiconductor laser through optical fiber coupling. Furthermore, the semiconductor laser is provided with an optical power sensor, a thermistor, an optical fiber in-situ probe and a laser heat dissipation structure. The optical power sensor is arranged at the output port of the semiconductor laser for detecting the actual output optical power; the thermistor is arranged at the multi-single-tube laser; the optical fiber in-situ probe is arranged at the connection between the semiconductor laser and the treatment handle for monitoring whether the optical fiber connection status is normal; the laser heat dissipation structure is provided with a heat sink and a water system channel, and the laser temperature is reduced by circulating water cooling.

[0068] like Figure 1 As shown, an alarm module is provided on the treatment handle, and the alarm module is electrically connected to the control terminal. The alarm module is used to receive the temperature value of the meibomian gland treatment area sent by the control terminal, and compare it with the preset temperature value. When the temperature value of the meibomian gland treatment area is greater than the preset temperature value, an alarm reminder is issued. The preset temperature value is 42.5°C. 42.5°C can dissolve meibomian gland meibomian fat, eliminate sebaceous demodex mites, and indirectly inhibit the reproduction of Propionibacterium acnes and reduce inflammatory response.

[0069] like Figure 1 As shown, when the preprocessed meibomian gland image is marked by the marking model, it is graded according to the degree of variation of the meibomian glands and sent to the parameter control module. The parameter control module selects the control parameters of the corresponding laser module according to the degree of variation of the meibomian glands. The meibomian glands are divided into three levels according to the degree of variation and marked with different colors.

[0070] like Figure 1 As shown, the display module is a display screen, which is arranged on the side of the treatment handle. The temperature value of the meibomian gland treatment area and the marked meibomian gland image are displayed through the display screen.

[0071] Working principle: When in use, the infrared camera module captures the image information of the meibomian glands in real time and sends the image information of the meibomian glands to the control terminal. The infrared temperature measurement module detects the temperature information of the meibomian gland treatment area in real time and sends the temperature information of the meibomian gland treatment area to the control terminal.

[0072] The conversion module in the control terminal converts the image information of the meibomian gland into a meibomian gland image, and converts the temperature information of the meibomian gland treatment area into the temperature value of the meibomian gland treatment area. The preprocessing module preprocesses the meibomian gland image to obtain a preprocessed meibomian gland image. The image marking module marks the area with abnormal meibomian glands in the preprocessed meibomian gland image to obtain a marked meibomian gland image, and grades the degree of variation of the meibomian glands, with each level corresponding to a color. The sending module sends the temperature value of the meibomian gland treatment area and the marked meibomian gland image to the display module, and the display module displays the temperature value of the meibomian gland treatment area and the marked meibomian gland image in real time. At the same time, the image marking module sends the variation level of the meibomian glands to the parameter control module, and the parameter control module selects the corresponding control parameters of the laser module according to the variation level of the meibomian glands to control the laser module to output laser.

[0073] At the same time, the control terminal sends the temperature value of the meibomian gland treatment area to the alarm module, and the alarm module compares the temperature value of the treatment area with the preset temperature value. When the temperature value of the meibomian gland treatment area is greater than the preset temperature value, the alarm module is controlled to issue an alarm.

[0074] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A meibomian gland laser treatment device, characterized in that: include: A treatment handle, on which a laser module, an infrared camera module and a display module are arranged, wherein the laser module is used to output laser light, the infrared camera module is used to capture image information of the meibomian glands in real time, and the display module is used to display the image of the meibomian glands in real time; The control terminal is electrically connected to the laser module, the infrared camera module and the display module respectively. The control terminal is used to receive the image information of the meibomian gland, convert the image information of the meibomian gland into a meibomian gland image, pre-process the meibomian gland image to obtain a pre-processed meibomian gland image, mark the area of ​​abnormal meibomian glands in the pre-processed meibomian gland image to obtain a marked meibomian gland image, and then send the marked meibomian gland image to the display module.

2. The meibomian gland laser treatment device according to claim 1, characterized in that: The control terminal comprises: A conversion module, used for converting the received meibomian gland image information into a meibomian gland image; A preprocessing module, used for performing denoising, enhancement and smoothing on the meibomian gland image in sequence to obtain a preprocessed meibomian gland image; An image marking module is used to mark the area where the meibomian gland is abnormal in the preprocessed meibomian gland image to obtain a marked meibomian gland image; The sending module is used to send the marked meibomian gland image to the display module.

3. The meibomian gland laser treatment device according to claim 2, characterized in that: The image marking module comprises: An input unit for receiving a preprocessed meibomian gland image; A marking model is used to mark the abnormal area of ​​the meibomian gland in the preprocessed meibomian gland image to obtain a marked meibomian gland image; The output unit is used to output the marked meibomian gland image to the sending module.

4. The meibomian gland laser treatment device according to claim 3, characterized in that: The marking model includes: An acquisition unit is used to acquire multiple meibomian gland images, and mark the meibomian gland range and each meibomian gland in each meibomian gland image as a training sample; A segmentation unit is used to segment each training sample to obtain the number, length, width, area ratio, curvature and loss ratio of the meibomian glands in each training sample; A labeling model unit is constructed to convert the number, length, width, area ratio, curvature and loss ratio of the meibomian glands in each training sample into a feature vector and its corresponding classification label. All feature vectors and their corresponding classification labels are input into the SVM model for training to obtain a test model, which is then corrected to obtain a labeling model.

5. The meibomian gland laser treatment device according to claim 4, characterized in that: The segmentation unit comprises: The MobileNet network structure is used to build a Mobile-U-Net algorithm based on the meibomian gland area image and each meibomian gland marked on each training sample, and to segment the meibomian gland area and meibomian gland in the image.

6. The meibomian gland laser treatment device according to claim 2, characterized in that: The preprocessing module comprises: De-noising unit, used to convert the meibomian gland image into a grayscale image, use Gaussian blur to reduce noise, apply Canny edge detection to quickly identify the eyelashes of the meibomian gland, apply binary threshold to detect highlight reflective points, and use dilation and erosion operations to remove unnecessary image features; An enhancement processing unit, used for improving the contrast of the denoised meibomian gland image using CLAHE; The smoothing processing unit is used to sharpen the enhanced meibomian gland image using a Laplace operator, reduce image noise, and smooth the image.

7. The meibomian gland laser treatment device according to claim 1, characterized in that: The treatment handle is provided with a manual control switch, the manual control switch is electrically connected to the laser module, and the manual control switch is used to control the start and stop of the laser module.

8. The meibomian gland laser treatment device according to claim 1, characterized in that: An infrared temperature measurement module and an alarm module are provided on the treatment handle. The infrared temperature measurement module is electrically connected to the control terminal and the display module respectively. The alarm module is electrically connected to the control terminal. The infrared temperature measurement module is used to detect the temperature information of the meibomian gland treatment area in real time and send it to the control terminal. The control terminal converts the temperature information of the meibomian gland treatment area into the temperature value of the meibomian gland treatment area, and sends it to the display module and the alarm module respectively. The display module displays the temperature value of the meibomian gland treatment area in real time. The alarm module is used to compare the temperature value of the meibomian gland treatment area with a preset temperature value. When the temperature value of the meibomian gland treatment area is greater than the preset temperature value, an alarm is triggered.

Images