Laser medical equipment control method and device, equipment and storage medium

Abstract

The invention discloses a laser medical equipment control method and device, equipment and a storage medium, and belongs to the field of laser medical equipment. Comprises: an information acquisition module for acquiring an image of a pigmented skin disease of a patient and a result of a skin pigment analyzer; the auxiliary diagnosis module is used for obtaining the type of the pigmented skin disease of the patient through image classification; the laser control module obtains a proper laser spot size, a proper laser energy density and a proper laser pulse width through a machine learning method; the laser positioning module is used for positioning the position of the pigmented skin disease of the patient through target detection; the skin feedback module is used for detecting the skin of the laser treatment part in real time through a camera; the laser treatment module positions the affected part according to the laser positioning module and treats the pigmented skin disease according to the parameters set by the laser control module. According to the invention, the accuracy and safety of laser treatment are improved by adopting a machine learning method, and the operation difficulty of laser treatment is reduced.

Description

Technical Field

[0001] The present application relates to the field of laser medical equipment, and more specifically, to a laser medical equipment control method, device, equipment and storage medium. Background Art

[0002] Laser therapy is a method of diagnosing and treating skin diseases using the physical and biological properties of lasers. Laser therapy can be applied to the treatment of pigmentary skin diseases, which are diseases that affect skin pigmentation, usually due to abnormal function or distribution of pigment cells (melanocytes). These diseases can cause the skin to appear hyperpigmented or hypopigmented, resulting in uneven or spotty skin pigmentation. Q-switched Nd:YAG lasers are lasers that use Q-switching technology to control the laser output of neodymium-doped yttrium aluminum garnet (Nd:YAG) crystals. They usually operate at a wavelength of 1064nm and can be converted to a wavelength of 532nm by a frequency doubling rod. Q-switched Nd:YAG lasers are characterized by low pulse energy (up to several millijoules), high peak power (up to several megawatts), short pulse width (up to several nanoseconds), and fixed pulse repetition rate or variable with pump power.

[0003] The document with the prior art publication number CN116578034A provides an electric control module, a laser medical device and a laser beam output control method. The electric control module is used to control the displacement of internal devices to accurately control the position of laser emission, thereby improving the position control accuracy and reducing the cost.

[0004] The above-mentioned prior art solutions only consider the position of the laser beam, but do not consider the parameters of the laser itself, such as the size of the laser spot, the energy density of the laser, and the pulse width of the laser. Different laser treatments also have different ranges.

[0005] In view of this, we propose a laser medical equipment control method, device, equipment and storage medium. Summary of the invention

[0006] 1. Technical problems to be solved

[0007] The purpose of this application is to provide a laser medical equipment control method, device, equipment and storage medium. The laser medical equipment is limited to the use of Q-switched Nd:YAG to treat pigmented skin diseases, and the machine learning method is used to help doctors determine the type of pigmented skin diseases of patients, locate the location of pigmented skin diseases, adjust the parameters of laser medical equipment, and ensure the safety of laser treatment by real-time detection of abnormalities in the patient's skin during the treatment process.

[0008] 2. Technical solution

[0009] In order to solve the above technical problems, in a first aspect, the technical solution of the present application provides a laser medical device control method, comprising the following steps:

[0010] S1. Collect images of patients' pigmented skin diseases through a camera, and classify and diagnose the types of patients' pigmented skin diseases;

[0011] S2. Detect the pigment content, pigment distribution, pigment depth and pigment color of the patient's pigmented skin disease by skin pigment analyzer;

[0012] S3, input the collected data into the machine learning regression model trained by historical successful cases to obtain appropriate laser parameters;

[0013] S4, using the trained target detection model to determine the patient's diseased location through the patient's pigmented skin disease image collected by the camera, and positioning the laser emitting device;

[0014] S5, adjusting the beam output of the laser treatment device using the obtained laser adjustment parameters and the location of the pigmented skin disease;

[0015] S6. During the treatment, the patient's skin condition is monitored in real time. When any abnormality is detected on the patient's skin, an alarm is issued in time and the treatment progress is suspended.

[0016] In an optional solution of the first aspect, the step of collecting images of the patient's pigmented skin disease through a camera and classifying the images to diagnose the type of the patient's pigmented skin disease includes: using the collected various types of pigmented skin disease images to be annotated using One-Hot encoding to form a data set, dividing the data set into a training set and a validation set, training a convolutional neural network, using ResNet as the architecture of the convolutional neural network, and using a softmax function as a classification function, where the softmax function is a function that maps any real number to the interval (0,1) and the sum is 1, and the softmax function formula is y k =exp(a k ) / ∑ n i＝1 exp(a i ),a k is the kth input signal in the output layer, y k is the kth output signal, n is the number of neurons in the output layer, and the cross entropy loss function is used as the loss function of the model. The back propagation optimization model has the formula of L = -∑ n k＝1 t k logy k , where t k is the true label of the kth category (represented by one-hot encoding), yk is the predicted probability of the kth category (obtained by the softmax function), n is the number of categories, and ten-fold cross validation is used to evaluate the F1 value to adjust the hyperparameters of the model. Pigmented skin diseases are divided into freckles, chloasma, dark spots, vitiligo, senile plaques, and melanoma. One-Hot encoding is used to represent the categories. In the diagnostic classification unit, the patient's pigmented skin disease images collected by the information acquisition unit are input into the trained image classification model to obtain the One-Hot encoding representing the patient's pigmented skin disease type. Image classification can help doctors classify patients' conditions and provide a basis for subsequent adjustment of laser parameters.

[0017] In an optional scheme of the first aspect, the step of inputting the collected data into a machine learning regression model trained by historical successful cases to obtain appropriate laser parameters includes: converting the type of pigmented skin disease, pigment content, pigment distribution, pigment depth, and pigment color in the successful laser treatment cases into data, annotated as laser parameters for laser treatment, wherein the laser parameters are spot size, energy density, and pulse width, performing data processing to form a data set, dividing the data set into a training set and a validation set to train a neural network, wherein the input layer has 5 nodes and the output layer has 3 nodes, ReLU is used as the activation function, the ReLU formula is f(x)=max(0,x), the output function is the identity function, i.e., f(x)=x, the loss function is the mean square error, and the mean square error formula is Where y is the true value, is the predicted value, n is the number of samples, and the neural network is optimized by back propagation to obtain a neural network model for multivariate regression.

[0018] In an optional solution of the first aspect, the judgment data of the patient's pigmented skin disease is combined with the analysis results of the skin pigment analyzer for data processing and then input into the neural network model to obtain the laser parameters of the laser medical equipment.

[0019] Through the above technical solution, by learning from historical successful cases, the parameters of laser treatment equipment can be adjusted, the accuracy of laser treatment can be improved, doctors can be assisted in treating patients' pigmented skin diseases, and the parameters of laser medical equipment can be adjusted.

[0020] In an optional scheme of the first aspect, the step of using a trained target detection model to determine the patient's diseased location through the patient's pigmented skin disease image collected by a camera, and positioning the laser emitting device includes: training the target detection model YOlOv5 after edge annotation of the collected different types of pigmented skin disease images, storing the images in the images and labels folders according to the requirements of YOLOv5, configuring the yaml file, selecting YOLOv5s to train with the configured data set, and obtaining a target detection model that can detect the location of the pigmented skin disease.

[0021] In an optional solution of the first aspect, the information acquisition module collects the patient's pigmented skin disease image and inputs it into a trained YOLOv5s model to confirm the location of the patient's pigmented skin disease and provide positioning information for laser treatment.

[0022] In an optional scheme of the first aspect, the step of detecting the patient's skin condition in real time during the treatment process includes: collecting images of abnormal reactions of the patient's skin during laser treatment, configuring the yaml file according to the requirements of YOLOv5 to support the data set, so that the trained YOLOv5 model can detect abnormal reactions of the patient's skin during laser treatment, and detecting whether the laser treatment has caused abnormal skin reactions by using the patient's skin images obtained in real time by the camera.

[0023] Through the above technical solution, machine learning methods are used to assist laser treatment, helping doctors to diagnose the condition, formulate treatment plans, control laser treatment equipment, reduce the operational difficulty of laser treatment, and improve the efficiency of treatment.

[0024] In a second aspect, the technical solution of the present application provides a laser medical equipment control device, comprising:

[0025] An information collection module, which collects images of the patient's pigmented skin disease and results of a skin pigment analyzer;

[0026] An auxiliary diagnosis module, which classifies the patient's pigmented skin disease by image classification to obtain the type of the patient's pigmented skin disease;

[0027] A laser control module, which obtains a suitable laser spot size, laser energy density, and laser pulse width through a machine learning method based on the image classification result, the pigment depth and distribution of the skin disease, the pigment property and density, and the patient's skin type;

[0028] A laser positioning module, wherein the laser positioning module locates the position of the patient's pigmented skin disease through target detection and uses the edge of the pigmented skin disease to locate the pigmented skin disease;

[0029] A skin feedback module, which detects the laser treatment part in real time through a camera, and when the skin shows abnormal reaction, an alarm is issued and the treatment is stopped;

[0030] The laser treatment module locates the affected area according to the laser positioning module and treats the pigmented skin disease according to the parameters set by the laser control module.

[0031] As an optional solution of the present application document, the information acquisition module includes an image acquisition unit, which acquires the patient's pigmented skin disease image, and a data acquisition unit, which obtains the analysis result of the skin pigmentation analyzer through communication.

[0032] As an optional solution of the present application document, the auxiliary diagnosis module includes: a diagnosis training unit, the disease classification training unit trains a convolutional neural network after annotating images of different types of pigmented skin diseases, and obtains an image classification model that can classify pigmented skin diseases, and the classification categories can be freckles, chloasma, black spots, vitiligo, age spots, melanoma, etc. A diagnosis classification unit, the diagnosis classification unit inputs the image of the patient's pigmented skin disease obtained by the information acquisition module into the trained image classification model to obtain the type of the patient's pigmented skin disease.

[0033] As an optional solution of the present application document, the laser control module includes: a regression training unit, which trains a machine learning regression model through different disease types (freckles, chloasma, vitiligo), pigment content, pigment distribution, pigment depth, and pigment color marking corresponding to the laser spot size, laser energy density, and laser pulse width. A parameter determination unit, which inputs the results of the information acquisition module and the auxiliary diagnosis module into the machine learning regression model to obtain a suitable spot size, laser energy density, and laser pulse width.

[0034] As an optional solution of the present application document, the laser positioning module includes: a positioning training unit, the positioning training module uses the patient's pigmented skin disease image to mark the position of the pigmented skin disease to make a data set, and uses the marked data set to train the target detection model to obtain a target detection model that can locate the position of the pigmented skin disease. A position determination unit performs target detection on the patient's pigmented skin disease image collected by a camera, locates the position of the pigmented skin disease, and enables the laser emitting device to align with the position of the pigmented skin disease.

[0035] As an optional solution of the present application document, the skin feedback module includes: a feedback training unit, which also uses a target detection algorithm to annotate the collected images of abnormal conditions caused by laser stimulation of the skin to form a data set, and uses the annotated data set to train the target detection model to obtain a target detection model that can detect abnormal skin reactions. An abnormality detection unit, which obtains the patient's skin image in real time through a camera during treatment, and when an abnormal reaction of the skin is detected, an alarm is issued in time and the treatment progress is suspended.

[0036] As an optional solution of the present application document, the laser treatment module includes a laser emitting device and a corresponding adjustment device, using a Q-switch Nd:YAG, and the laser parameters (spot size, energy density, pulse width) are controlled by the results output by the laser control module. Treatment is performed based on the location of the patient's pigmented skin disease located by the laser positioning module, and a timely response is made based on the results of the skin feedback module.

[0037] In a third aspect, the present application further provides a computer device, comprising a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, the laser medical device control method as described in the first aspect is implemented.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the laser medical equipment control method as described in the first aspect.

[0039] 3. Beneficial effects

[0040] One or more technical solutions provided in the technical solution of this application have at least the following technical effects or advantages:

[0041] Help doctors diagnose the type of pigmented skin disease through image classification, and analyze the laser parameters suitable for patients by learning from historical successful cases, thereby increasing the accuracy and success rate of laser treatment;

[0042] Through the laser positioning module, the target detection is used to automatically align the laser to the position of the patient's pigmented skin disease, which increases the accuracy of the laser treatment position, reduces the risks that may be caused by manual operation by doctors, avoids damage to normal skin due to position reasons, and reduces the difficulty of laser treatment operation;

[0043] Through the skin feedback module, target detection is used to monitor the patient's skin in real time to ensure that the patient's skin reaction can be discovered in time. When an abnormal reaction occurs, the doctor can be reminded and the treatment can be suspended in time to prevent the laser from causing unnecessary damage to the patient's skin and ensure the safety of laser treatment.

[0044] The laser treatment module cooperates with multiple modules to increase the accuracy and safety of laser treatment, provide assistance to doctors in diagnosis and treatment, reduce doctors' workload and improve work efficiency;

[0045] By using machine learning methods to assist laser treatment, doctors can be helped to diagnose the condition, develop treatment plans, control laser treatment equipment, reduce the operational difficulty of laser treatment, and improve the efficiency of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is an overall schematic diagram of a laser medical equipment control method disclosed in a preferred embodiment of the present application;

[0047] Figure 2 This is a structural diagram of an auxiliary diagnosis module of a laser medical equipment control device disclosed in a preferred embodiment of the present application;

[0048] Figure 3 A structure diagram of a laser control module of a laser medical equipment control device disclosed in a preferred embodiment of the present application;

[0049] Figure 4 This is a structural diagram of a laser positioning module of a laser medical equipment control device disclosed in a preferred embodiment of the present application;

[0050] Figure 5 This is a structural diagram of a skin feedback module of a laser medical equipment control device disclosed in a preferred embodiment of the present application; DETAILED DESCRIPTION

[0051] The present application is further described in detail below in conjunction with the accompanying drawings.

[0052] Reference Figure 1 , the embodiment of the present application provides a laser medical equipment control method, comprising the following steps:

[0053] S1. Collect images of patients' pigmented skin diseases through a camera, and classify and diagnose the types of patients' pigmented skin diseases;

[0054] S2. Detect the pigment content, pigment distribution, pigment depth and pigment color of the patient's pigmented skin disease by skin pigment analyzer;

[0055] S3, input the collected data into the machine learning regression model trained by historical successful cases to obtain appropriate laser parameters;

[0056] S4, using the trained target detection model to determine the patient's diseased location through the patient's pigmented skin disease image collected by the camera, and positioning the laser emitting device;

[0057] S5, adjusting the beam output of the laser treatment device using the obtained laser adjustment parameters and the location of the pigmented skin disease;

[0058] S6. During the treatment, the patient's skin condition is monitored in real time. When any abnormality is detected on the patient's skin, an alarm is issued in time and the treatment progress is suspended.

[0059] The steps of collecting images of patients with pigmented skin diseases through a camera and classifying and diagnosing the types of patients with pigmented skin diseases include: using the collected images of various types of pigmented skin diseases to annotate using One-Hot encoding to form a data set, dividing the data set into a training set and a validation set, training a convolutional neural network, using ResNet as the architecture of the convolutional neural network, and using the softmax function as the classification function. The softmax function is a function that maps any real number to the interval (0,1) and the sum is 1. The formula of the softmax function is y k =exp(a k ) / ∑ n i＝1 exp(a i ),a k is the kth input signal in the output layer, y k is the kth output signal, n is the number of neurons in the output layer, and the cross entropy loss function is used as the loss function of the model. The back propagation optimization model has the formula of L = -∑ n k＝1 t k logy k , where t k is the true label of the kth category (represented by one-hot encoding), y k is the predicted probability of the kth category (obtained by the softmax function), n is the number of categories, and ten-fold cross validation is used to evaluate the F1 value to adjust the hyperparameters of the model. Pigmented skin diseases are divided into freckles, chloasma, dark spots, vitiligo, senile plaques, and melanoma. One-Hot encoding is used to represent the categories. In the diagnostic classification unit, the patient's pigmented skin disease images collected by the information acquisition unit are input into the trained image classification model to obtain the One-Hot encoding representing the patient's pigmented skin disease type. Image classification can help doctors classify patients' conditions and provide a basis for subsequent adjustment of laser parameters.

[0060] The steps of inputting the collected data into the machine learning regression model trained by historical successful cases to obtain appropriate laser parameters include: converting the type of pigmented skin disease, pigment content, pigment distribution, pigment depth, and pigment color in the successful laser treatment cases into data, annotating them as laser parameters for laser treatment, where the laser parameters are spot size, energy density, and pulse width, processing the data to make a data set, dividing the data set into a training set and a validation set to train the neural network, with 5 nodes in the input layer and 3 nodes in the output layer, using ReLU as the activation function, the ReLU formula is f(x)=max(0,x), the output function is the identity function, i.e., f(x)=x, the loss function is the mean square error, and the mean square error formula is Where y is the true value, is the predicted value, n is the number of samples, and the neural network is optimized by back propagation to obtain a neural network model for multivariate regression.

[0061] The judgment data of the patient's pigmented skin disease is combined with the analysis results of the skin pigment analyzer for data processing and then input into the neural network model to obtain the laser parameters of the laser medical equipment.

[0062] The trained target detection model is used to determine the patient's diseased location through the patient's pigmented skin disease images collected by the camera. The steps of positioning the laser emitting device include: training the target detection model YOlOv5 after edge annotation of the collected different types of pigmented skin disease images, storing the images in the images and labels folders according to the requirements of YOLOv5, configuring the yaml file, selecting YOLOv5s and using the configured data set for training, and obtaining a target detection model that can detect the location of pigmented skin diseases.

[0063] By collecting the patient's pigmented skin disease images through the information acquisition module and inputting them into the trained YOLOv5s model, the location of the patient's pigmented skin disease can be confirmed, providing positioning information for laser treatment.

[0064] During the treatment process, the steps of real-time detection of the patient's skin condition include: collecting images of abnormal skin reactions during laser treatment, configuring the yaml file according to the requirements of YOLOv5 to support the data set, so that the trained YOLOv5 model can detect abnormal skin reactions during laser treatment, and using the patient's skin images obtained in real time by the camera to detect whether the laser treatment has caused abnormal skin reactions.

[0065] The embodiment of the present application also provides a laser medical equipment control device, comprising: an information acquisition module, the information acquisition module collects images of the patient's pigmented skin disease and the results of the skin pigment analyzer. An auxiliary diagnosis module, the auxiliary diagnosis module classifies the patient's pigmented skin disease through image classification, and obtains the type of the patient's pigmented skin disease, such as vitiligo, freckles, chloasma, age spots, etc. A laser control module, the laser control module obtains the appropriate laser spot size, laser energy density, and laser pulse width through a machine learning method based on the results of image classification (type of skin disease), the pigment depth and distribution of the skin disease, the pigment properties and density, and the patient's skin type. A laser positioning module, the laser positioning module locates the position of the patient's pigmented skin disease through target detection, and uses the edge of the pigmented skin disease to locate the pigmented skin disease. A skin feedback module, the skin feedback module performs real-time detection of the laser treatment part through a camera, and when the skin has an abnormal reaction, an alarm is issued and the treatment is suspended. A laser treatment module, the laser treatment module locates the patient's part according to the laser positioning module, and treats the pigmented skin disease according to the parameters set by the laser control module.

[0066] The auxiliary diagnosis module and the laser positioning module analyze the patient's pigmented skin disease images collected by the information collection module to obtain the location and type of the patient's pigmented skin disease. According to the type of the patient's pigmented skin disease combined with the results of the skin pigment analyzer collected by the information collection module, the laser control module obtains the laser parameters (spot size, energy density, pulse width) suitable for the patient. The laser control module controls the laser treatment module according to the parameters. During the treatment, the patient's skin is monitored in real time. The skin feedback module detects whether the patient's skin is abnormal in real time. When the patient has skin abnormalities, an alarm is issued and the treatment is suspended in time.

[0067] Reference Figure 1 and Figure 2 The embodiment of the present application provides a laser medical equipment control device, wherein the auxiliary diagnosis module includes: a diagnosis training unit, which uses various types of pigmented skin disease images collected and annotated using One-Hot coding to form a data set, divides the data set into a training set and a validation set, trains a convolutional neural network, uses ResNet as the architecture of the convolutional neural network, and uses a softmax function as a classification function. The softmax function is a function that maps any real number to the interval (0,1) and the sum is 1. The formula of the softmax function is y k =exp(a k ) / ∑ n i＝1 exp(a i ),a k is the kth input signal in the output layer, y kis the kth output signal, and n is the number of neurons in the output layer. The cross entropy loss function is used as the loss function of the model, and the back propagation optimization model is optimized. The cross entropy loss function formula is L = -∑ n k＝1 t k logy k , where t k is the true label of the kth category (represented by one-hot encoding), y k is the predicted probability of the kth category (obtained by the softmax function), and n is the number of categories. Using ten-fold cross validation, the F1 value is evaluated to adjust the model's hyperparameters, and pigmented skin diseases are divided into freckles, chloasma, dark spots, vitiligo, senile plaques, and melanoma, using one-hot encoding to represent the categories. The diagnostic classification unit inputs the patient's pigmented skin disease images collected by the information collection unit into the trained image classification model to obtain a one-hot encoding representing the patient's pigmented skin disease type. Image classification can help doctors classify the patient's condition and provide a basis for the subsequent adjustment of laser parameters.

[0068] Reference Figure 1 and Figure 3 , the embodiment of the present application provides a laser medical equipment control device, the laser control module includes: a regression training unit, which converts the type of pigmented skin disease, pigment content, pigment distribution, pigment depth, and pigment color in the successful laser treatment case into data, annotated as laser parameters (spot size, energy density, pulse width) of laser treatment, performs data processing (standardization) to make a data set, divides the data set into a training set and a validation set to train a neural network, the input layer has 5 nodes, the output layer has 3 nodes, uses ReLU as the activation function, the ReLU formula is f(x)=max(0,x), the output function is the identity function, that is, f(x)=x, the loss function is the mean square error, and the mean square error formula is Where y is the true value, is the predicted value, and n is the number of samples. Back propagation optimizes the neural network to obtain a neural network model for multivariate regression. The parameter determination unit uses the auxiliary diagnosis module to judge the patient's pigmented skin disease and the analysis results of the skin pigment analyzer for data processing and then inputs the data into the neural network model to obtain the laser parameters (spot size, energy density, pulse width) of the laser medical device. For example, in a freckle case, the output spot size is 3.024 mm and the energy density is 7.382 J / cm 2 , the pulse width is 6ns.

[0069] Reference Figure 1 and Figure 4The embodiment of the present application provides a laser medical equipment control device, wherein the laser positioning module includes: a positioning training unit, which uses the collected different types of pigmented skin disease images to perform edge annotation and then trains the target detection model YOlOv5, stores the images in the images and labels folders according to the requirements of YOLOv5, and configures the yaml file, selects YOLOv5s to use the configured data set for training, and obtains a target detection model that can detect the location of the pigmented skin disease. A position determination unit, which inputs the patient's pigmented skin disease image collected by the information acquisition module into the trained YOLOv5s model, confirms the location of the patient's pigmented skin disease, and provides positioning information for laser treatment.

[0070] Reference Figure 1 and Figure 5 The embodiment of the present application provides a laser medical equipment control device, wherein the skin feedback module includes: a feedback training unit, which uses the collected abnormal skin reaction images of patients during laser treatment to support the data set according to the requirements of YOLOv5, and configures the yaml file so that the trained YOLOv5 model can detect abnormal skin reactions of patients during laser treatment, such as abnormal redness and swelling. An abnormality detection unit, which detects whether the laser treatment causes abnormal skin reactions through the patient skin images obtained in real time by the camera.

[0071] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0072] An embodiment of the present application provides a computer program product. When the computer program product is run on a computer device, the computer device implements the steps in the above-mentioned method embodiments when executing the computer device.

[0073] When the laser medical equipment control device disclosed in the embodiment of the present application is used, the corresponding model is first trained through the collected image data, and the image classification model for determining the type of disease is trained using images of different types of pigmented skin diseases, and the target detection model for locating the location of the disease is trained. The machine learning regression model is trained using the disease type combined with the results of analysis by a skin pigment analyzer. The target detection model for detecting abnormalities in the patient's skin is trained using the collected images of abnormalities in the patient's skin during the laser treatment process. These models are deployed in different modules. When in use, the image of the patient's pigmented skin disease is collected by a camera to determine the disease type, disease location, and parameters required for the laser, and the patient's skin condition is detected in real time during the treatment process to ensure patient safety.

Claims

1. A laser medical equipment control method, characterized in that: The following steps are involved: S1. Collect images of patients' pigmented skin diseases through a camera, and classify and diagnose the types of patients' pigmented skin diseases; S2. Detect the pigment content, pigment distribution, pigment depth and pigment color of the patient's pigmented skin disease by skin pigment analyzer; S3, input the collected data into the machine learning regression model trained by historical successful cases to obtain appropriate laser parameters; S4, using the trained target detection model to determine the patient's diseased location through the patient's pigmented skin disease image collected by the camera, and positioning the laser emitting device; S5, adjusting the beam output of the laser treatment device using the obtained laser adjustment parameters and the location of the pigmented skin disease; S6. During the treatment, the patient's skin condition is monitored in real time. When any abnormality is detected on the patient's skin, an alarm is issued in time and the treatment progress is suspended.

2. The laser medical equipment control method according to claim 1, characterized in that: The step of collecting images of patients with pigmented skin diseases through a camera and classifying the images to diagnose the types of patients with pigmented skin diseases includes: using the collected images of various types of pigmented skin diseases to be annotated using One-Hot encoding to form a data set, dividing the data set into a training set and a validation set, training a convolutional neural network, using ResNet as the architecture of the convolutional neural network, and using a softmax function as a classification function. The softmax function is a function that maps any real number to the interval (0,1) and the sum is 1. The formula of the softmax function is y k =exp(a k ) / ∑ n i＝1 exp(a i ),a k is the kth input signal in the output layer, y k is the kth output signal, n is the number of neurons in the output layer, and the cross entropy loss function is used as the loss function of the model. The back propagation optimization model has the formula of L = -∑ n k＝1 t k logy k , where t k is the true label of the kth category (represented by one-hot encoding), y k is the predicted probability of the kth category (obtained by the softmax function), n is the number of categories, and ten-fold cross validation is used to evaluate the F1 value to adjust the hyperparameters of the model. Pigmented skin diseases are divided into freckles, chloasma, dark spots, vitiligo, senile plaques, and melanoma. One-Hot encoding is used to represent the categories. In the diagnostic classification unit, the patient's pigmented skin disease images collected by the information acquisition unit are input into the trained image classification model to obtain the One-Hot encoding representing the patient's pigmented skin disease type. Image classification can help doctors classify patients' conditions and provide a basis for subsequent adjustment of laser parameters.

3. The laser medical equipment control method according to claim 1, characterized in that: The step of inputting the collected data into a machine learning regression model trained by historical successful cases to obtain appropriate laser parameters includes: converting the type of pigmented skin disease, pigment content, pigment distribution, pigment depth, and pigment color in the successful laser treatment cases into data, annotating them as laser parameters for laser treatment, wherein the laser parameters are spot size, energy density, and pulse width, performing data processing to produce a data set, dividing the data set into a training set and a validation set to train a neural network, wherein the input layer has 5 nodes and the output layer has 3 nodes, ReLU is used as the activation function, the ReLU formula is f(x)=max(0,x), the output function is the identity function, i.e., f(x)=x, the loss function is the mean square error, and the mean square error formula is Where y is the true value, is the predicted value, n is the number of samples, and the neural network is optimized by back propagation to obtain a neural network model for multivariate regression.

4. The laser medical equipment control method according to claim 3, characterized in that: The judgment data of the patient's pigmented skin disease is combined with the analysis results of the skin pigment analyzer for data processing and then input into the neural network model to obtain the laser parameters of the laser medical equipment.

5. The laser medical equipment control method according to claim 1, characterized in that: The step of using the trained target detection model to determine the patient's diseased location through the patient's pigmented skin disease image collected by the camera and positioning the laser emitting device includes: training the target detection model YOlOv5 after edge annotation of the collected different types of pigmented skin disease images, storing the images in the images and labels folders according to the requirements of YOLOv5, configuring the yaml file, selecting YOLOv5s to train with the configured data set, and obtaining a target detection model that can detect the location of the pigmented skin disease.

6. The laser medical equipment control method according to claim 5, characterized in that: By collecting the patient's pigmented skin disease images through the information acquisition module and inputting them into the trained YOLOv5s model, the location of the patient's pigmented skin disease can be confirmed, providing positioning information for laser treatment.

7. The laser medical equipment control method according to claim 1, characterized in that: During the treatment, the step of real-time detection of the patient's skin condition includes: collecting images of abnormal skin reactions during laser treatment, configuring a YOLOv5 support data set according to the requirements of YOLOv5, so that the trained YOLOv5 model can detect abnormal skin reactions during laser treatment, and detecting whether the laser treatment has caused abnormal skin reactions by using the patient's skin images obtained in real time by the camera.

8. A laser medical equipment control device, characterized in that: include: An information collection module, which collects images of the patient's pigmented skin disease and results of a skin pigment analyzer; An auxiliary diagnosis module, which classifies the patient's pigmented skin disease by image classification to obtain the type of the patient's pigmented skin disease; A laser control module, which obtains a suitable laser spot size, laser energy density, and laser pulse width through a machine learning method based on the image classification result, the pigment depth and distribution of the skin disease, the pigment property and density, and the patient's skin type; A laser positioning module, wherein the laser positioning module locates the position of the patient's pigmented skin disease through target detection and uses the edge of the pigmented skin disease to locate the pigmented skin disease; A skin feedback module, which detects the laser treatment part in real time through a camera, and when the skin shows abnormal reaction, an alarm is issued and the treatment is stopped; The laser treatment module locates the affected area according to the laser positioning module and treats the pigmented skin disease according to the parameters set by the laser control module.

9. A computer device, characterized in that: The method comprises a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, the laser medical equipment control method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements the laser medical equipment control method as described in any one of claims 1 to 7.

Images