CN114618090A - Energy control method of strong pulse laser xerophthalmia therapeutic instrument - Google Patents
Energy control method of strong pulse laser xerophthalmia therapeutic instrument Download PDFInfo
- Publication number
- CN114618090A CN114618090A CN202210247772.XA CN202210247772A CN114618090A CN 114618090 A CN114618090 A CN 114618090A CN 202210247772 A CN202210247772 A CN 202210247772A CN 114618090 A CN114618090 A CN 114618090A
- Authority
- CN
- China
- Prior art keywords
- pulse laser
- layer
- dry eye
- strong pulse
- rbm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000001225 therapeutic effect Effects 0.000 title claims abstract description 22
- 208000005494 xerophthalmia Diseases 0.000 title claims abstract description 16
- 208000003556 Dry Eye Syndromes Diseases 0.000 claims abstract description 48
- 206010013774 Dry eye Diseases 0.000 claims abstract description 48
- 238000011282 treatment Methods 0.000 claims abstract description 32
- 230000000694 effects Effects 0.000 claims abstract description 11
- 230000003631 expected effect Effects 0.000 claims abstract description 11
- 230000002159 abnormal effect Effects 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 28
- 210000003491 skin Anatomy 0.000 claims description 26
- 210000002569 neuron Anatomy 0.000 claims description 24
- 230000000007 visual effect Effects 0.000 claims description 19
- 230000037325 pain tolerance Effects 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 7
- 210000000434 stratum corneum Anatomy 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000012790 confirmation Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 206010048245 Yellow skin Diseases 0.000 claims description 2
- 230000000630 rising effect Effects 0.000 claims description 2
- 239000000126 substance Substances 0.000 claims description 2
- 230000009471 action Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 2
- 150000002632 lipids Chemical class 0.000 description 2
- 210000004175 meibomian gland Anatomy 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 241000238876 Acari Species 0.000 description 1
- 241000894006 Bacteria Species 0.000 description 1
- 206010011732 Cyst Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 208000031513 cyst Diseases 0.000 description 1
- 230000002888 effect on disease Effects 0.000 description 1
- 208000008025 hordeolum Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000028327 secretion Effects 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N5/00—Radiation therapy
- A61N5/06—Radiation therapy using light
- A61N5/0613—Apparatus adapted for a specific treatment
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61F—FILTERS IMPLANTABLE INTO BLOOD VESSELS; PROSTHESES; DEVICES PROVIDING PATENCY TO, OR PREVENTING COLLAPSING OF, TUBULAR STRUCTURES OF THE BODY, e.g. STENTS; ORTHOPAEDIC, NURSING OR CONTRACEPTIVE DEVICES; FOMENTATION; TREATMENT OR PROTECTION OF EYES OR EARS; BANDAGES, DRESSINGS OR ABSORBENT PADS; FIRST-AID KITS
- A61F9/00—Methods or devices for treatment of the eyes; Devices for putting-in contact lenses; Devices to correct squinting; Apparatus to guide the blind; Protective devices for the eyes, carried on the body or in the hand
Landscapes
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Heart & Thoracic Surgery (AREA)
- Vascular Medicine (AREA)
- Ophthalmology & Optometry (AREA)
- Pathology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Radiation-Therapy Devices (AREA)
Abstract
The invention belongs to the technical field of xerophthalmia treatment, and particularly discloses an energy control method of a strong pulse laser xerophthalmia treatment instrument, which comprises the following steps: s1: collecting data; s2: data input: inputting patient data into a dry eye data prediction model; s3: and (3) data output: the dry eye data prediction model outputs relevant information of the strong pulse laser; s4: the treatment application is as follows: applying the relevant information of the output strong pulse laser to the strong pulse laser xerophthalmia therapeutic instrument; s5: abnormal feedback: if the dry eye treatment effect based on the strong pulse laser does not reach the expected effect, feeding the acquired data back to the dry eye data prediction model; s6: and (3) adjusting again: and carrying out fine adjustment on the parameters. The method can solve the problems that the judgment and selection of relevant characteristics of the strong pulse laser are carried out by the experience of medical staff, the requirements on the skill, the experience and the like of the medical staff are high, the application of the strong pulse laser in the field of dry eye treatment is not facilitated, and the automation degree is low.
Description
Technical Field
The invention belongs to the technical field of xerophthalmia treatment, and particularly relates to an energy control method of a strong pulse laser xerophthalmia treatment instrument.
Background
A strong pulsed laser, abbreviated by IPL in english, is a broad spectrum visible light. The IPL is based on the selective photothermal action principle, namely, light with longer wavelength in the output intense pulse light can penetrate through deeper tissues of the skin to generate photothermal action and photochemical action, so that the IPL plays a role in improving the function of meibomian glands, can soften lipid, promote the secretion of the lipid, can kill mites and partial bacteria, and has a treatment effect on diseases such as meibomian gland cyst, hordeolum and the like.
At present, medical staff judge the intensity, time and energy waveform of the strong pulse laser according to own experience and adjust the intensity, time and energy waveform according to treatment results. The judgment and selection of the relevant characteristics of the strong pulse laser are carried out by the experience of medical staff, the requirements on the skill, the experience and the like of the medical staff are high, the application of the strong pulse laser in the field of dry eye treatment is not facilitated, and the automation degree is low.
Disclosure of Invention
The invention aims to provide an energy control method of a strong pulse laser xerophthalmia therapeutic instrument, which aims to solve the problems that the judgment and selection of relevant characteristics of the strong pulse laser are carried out by the experience of medical staff, the requirements on the skills, the experience and the like of the medical staff are high, the application of the strong pulse laser in the xerophthalmia therapeutic field is not facilitated, and the automation degree is low.
In order to achieve the purpose, the technical scheme of the invention is as follows: an energy control method of a strong pulse laser dry eye therapeutic apparatus comprises the following steps:
s1: data acquisition: collecting patient data including patient skin tone, stratum corneum thickness, and pain tolerance;
s2: data input: inputting patient data into a deep learning based dry eye data prediction model;
s3: and (3) data output: the dry eye data prediction model outputs relevant information of the strong pulse laser, wherein the relevant information comprises an energy waveform of the strong pulse laser, duration of each sub-pulse in the strong pulse laser and energy intensity of each sub-pulse in the strong pulse laser;
s4: the treatment application is as follows: and the relevant information of the output strong pulse laser is applied to the strong pulse laser xerophthalmia therapeutic apparatus, so that the strong pulse laser emitted by the strong pulse laser xerophthalmia therapeutic apparatus conforms to the relevant information.
Further, in step S2, the establishment of the dry eye data prediction model specifically includes the following steps:
s2.1: calibrating a sample: the sample adopts the processed relevant data of each patient, and the relevant data comprises the skin color of the patient, the thickness of the stratum corneum, the pain tolerance, the energy waveform of the strong pulse laser in treatment, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser; calibrating the sample with the treatment effect reaching the expected effect, and not calibrating the sample with the treatment effect not reaching the expected effect;
s2.2: confirmation of input and output quantities: the skin color, the thickness of the cuticle layer and the pain tolerance of a patient are used as input quantities, and the energy waveform of the strong pulse laser, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser during treatment are used as output quantities;
s2.3: training data set generation: embedding the skin color, the thickness of the cuticle layer and the pain tolerance of each patient in the calibration sample into RGB components of the picture, and converting the RGB components into an information image; preprocessing an information image, generating a target information segment with the same picture size as a training data set;
s2.4: establishing a dry eye data prediction model based on a DBN structure: the method comprises the steps that a deep confidence network is adopted as an algorithm of a dry eye data prediction model, the deep confidence network comprises a plurality of RBM models, each RBM model comprises a visible layer and a hidden layer which are connected in a bidirectional mode, a training data set is used as an input vector of a first RBM model, and the first RBM model is trained; taking the hidden layer of the trained first RBM as an input vector of the visible layer of the second RBM, and continuing to train the second RBM; taking the hidden layer of the trained second RBM as an input vector of the visible layer of the third RBM, and continuing to train the third RBM; sequentially training until all RBM models are trained, connecting the hidden layer of the last RBM model to the output layer, and outputting a prediction result;
s2.5: parameter adjustment: after the unsupervised learning process is completed, performing supervised joint training on the DBN structure according to the label attached to the top layer of the network, and finely adjusting parameters of all network layer structures according to a prediction result to form a dry eye data prediction model.
Further, the method also comprises the following steps:
s5: abnormal feedback: if the dry eye treatment effect by the strong pulse laser does not reach the expected effect, feeding back the data of the patient collected in the step S1 to the dry eye data prediction model;
s6: and (3) adjusting again: and fine-tuning parameters of the dry eye data prediction model through supervised training again.
Further, in step S2.4, each RBM model includes n visible layers and m hidden layers, and the parameter θ is expressed as θ ═ ai,bj,wijIn which wijRepresents a connection weight; among them are:
v＝(v1,v2,…,vi,…,vn) Is a visual layer state, viIs the ith neuronal state;
h＝(h1,h2,…,hj,…,hm) For hiding layer state, hjIs the jth neuron state;
a＝(a1,a2,…,ai,…,an) For visual layer biasing, aiBiasing for the ith neuron;
b＝(b1,b2,…,bj,…,bm) For hidden layer biasing, bjBiasing for the jth neuron;
the energy function of the RBM model is:
from the above formula, the function value is related to the values of all neurons in the visible layer and the hidden layer, and the joint probability density of v and h is defined as:
the edge probability density of the visual layer is obtained by summing all hidden layer neurons:
the edge probability density of the hidden layer is obtained by summing all visible layer neurons:
the probability of neuron i being activated in the visual layer is:
the probability of neuron j being activated in the hidden layer is:
wherein, σ (a)i+∑jwijhj) And σ (b)j+∑iviwij) Is an activation function;
to maximize the probability distribution of the visual layer v, i.e. to maximize the likelihood function according to the training data set, as follows:
solving the maximum value of the likelihood function by a random gradient rising method, and solving the partial derivative of the parameter theta to obtain:
wherein the content of the first and second substances,<vihj>p(h|v)indicating the expectation of the data distribution p (h | v),<vihj>modelrepresents the expectation of the model distribution p (v, h);
according to the contrast divergence algorithm, the update criteria that can be derived for the weights and biases are as follows:
Δωij＝α(<vihj>p(h|v)-<vihj>recon) (9)
Δai＝α(<vi>p(h|v)-<vi>recon) (10)
Δbj＝α(<hj>p(h|v)-<hj>ercon) (11)
wherein, alpha represents the learning rate,<vihj>recon、<vi>reconand<hj>reconrespectively representing the expectations for p (v, h), p (v), and p (h) under the reconstructed model distribution;
and obtaining an updated parameter theta according to the updated weight and the offset.
Further, the thickness of the stratum corneum includes thick, medium, and thin.
Further, pain tolerance includes levels 1, 2, 3, 4 and 5, with level 1 being the weakest and level 5 being the strongest.
Further, the patient's skin color includes black, yellow and white skin.
Further, in step S6 and step S2.5, the parameters of all network layer structures are fine-tuned by using a gradient descent method.
The beneficial effects of this technical scheme lie in: according to the scheme, self data (skin color, thickness and pain tolerance of a patient) of the patient are used as input factors, a dry eye data prediction model is used for prediction processing, strong pulse laser which accords with the dry eye treatment of the patient is output, and the final strong pulse laser is generated from three aspects of waveform of the strong pulse laser, duration of each sub-pulse and energy intensity of each sub-pulse, so that judgment according to self experience and skill of medical staff is replaced, and the problems of low automation degree and inaccurate judgment are solved.
Drawings
FIG. 1 is a flow chart of a method for controlling the energy of a strong pulse laser dry eye therapeutic apparatus according to the present invention;
FIG. 2 is a flow chart of dry eye data prediction model building;
FIG. 3 is a schematic diagram of a first waveform;
FIG. 4 is a schematic diagram of a second waveform;
fig. 5 is a diagram illustrating a third waveform.
Detailed Description
The following is further detailed by way of specific embodiments:
the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment is basically as shown in the attached figure 1: an energy control method of a strong pulse laser dry eye therapeutic apparatus comprises the following steps:
s1: data acquisition: collecting patient data including patient skin tone, stratum corneum thickness, and pain tolerance; the skin color of the patient comprises black skin, yellow skin and white skin, and the thickness of the horny layer comprises thick, medium and thin; pain tolerance includes grade 1, 2, 3, 4 and 5, with grade 1 being the weakest and grade 5 being the strongest; the skin color of the patient is determined by observing the skin of the patient by medical staff, the pain tolerance is determined by inquiring the condition of the patient by the medical staff, the thickness of the cuticle layer is detected by adopting a skin detector, and the medical staff divides the boundary of the thickness, the middle thickness and the thin thickness;
s2: data input: inputting patient data into a deep learning based dry eye data prediction model; as shown in fig. 2, the establishment of the dry eye data prediction model specifically includes the following steps:
s2.1: sample calibration: the sample adopts the processed relevant data of each patient, and the relevant data comprises the skin color, the thickness of the cuticle layer, the pain tolerance of the patient, the energy waveform of the strong pulse laser in treatment, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser; calibrating the sample with the treatment effect reaching the expected effect, and not calibrating the sample with the treatment effect not reaching the expected effect; in the present case, there are three energy waveforms of the intense pulse laser, i.e., the first one in fig. 3, the second one in fig. 4, and the third one in fig. 5;
s2.2: confirmation of input and output quantities: the skin color, the thickness of the cuticle layer and the pain tolerance of a patient are used as input quantities, and the energy waveform of the strong pulse laser, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser during treatment are used as output quantities;
s2.3: training data set generation: embedding the skin color, the thickness of the cuticle layer and the pain tolerance of each patient in the calibration sample into RGB components of the picture, and converting the RGB components into information images; preprocessing an information image, generating a target information segment with the same picture size as a training data set;
s2.4: establishing a dry eye data prediction model based on a DBN structure: the method comprises the steps that a deep confidence network is adopted as an algorithm of a dry eye data prediction model, the deep confidence network comprises a plurality of RBM models, each RBM model comprises a visible layer and a hidden layer which are connected in a bidirectional mode, a training data set is used as an input vector of a first RBM model, and the first RBM model is trained; taking the hidden layer of the trained first RBM as an input vector of the visible layer of the second RBM, and continuing to train the second RBM; taking the hidden layer of the trained second RBM as an input vector of the visible layer of the third RBM, and continuing to train the third RBM; sequentially training until all RBM models are trained, connecting the hidden layer of the last RBM model to the output layer, and outputting a prediction result;
each RBM model comprises n visible layers and m hidden layers, and a parameter theta is recorded as theta ═ ai,bj,wijIn which wijRepresents a connection weight; among them are:
v＝(v1,v2,…,vi,…,vn) Is a visual layer state, viIs the ith neuronal state;
h＝(h1,h2,…,hj,…,hm) For hiding layer state, hjIs the jth neuron state;
a＝(a1,a2,…,ai,…,an) For visual layer biasing, aiBiasing for the ith neuron;
b＝(b1,b2,…,bj,…,bm) To hide layer bias, bjBiasing for the jth neuron;
the RBM model is actually an energy-based model, different model variable combinations correspond to different scalar energy, the scalar energy can be continuously changed in the RBM model training process, and the energy function of the RBM model is as follows:
from the above formula, the function value is related to the values of all neurons in the visible layer and the hidden layer, and the joint probability density of v and h is defined as:
the edge probability density of the visual layer is obtained by summing all hidden layer neurons:
the edge probability density of the hidden layer is obtained by summing all visible layer neurons:
the probability of neuron i being activated in the visual layer is:
the probability of neuron j being activated in the hidden layer is:
wherein, σ (a)i+∑jwijhj) And σ (b)j+∑iviwij) Is an activation function;
the training target of the RBM model is to solve a parameter theta ═ ai,bj,wijAnd (c) the RBM model can well fit the input data under the parameter, so that the probability distribution of the visual layer v is maximized, namely, the following likelihood function is maximized according to the training data set:
in order to obtain the optimal parameters, the maximum value of the likelihood function is solved through a random gradient rise method, and the partial derivative is solved for the parameter theta to obtain:
wherein, the first and the second end of the pipe are connected with each other,<vihj>p(h|v)indicating the expectation of the data distribution p (h | v),<vihj>modelrepresents the expectation of the model distribution p (v, h);
according to the contrast divergence algorithm, the update criteria that can be derived for the weights and biases are as follows:
Δωij＝α(vihj>p(h|v)-<vihj>recon) (9)
Δai＝α(<vi>p(h|v)-<vi>recon) (10)
Δbj＝α(<hj>p(h|v)-<hj>recon) (11)
wherein, alpha represents the learning rate,<vihj>recon、<vi>reconand<hj>reconrespectively representing the expectations for p (v, h), p (v), and p (h) under the reconstructed model distribution;
obtaining an updated parameter theta according to the updated weight and the offset;
s2.5: parameter adjustment: in the whole DBN structure, each layer structure is trained through learning rules, the value of the visual layer is transmitted to the hidden layer, the visual layer is reconstructed through the hidden layer, and the parameters of the network are updated according to the difference between the reconstructed visual layer and the original visual layer. By stacking the network structures layer by layer, not only the calculation process of the network can be simplified, but also higher-level feature expression can be obtained from input data; after the unsupervised learning process is completed, carrying out supervised joint training on the DBN structure according to a label attached to the top layer of the network, and finely adjusting parameters of all network layer structures according to a prediction result to form a dry eye data prediction model; supervised global training can reduce training errors and improve classification accuracy, typically by fine-tuning all network parameters through a gradient descent method. Because each RBM model can only ensure that parameters in the layer are optimal for the training of the current layer, but not for the training of the whole network during the training, the parameters of the whole network are finely adjusted by utilizing label information through back propagation errors;
s3: and (3) data output: the dry eye data prediction model outputs relevant information of the strong pulse laser, including energy waveform of the strong pulse laser, duration of each sub-pulse in the strong pulse laser and energy intensity of each sub-pulse in the strong pulse laser;
s4: the treatment application is as follows: applying the relevant information of the output strong pulse laser to the strong pulse laser xerophthalmia therapeutic apparatus to enable the strong pulse laser emitted by the strong pulse laser xerophthalmia therapeutic apparatus to conform to the relevant information;
s5: abnormal feedback: if the dry eye treatment effect by the strong pulse laser does not reach the expected effect, feeding back the data of the patient collected in the step S1 to the dry eye data prediction model; the expected effect can be set to be that after three treatments, no obvious improvement is obtained, and the medical care personnel judge according to the treatment experience;
s6: and (3) adjusting again: and fine-tuning parameters of the dry eye data prediction model through supervised training, specifically adopting a gradient descent method for fine tuning.
According to the scheme, self data (skin color, thickness and pain tolerance of a patient) of the patient are used as input factors, a dry eye data prediction model is used for prediction processing, strong pulse laser which accords with the dry eye treatment of the patient is output, and the final strong pulse laser is generated from three aspects of waveform of the strong pulse laser, duration of each sub-pulse and energy intensity of each sub-pulse, so that judgment according to self experience and skill of medical staff is replaced, and the problems of low automation degree and inaccurate judgment are solved.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
The foregoing is merely an example of the present invention, and common general knowledge in the field of known specific structures and characteristics is not described herein in any greater extent than that known in the art at the filing date or prior to the priority date of the application, so that those skilled in the art can now appreciate that all of the above-described techniques in this field and have the ability to apply routine experimentation before this date can be combined with one or more of the present teachings to complete and implement the present invention, and that certain typical known structures or known methods do not pose any impediments to the implementation of the present invention by those skilled in the art. It should be noted that, for those skilled in the art, without departing from the structure of the present invention, several changes and modifications can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicability of the patent. The scope of the claims of the present application shall be determined by the contents of the claims, and the description of the embodiments and the like in the specification shall be used to explain the contents of the claims.
Claims (8)
1. The energy control method of the strong pulse laser xerophthalmia therapeutic instrument is characterized by comprising the following steps of: the method comprises the following steps:
s1: data acquisition: collecting patient data including patient skin tone, stratum corneum thickness, and pain tolerance;
s2: data input: inputting patient data into a deep learning based dry eye data prediction model;
s3: and (3) data output: the dry eye data prediction model outputs relevant information of the strong pulse laser, wherein the relevant information comprises an energy waveform of the strong pulse laser, duration of each sub-pulse in the strong pulse laser and energy intensity of each sub-pulse in the strong pulse laser;
s4: the treatment application is as follows: and the output relevant information of the strong pulse laser is applied to the strong pulse laser xerophthalmia therapeutic instrument, so that the strong pulse laser emitted by the strong pulse laser xerophthalmia therapeutic instrument conforms to the relevant information.
2. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 1, wherein: in step S2, the establishment of the dry eye data prediction model specifically includes the following steps:
s2.1: sample calibration: the sample adopts the processed relevant data of each patient, and the relevant data comprises the skin color of the patient, the thickness of the stratum corneum, the pain tolerance, the energy waveform of the strong pulse laser in treatment, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser; calibrating the sample with the treatment effect reaching the expected effect, and not calibrating the sample with the treatment effect not reaching the expected effect;
s2.2: confirmation of input and output quantities: the skin color, the thickness of the cuticle layer and the pain tolerance of a patient are used as input quantities, and the energy waveform of the strong pulse laser, the duration of each sub-pulse in the strong pulse laser and the energy intensity of each sub-pulse in the strong pulse laser during treatment are used as output quantities;
s2.3: training data set generation: embedding the skin color, the thickness of the cuticle layer and the pain tolerance of each patient in the calibration sample into RGB components of the picture, and converting the RGB components into an information image; preprocessing an information image, generating a target information segment with the same picture size as a training data set;
s2.4: establishing a dry eye data prediction model based on a DBN structure: the method comprises the steps that a deep confidence network is adopted as an algorithm of a dry eye data prediction model, the deep confidence network comprises a plurality of RBM models, each RBM model comprises a visible layer and a hidden layer which are connected in a two-way mode, a training data set is used as an input vector of a first RBM model, and the first RBM model is trained; taking the hidden layer of the trained first RBM as an input vector of the visible layer of the second RBM, and continuing to train the second RBM; taking the hidden layer of the trained second RBM as an input vector of the visible layer of the third RBM, and continuing to train the third RBM; sequentially training until all RBM models are trained, connecting the hidden layer of the last RBM model to the output layer, and outputting a prediction result;
s2.5: parameter adjustment: after the unsupervised learning process is completed, performing supervised joint training on the DBN structure according to the label attached to the top layer of the network, and finely adjusting parameters of all network layer structures according to a prediction result to form a dry eye data prediction model.
3. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 2, wherein: further comprising the steps of:
s5: abnormal feedback: if the dry eye treatment effect by the strong pulse laser does not reach the expected effect, feeding back the data of the patient collected in the step S1 to the dry eye data prediction model;
s6: and (3) adjusting again: and fine-tuning parameters of the dry eye data prediction model through supervised training again.
4. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 3, wherein: in step S2.4, each RBM model includes n visible layers and m hidden layers, and the parameter θ is expressed as θ ═ ai,bj,wijIn which wijRepresents a connection weight; among them are:
v＝(v1,v2,…,vi,…,vn) Is a visual layer state, viIs the ith neuronal state;
h＝(h1,h2,…,hj,…,hm) For hiding layer state, hjIs the jth neuron state;
a＝(a1,a2,…,ai,…,an) For visual layer biasing, aiBiasing for the ith neuron;
b＝(b1,b2,…,bj,…,bm) To hide layer bias, bjBiasing for the jth neuron;
the energy function of the RBM model is:
from the above formula, the function value is related to the values of all neurons in the visible layer and the hidden layer, and the joint probability density of v and h is defined as:
the edge probability density of the visual layer is obtained by summing all hidden layer neurons:
the edge probability density of the hidden layer is obtained by summing all visible layer neurons:
the probability of neuron i being activated in the visual layer is:
the probability of neuron j being activated in the hidden layer is:
wherein, σ (a)i+∑jwijhj) And σ (b)j+∑iviwij) Is an activation function;
to maximize the probability distribution of the visual layer v, i.e. from the training data set, the likelihood function of:
solving the maximum value of the likelihood function by a random gradient rising method, and solving the partial derivative of the parameter theta to obtain:
wherein the content of the first and second substances,<vihj>p(h|v)indicating the expectation of the data distribution p (h | v),<vihj>modelrepresents the expectation of the model distribution p (v, h);
according to the contrast divergence algorithm, the update criteria that can be derived for the weights and biases are as follows:
wherein, alpha represents the learning rate,<vihj>recon、<vi>reconand<hj>reconrespectively representing the expectations for p (v, h), p (v), and p (h) under the reconstructed model distribution;
and obtaining an updated parameter theta according to the updated weight and the offset.
5. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 1, wherein: the thickness of the stratum corneum includes thick, medium and thin.
6. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 1, wherein: pain tolerance includes grade 1, 2, 3, 4 and 5, with grade 1 being the weakest and grade 5 being the strongest.
7. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 1, wherein: the skin color of the patient includes black skin, yellow skin and white skin.
8. The energy control method of the intense pulse laser dry eye therapeutic apparatus according to claim 3, wherein: in step S6 and step S2.5, the parameters of all network layer structures are fine-tuned using the gradient descent method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210247772.XA CN114618090A (en) | 2022-03-14 | 2022-03-14 | Energy control method of strong pulse laser xerophthalmia therapeutic instrument |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210247772.XA CN114618090A (en) | 2022-03-14 | 2022-03-14 | Energy control method of strong pulse laser xerophthalmia therapeutic instrument |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114618090A true CN114618090A (en) | 2022-06-14 |
Family
ID=81902655
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210247772.XA Pending CN114618090A (en) | 2022-03-14 | 2022-03-14 | Energy control method of strong pulse laser xerophthalmia therapeutic instrument |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114618090A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116959696A (en) * | 2023-09-20 | 2023-10-27 | 武汉光盾科技有限公司 | Data processing method and device based on laser therapeutic instrument |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1642500A (en) * | 2002-03-28 | 2005-07-20 | 博士伦公司 | System and method for predictive ophthalmic correction |
| CN110021428A (en) * | 2019-04-03 | 2019-07-16 | 北京大学第三医院（北京大学第三临床医学院） | A method of laser freckle effect is improved using neural network |
| CN110338906A (en) * | 2019-07-10 | 2019-10-18 | 清华大学深圳研究生院 | Smart therapeutics system and method for building up for photo-crosslinking operation |
| US20210290154A1 (en) * | 2020-03-17 | 2021-09-23 | Lumenis Ltd. | Method and system for determining an optimal set of operating parameters for an aesthetic skin treatment unit |
| CN113679355A (en) * | 2021-09-17 | 2021-11-23 | 山西瑞豪生物科技有限公司 | Skin complexion measuring device and method for intense pulse light xerophthalmia treatment |
| KR102365783B1 (en) * | 2021-11-30 | 2022-02-21 | 주식회사 미모바이오 | Method and device for generating customized prescription data based on artificial intelligence |
-
2022
- 2022-03-14 CN CN202210247772.XA patent/CN114618090A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1642500A (en) * | 2002-03-28 | 2005-07-20 | 博士伦公司 | System and method for predictive ophthalmic correction |
| CN110021428A (en) * | 2019-04-03 | 2019-07-16 | 北京大学第三医院（北京大学第三临床医学院） | A method of laser freckle effect is improved using neural network |
| CN110338906A (en) * | 2019-07-10 | 2019-10-18 | 清华大学深圳研究生院 | Smart therapeutics system and method for building up for photo-crosslinking operation |
| US20210290154A1 (en) * | 2020-03-17 | 2021-09-23 | Lumenis Ltd. | Method and system for determining an optimal set of operating parameters for an aesthetic skin treatment unit |
| CN113679355A (en) * | 2021-09-17 | 2021-11-23 | 山西瑞豪生物科技有限公司 | Skin complexion measuring device and method for intense pulse light xerophthalmia treatment |
| KR102365783B1 (en) * | 2021-11-30 | 2022-02-21 | 주식회사 미모바이오 | Method and device for generating customized prescription data based on artificial intelligence |
Non-Patent Citations (1)
| Title |
|---|
| 梁礼明: "《优化方法导论》", 北京：北京理工大学出版社 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116959696A (en) * | 2023-09-20 | 2023-10-27 | 武汉光盾科技有限公司 | Data processing method and device based on laser therapeutic instrument |
| CN116959696B (en) * | 2023-09-20 | 2023-12-08 | 武汉光盾科技有限公司 | Data processing method and device based on laser therapeutic instrument |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wei et al. | Automatic image segmentation based on PCNN with adaptive threshold time constant | |
| CN114613492B (en) | Multi-mode laser dry eye therapeutic instrument wavelength control method | |
| DE102011102333A1 (en) | Systems and methods for constructing a patient-specific model of neutral electrical stimulation | |
| CN103324981B (en) | Based on the quantization method that the chemicotherapy normalized quality of neural network controls | |
| CN105792745A (en) | Detecting neuronal action potentials using a sparse signal representation | |
| CN114618090A (en) | Energy control method of strong pulse laser xerophthalmia therapeutic instrument | |
| CN105828871B (en) | For the system and method from electrical stimulation tissue detection neuron action potential signal | |
| CN109414581A (en) | The bionical fast adaptation of cochlear implant | |
| Granley et al. | Hybrid neural autoencoders for stimulus encoding in visual and other sensory neuroprostheses | |
| Pimentel et al. | Human mental state monitoring in the wild: Are we better off with deeperneural networks or improved input features? | |
| CN116564477B (en) | Cognitive dysfunction rehabilitation training system | |
| Ming-Ai et al. | Feature extraction and classification of mental EEG for motor imagery | |
| CN116452855A (en) | Wound image classification and laser assisted treatment method based on deep learning | |
| JP2002366927A (en) | Neural network, neural network system and neural network processing program | |
| Corbin et al. | Assessing bias in skin lesion classifiers with contemporary deep learning and post-hoc explainability techniques | |
| Granley et al. | A hybrid neural autoencoder for sensory neuroprostheses and its applications in bionic vision | |
| Hallum et al. | Simulating auditory and visual sensorineural prostheses: a comparative review | |
| CN110931123B (en) | Matrix type brain network and construction method thereof | |
| Padmavilochanan et al. | Minimal features based non invasive cholesterol computation using machine learning | |
| CN115670639B (en) | High-voltage steep pulse treatment control system | |
| Bai | Neural representation of speech segmentation and syntactic structure discrimination | |
| Kharchenko et al. | Fundamental frequency evaluation subsystem for natural speech rehabilitation software calculation module for cancer patients after larynx resection | |
| Deepak et al. | A convolutional neural-network model of human cochlear mechanics and filter tuning for real-time applications | |
| Leite de Castro et al. | Neural Activity Shaping in Visual Prostheses with Deep Learning | |
| CN118152860A (en) | Emotion recognition-based regulation and control content generation method, system, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220614 |