Ventilator rapid detection method and device, electronic device and storage medium

Abstract

The present disclosure relates to a method and device for rapid detection of a ventilator, an electronic device, and a storage medium. The embodiments of the present disclosure can be implemented and relate to the technical field of ventilators. Among them, the method for rapid detection of a ventilator includes: obtaining a first label corresponding to a patient suffering from respiratory distress syndrome, a second label corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label; obtaining the second setting parameter corresponding to when the subject wears the ventilator, and completing the rapid detection of the ventilator. The embodiments of the present disclosure can realize rapid detection of the setting parameters of the ventilator.

Description

Technical Field

[0001] The present disclosure relates to the technical field of ventilators, and in particular to a method and device for rapid detection of ventilators, an electronic device, and a storage medium. Background Art

[0002] A ventilator is a medical device that replaces, controls, and modifies normal physiological breathing, increasing lung ventilation, improving respiratory function, reducing respiratory effort, and conserving cardiac reserve capacity. As emergency medical equipment, ventilators are currently widely used in clinical departments of large and medium-sized hospitals and carry the highest clinical risk of all medical devices. Ventilators are high-risk medical devices, and their safety and effectiveness in clinical use are directly related to the patient's life safety. Effective quality control of ventilators during use, regular testing, and prompt resolution of any quality issues and potential hazards discovered, to improve the safety and reliability of ventilator clinical use, are fundamental requirements for clinical emergency care. Especially in the event of a major public health emergency, such as a major epidemic, rapidly confirming the core performance parameters of a ventilator at the scene and ensuring that key safety indicators meet the clinical requirements of on-site emergency care have become urgent challenges for hospital medical staff and clinical medical engineering technicians.

[0003] The key issues of rapid on-site ventilator testing technology are effectiveness, ease of use, and speed. Effectiveness means that the test items for rapid ventilator testing should cover the core parameters, key functions, and common faults of the ventilator. This needs to be confirmed and optimized through big data analysis technology in combination with clinical applications. Ease of use should take into account that rapid on-site ventilator testing is often performed by medical staff rather than professional testing technicians. The testing process needs to be simplified and appropriate operational guidance needs to be provided. Rapid testing itself also has high requirements for timeliness and speed. At present, there is no mature solution for rapid on-site ventilator testing.

[0004] Especially in emergency situations during special periods such as epidemics, it is even more necessary to regularly (weekly / monthly) confirm the safety of ventilators before use. This requires medical institutions to have the ability to quickly test ventilators on site. Summary of the Invention

[0005] The present disclosure proposes a method and device for rapid detection of a ventilator, an electronic device, and a storage medium technical solution.

[0006] According to one aspect of the present disclosure, a method for rapid detection of a ventilator is provided, comprising:

[0007] Obtaining a first tag corresponding to a patient suffering from respiratory distress syndrome, a second tag corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode;

[0008] selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label;

[0009] The second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator is obtained to complete the ventilator rapid detection.

[0010] Preferably, before obtaining the first label corresponding to the patient suffering from respiratory distress syndrome and the second label corresponding to the patient not suffering from respiratory distress syndrome, determining whether the patient suffers from respiratory distress syndrome comprises:

[0011] Acquiring a lung imaging image and extracting a first set imaging feature of the lung imaging image;

[0012] Using a preset classifier and the first set imaging features, it is determined whether the patient suffers from respiratory distress syndrome.

[0013] Preferably, before acquiring the lung imaging image, an image conversion model is acquired; the lung imaging image is enhanced using the image conversion model to obtain an enhanced image; and then, a first set imaging feature of the enhanced image is extracted;

[0014] and / or,

[0015] Before acquiring the lung imaging image, a lung region segmentation model and a chest lung imaging image to be segmented are acquired, and the chest lung imaging image is segmented using the lung region segmentation model to obtain a lung imaging image;

[0016] and / or,

[0017] The method for extracting the first set imaging feature of the lung imaging image further includes:

[0018] Get the set imaging selection model;

[0019] Selecting the second set imaging feature from the first set imaging feature using the imaging selection model, the first label, and the second label;

[0020] Using a preset classifier and the second set imaging feature, it is determined whether the patient suffers from respiratory distress syndrome.

[0021] Preferably, the method for determining whether a patient suffers from respiratory distress syndrome further comprises:

[0022] Obtain an imaging fusion model;

[0023] Using the imaging fusion model, fusing the first set imaging feature or the second set imaging feature to obtain a fusion feature;

[0024] splicing the first set imaging feature or the second set imaging feature and the fusion feature to obtain a spliced ​​feature vector;

[0025] Using a preset classifier and the concatenated feature vector, it is determined whether the patient suffers from respiratory distress syndrome.

[0026] Preferably, the method of selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes:

[0027] Get the setting parameter selection model;

[0028] The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

[0029] According to one aspect of the present disclosure, a rapid detection device for a ventilator is provided, comprising:

[0030] an acquiring unit, configured to acquire a first tag corresponding to a patient suffering from respiratory distress syndrome, a second tag corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode;

[0031] a selection unit, configured to select a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label;

[0032] The detection unit is used to obtain the second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator, and complete the rapid detection of the ventilator.

[0033] According to one aspect of the present disclosure, there is provided a rapid detection device for a ventilator, comprising: a first flow sensor, a second flow sensor, a first pressure sensor, and an oxygen concentration sensor;

[0034] The first flow sensor, the first pressure sensor and the oxygen concentration sensor are arranged at the air supply port of the ventilator, and are used to detect the air supply flow of the ventilator / set the first tidal volume, air supply pressure and air supply oxygen concentration of the lung model or the subject respectively;

[0035] The second flow sensor is arranged at the collection port of the ventilator and is used to detect the second tidal volume of the set lung model or the subject.

[0036] Preferably, the detection device further comprises: a processor; the processor is connected to the first flow sensor, the second flow sensor, the first pressure sensor and the oxygen concentration sensor respectively;

[0037] The processor is configured to determine a respiratory rate based on the first tidal volume and the second tidal volume and / or determine a positive end-expiratory pressure based on the supplied air pressure;

[0038] or,

[0039] Also included: a processor and a second pressure sensor;

[0040] The processor is connected to the first flow sensor, the second flow sensor, the first pressure sensor and the oxygen concentration sensor respectively;

[0041] The second pressure sensor is configured on the outside of the chest of the set lung model or the subject, and is used to detect the respiratory cycle of the set lung model or the subject;

[0042] The processor is configured to determine a respiratory rate based on a respiratory cycle and / or determine a positive end-expiratory pressure according to the supplied air pressure.

[0043] Preferably, an analog / digital conversion circuit is provided between the processor and the oxygen concentration sensor;

[0044] The analog / digital conversion circuit is used to convert the analog value of the oxygen concentration of the supplied air into a digital value.

[0045] Preferably, the first flow sensor, the second flow sensor, and the first pressure sensor are connected to the processor via an IIC bus.

[0046] Preferably, the IIC bus includes: a data line and a clock line;

[0047] The data line and the clock line are connected to one end of a first pull-up resistor and a second pull-up resistor respectively, and the other ends of the first pull-up resistor and the second pull-up resistor are connected to a set power supply.

[0048] Preferably, the detection device further comprises: a display mechanism, wherein the display mechanism is connected to the processor and the oxygen concentration sensor respectively;

[0049] The display mechanism is used to display the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory rate and / or the positive end-expiratory pressure.

[0050] Preferably, the display mechanism is connected to the processor via a serial port and / or Ethernet.

[0051] Preferably, a memory is provided between the display mechanism and the processor;

[0052] The memory is used to store the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory frequency and / or positive end-expiratory pressure.

[0053] Preferably, the detection device further comprises: an atmospheric pressure sensor;

[0054] The atmospheric pressure sensor is used to detect the atmospheric pressure value to calibrate the air supply flow rate of the ventilator / the first tidal volume of the set lung model or the subject and the second tidal volume of the set lung model or the subject.

[0055] According to one aspect of the present disclosure, a ventilator is provided, comprising: the detection device as described above, wherein the ventilator is connected to an input mechanism; wherein the input mechanism is used to set a ventilation mode of the ventilator.

[0056] According to one aspect of the present disclosure, there is provided an electronic device, including:

[0057] processor;

[0058] a memory for storing processor-executable instructions;

[0059] Wherein, the processor is configured to: execute the above-mentioned ventilator rapid detection method.

[0060] According to one aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the above-mentioned ventilator rapid detection method is implemented.

[0061] In the embodiments of the present disclosure, rapid detection of set respiratory parameters can be achieved to solve the current problems of low timeliness and speed of detection. Specifically, the present disclosure combines the clinical needs of epidemic prevention work, targets ventilators (invasive ventilators, emergency ventilators, etc.), and confirms the detection methods of their core performance parameters and key safety indicators according to their working principles and application scenarios, and develops corresponding on-site rapid detection devices to solve the problem of rapid confirmation of the safety and effectiveness of ventilators in clinical use, and effectively improve the emergency response capabilities of medical institutions to prevent and deal with major epidemics and other public health emergencies.

[0062] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

[0063] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings herein are incorporated into and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and, together with the specification, are used to explain the technical solutions of the present disclosure.

[0065] Figure 1 A flow chart showing a method for rapid detection of a ventilator according to an embodiment of the present disclosure is shown;

[0066] Figure 2 shows a network structure of a setting synthesizer according to an embodiment of the present disclosure;

[0067] Figure 3 A block diagram of a rapid detection device for a ventilator according to an embodiment of the present disclosure is shown;

[0068] Figure 4 is a block diagram of an electronic device 800 according to an exemplary embodiment;

[0069] Figure 5 is a block diagram of an electronic device 1900 according to an exemplary embodiment. DETAILED DESCRIPTION

[0070] Various exemplary embodiments, features, and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise indicated.

[0071] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0072] The term "and / or" herein simply describes an association relationship between associated objects, indicating that three relationships can exist. For example, "A and / or B" can represent the existence of three situations: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" herein refers to any combination of at least two of any one or more of a plurality of items. For example, "at least one of A, B, and C" can represent any one or more elements selected from the set consisting of A, B, and C.

[0073] In addition, numerous specific details are provided in the following detailed description to better illustrate the present disclosure. Those skilled in the art will appreciate that the present disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art are not described in detail in order to highlight the main points of the present disclosure.

[0074] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form combined embodiments without violating the principle logic. Due to space limitations, the present disclosure will not elaborate on them.

[0075] In addition, the present disclosure also provides a ventilator rapid detection device, an electronic device, a computer-readable storage medium, and a program, all of which can be used to implement any ventilator rapid detection method provided by the present disclosure. The corresponding technical solutions and descriptions are referred to the corresponding records in the method section and will not be repeated here.

[0076] Figure 1 A flowchart of a rapid detection method for a ventilator according to an embodiment of the present disclosure is shown. Figure 1 As shown, the ventilator rapid detection method includes: step S101: obtaining a first label corresponding to a patient with respiratory distress syndrome, a second label corresponding to a patient without respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; step S102: selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label; step S103: obtaining the second setting parameter corresponding to a subject with respiratory distress syndrome when wearing a ventilator, and completing the ventilator rapid detection. Rapid detection of set respiratory parameters can be achieved to solve the current problem of low timeliness and speed of detection. Specifically, the present disclosure combines the clinical needs of epidemic prevention work, targets ventilators (invasive ventilators, emergency ventilators, etc.), and, based on their working principles and application scenarios, confirms the detection methods of their core performance parameters and key safety indicators, and develops corresponding on-site rapid detection devices to solve the problem of rapid confirmation of the safety and effectiveness of the clinical use of ventilators, and effectively improves the emergency response capabilities of medical institutions to prevent and deal with major epidemics and other public health emergencies.

[0077] Step S101: obtaining a first tag corresponding to a patient suffering from respiratory distress syndrome, a second tag corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of a ventilator in a set ventilation mode.

[0078] In the embodiments of the present disclosure and other possible embodiments, the first label corresponding to respiratory distress syndrome can be configured as 1, while the second label corresponding to non-respiratory distress syndrome can be configured as 0. Those skilled in the art can configure the first label corresponding to respiratory distress syndrome and the second label corresponding to non-respiratory distress syndrome according to actual needs, but the first label and the second label should have different values. For example, the first label corresponding to respiratory distress syndrome can also be configured as 2, and the second label corresponding to non-respiratory distress syndrome can also be configured as 1.

[0079] In the embodiments of the present disclosure and other possible embodiments, the first setting parameters of the ventilator may include: inspiratory oxygen concentration FiO2, tidal volume Vt, respiratory rate f, peak airway pressure Ppeak, positive end-expiratory pressure PEEP, minute ventilation MV, inspiration-expiration ratio I:E (inspiratory time: expiratory time), inspiratory pressure level IPL, lung compliance C, airway resistance R, etc.

[0080] In addition, in the embodiments of the present disclosure and other possible embodiments, the set ventilation modes of the ventilator mainly include: volume-controlled ventilation VCV, pressure-controlled ventilation PCV, assisted ventilation AV, synchronized intermittent mandatory ventilation SIMV, spontaneous breathing SPONT, intermittent positive pressure ventilation IPPV, etc.

[0081] In addition, the present disclosure proposes a new method for determining respiratory distress syndrome based on lung imaging images to overcome the problem that the current identification of respiratory distress syndrome relies too much on subjective judgment, so that mild respiratory distress syndrome cannot be identified.

[0082] In an embodiment of the present disclosure, before obtaining a first label corresponding to having respiratory distress syndrome and a second label corresponding to not having respiratory distress syndrome, determining whether a patient suffers from respiratory distress syndrome comprises: obtaining a lung imaging image and extracting a first set imaging feature of the lung imaging image; and determining whether a patient suffers from respiratory distress syndrome using a preset classifier and the first set imaging feature.

[0083] In the embodiments of the present disclosure and other possible embodiments, before determining whether a patient has respiratory distress syndrome, a preset classifier is trained using a set number of first labels, second labels, and first set imaging features. Then, based on the trained preset classifier, a determination is made whether a patient has respiratory distress syndrome. The process of training the preset classifier is a conventional technique used by those skilled in the art and will not be further described here.

[0084] In the embodiments of the present disclosure and other possible embodiments, the preset classifier may be a machine learning (ML) classification model, for example, one or more of a support vector machine, a decision tree, a random forest, a K-nearest neighbor, a logistic regression, adaptive boosting, a linear discriminant analysis, and a multilayer perceptron.

[0085] In the embodiments of the present disclosure and other possible embodiments, a decision tree (DT) is a tree structure in which each internal node represents a judgment on an attribute, each branch represents the output of a judgment result, and finally each leaf node represents a classification result. The steps for generating a decision tree include: (a) node splitting. When the attribute represented by a node cannot be judged, the node is split into two child nodes (if it is not a binary tree, it will be divided into n child nodes); (b) threshold determination. The appropriate threshold is selected to minimize the training classification error rate.

[0086] Common DT models include ID3, C4.5, and Classification and Regression Tree (CART). CART generally performs better than other decision trees. They are introduced below.

[0087] ID3 uses the principle of increasing entropy to determine the parent node and split node. For a set of data, the smaller the entropy, the better the classification result. The definition of entropy is shown in formula (3).

[0088]

[0089] Among them, p(x i ) is x i The probability of a sample appearing. As can be seen, when entropy is at its maximum, 1, the classification is at its worst; when it is at its minimum, 0, it is a state of complete classification. While zero entropy is ideal, in practice, entropy generally lies between 0 and 1. Minimizing entropy effectively improves classification accuracy.

[0090] Because ID3's classification error rate decreases with finer segmentation, it tends to overfit as the segmentation becomes finer. To prevent over-segmentation, C4.5 improves on ID3. In C4.5, the optimization term is divided by the cost of over-segmentation. This ratio is called the information gain. Clearly, as the denominator of over-segmentation increases, the information gain decreases. Other than that, the principles are the same as those of ID3.

[0091] CART is a binary tree used for classification and regression. It can only split a parent node into two child nodes, using the Gini index to determine the split. CART also suffers from a bias toward small segments, a problem known as overlearning. To address this, the CART algorithm can prune particularly long trees.

[0092] In the embodiments of the present disclosure and other possible embodiments, a random forest (RF) model constructs multiple decision trees and comprehensively evaluates the predictions of the multiple decision trees to obtain a final result. The RF model is an extension of the bagging algorithm, combining the advantages of bagging and decision trees.

[0093] Specifically, the RF model uses bootstrap resampling techniques to repeatedly extract n samples from the dataset with replacement to form new training samples to train a decision tree. Then, the n decision trees are used to form m random forests. The final prediction value is determined based on the voting or mean structure of the m random forests. When the number of decision trees is large enough, the generalization ability of the random forest is between the following equation (4).

[0094]

[0095] Among them, R represents the generalization error of the random forest upper bound converges to the lower bound, It represents the average correlation coefficient between decision trees, and s is a measure of the strength of the decision tree.

[0096] Usually the strength of a model represents its average performance, and the performance can be expressed by the model's margin M, as shown in Equation (5).

[0097] M(x,y)=P(y θ =y)-maxP(y θ =z) (5)

[0098] Among them, y θ is the predicted classification result for attribute x by the decision tree constructed from the random vector θ, with y and z representing different categories. Generally speaking, a larger margin indicates a greater likelihood that the model will correctly predict the attribute value x of the unknown data instance. Therefore, as decision trees are ensembled, the relevance of the random forest trees increases, their generalization ability improves, and the classification error decreases.

[0099] In the embodiments of the present disclosure and other possible embodiments, the method for extracting the first set imaging feature of the lung imaging image includes: obtaining a set imaging feature extraction model; and performing feature extraction on the lung imaging image using the set imaging feature extraction model to obtain the first set imaging feature.

[0100] In the embodiment of the present disclosure and other possible embodiments, the set imaging feature extraction model may be a set imaging feature extraction model based on PyRadiomics and / or Med3D model.

[0101] Since radiomics features can well characterize the morphology and texture information of lesions or tissues, and with the rapid development of CNN models, both have been widely used in the analysis and processing of medical images. At present, intelligent analysis methods for medical images based on radiomics and CNN models mainly include: extracting radiomics features of medical images based on the PyRadiomics model, and combining it with machine learning for classification, prediction, and other tasks; and directly using CNN models to perform segmentation, target detection, classification, and other tasks on medical images. Among them, CNN feature extraction of medical images based on CNN models is a very meaningful work, which can deeply explore the information of medical images for the set tasks. Furthermore, effectively combining CNN features with machine learning can improve the level of disease diagnosis, treatment, and evaluation.

[0102] In the embodiments of the present disclosure and other possible embodiments, the PyRadiomics model is an open source model of the PyRadiomics model proposed by Van Griethuysen in 2017. The method for extracting lung imaging omics features based on the PyRadiomics model includes: first, determining the type of lung region image (lung imaging image); second, extracting lung imaging omics features of different types of lung region images (lung imaging images) based on the type of lung region image (lung imaging image) and the type of lung imaging omics feature setting. Wavelet filter, as an "image microscope", has been widely used in the field of image processing. Images of different scales are obtained through its multi-resolution decomposition, and have been applied accordingly, especially in complex chest lung images. At the same time, Laplacian of Gaussian filter (LoG) as an edge enhancement filter can emphasize the areas of grayscale changes in the image, which is crucial for COPD chest lung images. Based on this, the types of lung images further include: original lung images without filtering, and two types of derived lung images obtained by filtering the original lung images using wavelet filters and LoG filters, respectively. Based on the original lung images (lung imaging images), the two types of derived lung images (lung imaging images), and the set types of lung imaging features, lung imaging features are extracted for different types of lung images.

[0103] Among them, in the embodiments of the present disclosure and other possible embodiments, the setting types of lung imaging omics features include: First Order Statistics (First Order), Shape-based 2D&3D (SHAPE), Gray Level Cooccurence Matrix (GLCM), Gray Level Run Length Matrix (GLRLM), Gray Level Size Zone Matrix (GLSZM), Neighbouring Gray Tone Difference Matrix (NGTDM) and Gray Level Dependence Matrix (GLDM).

[0104] Research has shown that CNN models pre-trained from massive datasets (such as ImageNet) have become a powerful tool for accelerating training convergence and improving accuracy. To extract 3D CNN features from medical images, Sihong et al. proposed a pre-trained Med3D model in 2019 to perform targeted 3D medical tasks corresponding to medical images.

[0105] In the embodiments of the present disclosure and other possible embodiments, 3D CNN feature extraction based on the pre-trained Med3D model includes: pre-processing the lung region image (lung imaging image) strictly according to the method of the 3DSeg-8 dataset in the Med3D model, and using the encoder backbone network (3D ResNet10) of the migrated Med3D model to extract features from the pre-processed lung region image (lung imaging image) to generate 3D CNN features. First, the lung region image of size 512×512×N is cropped to a lung region image of size 280×400×N' (N'<N). The cropping process retains the lung region and deletes the corresponding image in the lung region image (lung imaging image) that does not have the lung; secondly, the lung region of the lung region image (lung imaging image) of size 280×400×N' is normalized, and random values ​​are generated for the part outside the lung, and the random values ​​conform to the Gaussian distribution.

[0106] In an embodiment of the present disclosure, before acquiring a lung imaging image, a lung region segmentation model and a chest lung imaging image to be segmented are acquired, and the chest lung imaging image is segmented using the lung region segmentation model to obtain a lung imaging image.

[0107] In the embodiments of the present disclosure and other possible embodiments, the lung region segmentation model may be a standard pathological lung region segmentation model (U-net (R231)) with high robustness proposed by Hofmanninger et al. in 2020. This standard pathological lung region segmentation model is obtained by training a deep neural convolutional neural network ResU-Net using images of various lung diseases. At the same time, the lung region segmentation model may also be a U-net segmentation model or other segmentation model improved based on U-net, or other existing segmentation models.

[0108] In an embodiment of the present disclosure, before acquiring the lung imaging image, an image conversion model is acquired; the lung imaging image is enhanced using the image conversion model to obtain an enhanced image; and then, a first set imaging feature of the enhanced image is extracted.

[0109] In the embodiments of the present disclosure and other possible embodiments, Figure 2 FIG. 5 shows a network structure of a configuration synthesizer according to an embodiment of the present disclosure (taking synthesizer 1 as an example). Figure 2 As shown, the method of enhancing the lung imaging image using the image conversion model to obtain an enhanced image includes: obtaining a medical image to be converted (lung imaging image) and setting a synthesizer; wherein, the training method of the setting synthesizer includes: using the generator G1 of the setting synthesizer to perform convolution processing on the plain image / enhanced image in the dual-energy medical image to generate a corresponding synthetic enhanced image / synthesized plain image; based on the synthetic enhanced image / synthesized plain image and the enhanced image / plain image in the dual-energy medical image, using a preset discriminator D1, completing the parameter training of the generator G1 in the setting synthesizer; determining the type corresponding to the medical image to be converted (lung imaging image); wherein, the type is a plain medical image or an enhanced medical image; based on the type and the generator G1 in the setting synthesizer, performing convolution processing on the medical image to be converted (lung imaging image) to complete the conversion of the medical image to be converted (lung imaging image) from a plain medical image to an enhanced medical image or from an enhanced medical image to a plain medical image. It can realize the conversion between plain scan and enhanced medical images, and obtain different plain scan or enhanced medical images with one scan, thereby meeting the needs of medical image analysis.

[0110] In the embodiments of the present disclosure and other possible embodiments, a medical image to be converted and a set synthesizer are obtained; wherein, the training method of the set synthesizer includes: using the generator G1 of the set synthesizer to perform convolution processing on the plain image / enhanced image in the dual-energy medical image to generate a corresponding synthetic enhanced image / synthesized plain image; based on the synthetic enhanced image / synthesized plain image and the enhanced image / plain image in the dual-energy medical image, using a preset discriminator D1, completing the parameter training of the generator G1 in the set synthesizer.

[0111] For example, the generator G1 of the pre-set synthesizer performs convolution processing on the plain scan image in the dual-energy medical image to generate a corresponding synthesized enhanced image. Based on the plain scan image, the synthesized enhanced image, and the enhanced image in the dual-energy medical image, the parameters of the generator G1 in the pre-set synthesizer are trained using a preset discriminator D1. At this point, the generator G1 (synthesizer 1) can complete the conversion of the medical image to be converted from a plain scan medical image to an enhanced medical image.

[0112] For another example, the generator G1 of the pre-set synthesizer performs convolution processing on the enhanced image in the dual-energy medical image to generate a corresponding synthesized plain scan image. Based on the plain scan image, the plain scan image and the enhanced image in the dual-energy medical image are synthesized, and the parameters of the generator G1 in the pre-set synthesizer are trained using the preset discriminator D1. At this point, the generator G1 (synthesizer 2) can complete the conversion of the medical image to be converted from the enhanced medical image to the plain scan image.

[0113] In the embodiments of the present disclosure and other possible embodiments, the medical image or dual-energy medical image to be converted may be a CT image, a DR image, an MRI image, an ultrasound image, a PET image, a CT-PET image, or other medical image. Furthermore, the CT image, DR image, MRI image, ultrasound image, PET image, CT-PET image, or other medical image may be an image of any part of the human body, such as a CT image, DR image, MRI image, ultrasound image, PET image, CT-PET image, or other medical image corresponding to the chest (lungs), brain, heart, kidneys, liver, stomach, bones, etc.

[0114] In the embodiments of the present disclosure and other possible embodiments, a chest (lung) CT image is used for illustration. The generator G1 of the setting synthesizer is used to perform convolution processing on the plain scan image (plain scan CT image) in the dual-energy medical image to generate a corresponding synthetic enhanced image / synthesized plain scan image (synthesized enhanced / plain scan CT image). Based on the plain scan image (plain scan CT image), the synthetic enhanced image / synthesized plain scan image (synthesized enhanced / plain scan CT image), and the enhanced image / plain scan image (enhanced CT image / plain scan CT image) in the dual-energy medical image, the preset discriminator D1 is used to complete the parameter training of the generator G1 in the setting synthesizer.

[0115] In an embodiment of the present disclosure, the method for generating a corresponding synthetic enhanced image / synthetic plain scan image based on the plain scan image in the dual-energy medical image using the generator G1 of the set synthesizer includes: using the encoder and decoder of the generator G1 to perform convolution processing on the plain scan image / enhanced image in the dual-energy medical image to obtain the corresponding synthetic enhanced image / synthetic plain scan image; and / or, the preset discriminator D1 includes: a plurality of convolution units connected in sequence.

[0116] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 As shown, a synthesizer is set, including: a generator G1 and a discriminator D1; wherein the generator G1 is used to synthesize enhanced images / synthesized plain scan images; the discriminator D1 is used to distinguish between real enhanced images and synthesized enhanced images / synthesized plain scan images, so that the generator G1 synthesizes real plain scan CT images (plain scan medical images) or enhanced CT images (enhanced medical images), that is, completing the conversion of the medical image to be converted from plain scan medical images to enhanced medical images or from enhanced medical images to plain scan medical images.

[0117] In an embodiment of the present disclosure, the encoder includes: a plurality of downsampling convolution units and a plurality of pooling units connected in sequence; and / or the decoder includes: a plurality of upsampling convolution units and an activation function connected in sequence.

[0118] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 As shown in (a), the encoder and decoder of the generator G1 are used to perform convolution processing on the plain scan image (plain scan CT image) in the dual-energy medical image to obtain the corresponding synthetic enhanced image / synthetic plain scan image (synthetic enhanced CT image / synthetic plain scan image); based on the synthetic enhanced image / synthetic plain scan image and the enhanced image / plain scan image (enhanced CT image / plain scan CT image) in the dual-energy medical image, the preset discriminator D1 is used to complete the parameter training of the generator G1 in the set synthesizer.

[0119] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 As shown in (b), the overall architecture of the generator G1 can adopt a U-net structure.

[0120] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 As shown in (b), the multiple upsampling convolution units of the encoder include: a first downsampling convolution unit (Conv-BN-ReLU×2), a first pooling unit (Max Pooling), a second downsampling convolution unit (Conv-BN-ReLU×2), a second pooling unit (MaxPooling), a third downsampling convolution unit (Conv-BN-ReLU×2), a third pooling unit (Max Pooling) and a fourth downsampling convolution unit (Conv-BN-ReLU×2) connected in sequence; wherein Conv represents a convolution layer, BN represents a batch normalization layer, ReLU represents an activation function layer, and Max Pooling represents a maximum pooling layer; ×2 represents the number is 2, for example, Conv-BN-ReLU×2 represents 2 first downsampling convolution units. At the same time, those skilled in the art can set the convolution kernel size and step size of the convolution layer Conv as needed, and can also omit the batch normalization layer BN as needed, and configure the activation function of the activation function layer, for example, ReLU can be configured as ELU or other activation functions, and the maximum pooling layer Max Pooling can be configured as an average pooling layer, etc.

[0121] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 As shown in (b), the multiple upsampling convolution units of the decoder include: a first deconvolution (upsampling) unit (Transposed Cov), a first convolution unit (Conv-BN-ReLU×2), a second deconvolution (upsampling) unit (Transposed Cov), a second convolution unit (Conv-BN-ReLU×2), a third deconvolution (upsampling) unit (Transposed Cov), a third convolution unit (Conv-BN-ReLU×2) and an activation function (Conv-Sigmoid) connected in sequence.

[0122] In the embodiments of the present disclosure and other possible embodiments, as Figure 2As shown in (c), the multiple convolutional units connected in sequence in the preset discriminator D1 adopt residual connections. Specifically, the multiple convolutional units connected in sequence include: a first convolutional layer (5×5×5Conv-ELU), a second convolutional layer (2×2×2Conv S2), a first activation function layer (ELU), a third convolutional layer (5×5×5Conv-ELU×2), a second activation function layer (ELU), a fourth convolutional layer (2×2×2Conv S2), a third activation function layer (ELU), a fifth convolutional layer (5×5×5Conv-ELU×4), a fourth activation function layer (ELU), a sixth convolutional layer (2×2×2Conv S2), a fifth activation function layer (ELU), a seventh convolutional layer (5×5×5Conv-ELU×4) and a sixth activation function layer (ELU). Among them, S2 and ×2 both indicate that they have two identical layer structures, and similarly, ×4 indicates that they have four identical layer structures.

[0123] In the implementation of the present disclosure, the method for completing the parameter training of the generator in the setting synthesizer based on the synthesized enhanced image / synthesized plain scan image and the enhanced image / plain scan image / plain scan image in the dual-energy medical image using a preset discriminator includes: training the preset discriminator based on the synthesized enhanced image / synthesized plain scan image and the enhanced image / plain scan image in the dual-energy medical image to achieve recognition of the synthesized enhanced image / synthesized plain scan image and the enhanced image / plain scan image; and using the trained preset discriminator to output the synthesized enhanced image / synthesized plain scan image generated by the generator and the plain scan image / enhanced image to achieve the maximum probability value, thereby completing the parameter training of the generator in the setting synthesizer. The method includes training the preset discriminator based on the synthesized enhanced image / synthesized plain scan image and the enhanced image / plain scan image / plain scan image in the dual-energy medical image to achieve recognition of the synthesized enhanced image / synthesized plain scan image and the enhanced image / plain scan image.

[0124] Specifically, the method for completing the parameter training of the generator in the synthesizer using a preset discriminator based on the plain scan image, the synthetic enhanced image, and the enhanced image in the dual-energy medical image includes: training the preset discriminator based on the plain scan image, the synthetic enhanced image, and the enhanced image in the dual-energy medical image to achieve recognition of the synthetic enhanced image and the enhanced image; and using the trained preset discriminator to output the synthetic enhanced image generated by the generator and the plain scan image to achieve the maximum probability value, thereby completing the parameter training of the generator in the synthesizer. The method includes training the preset discriminator based on the plain scan image, the synthetic enhanced image, and the enhanced image in the dual-energy medical image to achieve recognition of the synthetic enhanced image / synthetic plain scan image and the enhanced image.

[0125] Specifically, the method for completing parameter training of the generator in the synthesizer using a preset discriminator based on the synthesized plain scan image and the enhanced image and plain scan image in the dual-energy medical image includes: training the preset discriminator based on the plain scan image, the synthesized plain scan image, and the enhanced image in the dual-energy medical image to achieve recognition of the synthesized plain scan image and the plain scan image; and using the trained preset discriminator to output the synthesized plain scan image generated by the generator and the plain scan image to achieve the maximum probability value, thereby completing parameter training of the generator in the synthesizer. The method for training the preset discriminator based on synthesizing the plain scan image and the plain scan image in the dual-energy medical image based on the plain scan image to achieve recognition of the synthesized plain scan image and the plain scan image is described.

[0126] For example, Figure 2 "True" and "False" in (a). Specifically, if a synthetic enhanced image / synthesized plain scan image and the plain scan image / enhanced image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "False"; whereas, if a real enhanced image / plain scan image (non-synthesized enhanced image / non-synthesized plain scan image) and the plain scan image / enhanced image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "True".

[0127] More specifically, if the synthesized enhanced image and the plain scan image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "false"; however, if the real enhanced image (non-synthesized enhanced image) and the plain scan image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "true". Similarly, if the synthesized plain scan image and the enhanced image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "false"; however, if the real enhanced image (non-synthesized enhanced image) and the plain scan image are input to the preset discriminator D1, the trained preset discriminator D1 outputs "true".

[0128] Take synthesizer 1 as an example, Figure 2 As shown in (a), the plain scan CT is input into the generator to obtain a synthesized enhanced CT. The plain scan CT and the synthesized enhanced CT are then combined in the channel dimension and input into the discriminator. The discriminator outputs a probability map, which represents the probability that the input image is a real enhanced CT. In addition, the plain scan CT and the real enhanced CT are combined and input into the discriminator to obtain a probability map. When the input is not a pair of real images, a small probability value is output, and when the input is a pair of real images, a large probability value is output. The training goal of the generator is to make the probability value output by the discriminator as large as possible when the synthesized enhanced CT and the plain scan CT are input into the discriminator. The generator and the discriminator compete until a balance is reached.

[0129] The training process of the synthesizer is as follows. The optimization objective of the synthesizer consists of two parts: the conditional generative adversarial network objective and the L1 distance between the synthesized enhanced CT image and the real enhanced CT image. The conditional generative adversarial network objective can be expressed as:

[0130]

[0131] Among them, the generator G(x,z) tries to minimize the target, and the discriminator D(x,y), D(x,G(x,z)) tries to maximize the target, x represents plain CT, y represents real enhanced CT, and z represents noise; among them, is the expectation function, which is used to calculate the expected value.

[0132] Used to find the expected value of the discriminator logD(x,y); Used to find the expected value of log(1-D(x,G(x,z)).

[0133] The L1 distance is used to constrain the difference between the synthetic enhanced CT image and the real enhanced CT image:

[0134]

[0135] The ultimate optimization goals of the synthesizer are:

[0136]

[0137] Where λ is the weight of the L1 loss between the synthetic image (synthetic enhanced image / synthetic plain image) and the real image (real enhanced image).

[0138] We use the 3D U-Ne architecture as the generator ( Figure 2 (b)), which includes the encoder and decoder parts. The encoder part consists of three blocks. Each block consists of two layers, each of which includes a convolution operation Cov, batch normalization BN, and a rectified linear unit ReLU. Downsampling is performed between each block through Max Pooling. The decoder consists of three blocks, each of which has the same structure as the encoder. Upsampling is performed between each block in the decoder through a transposed convolution layer Transposed Cov. In addition, we connect the layers with the same resolution in the encoder and decoder. Among them, S2 indicates that the stride S is configured to 2; ELU is the activation function, specifically the Gaussian error linear unit.

[0139] In an embodiment of the present disclosure, during the parameter training of the generator in the set synthesizer, the synthetic enhanced image / synthetic plain scan image output by the last layer activation function of the decoder in the generator and the synthetic enhanced image / synthetic plain scan image output by the set upsampling convolution unit before the last layer activation function are used to calculate the loss value of the generator.

[0140] In the embodiments of the present disclosure and other possible embodiments, we introduce deep supervision into the synthesizer. Since ordinary convolutional neural networks only use the last layer of the model to calculate the loss, the information contained in the feature map of the hidden layer is not effectively utilized. Deep supervision can inject gradients deeper into the network, facilitating the training of all layers in the network. Specifically, for output 1 and output 2 ( Figure 2 (b)), the corresponding downsampled image is used to calculate the L1 loss value.

[0141] In the embodiments of the present disclosure and other possible embodiments, as Figure 2 (c) shows the discriminator, which includes a convolutional layer, an exponential linear unit / Gaussian error linear unit (ELU), and four strided convolutional layers. Each time a strided convolutional layer is passed, the size of the feature map is reduced by half.

[0142] In an embodiment of the present disclosure, in the process of setting the parameter training of the generator in the synthesizer, it also includes: using real plain scan medical images and corresponding enhanced medical images to adjust the parameters of the generator trained based on the dual-energy medical images.

[0143] This disclosure trains the synthesizer in three steps. In the first step, the synthesizer generator is pre-trained using self-supervised learning. In the second step, the synthesizer is pre-trained using dual-energy CT from Dataset 3. In the third step, the synthesizer is fine-tuned using plain and contrast-enhanced CT from Dataset 1. Because the plain and contrast-enhanced CT images are not aligned, they must first be registered.

[0144] In the embodiments of the present disclosure and other possible embodiments, self-supervised learning is used to train the synthesizer (image transformation). Specifically, due to the difficulty in obtaining a large number of paired plain scan CT and enhanced CT, we introduced self-supervised learning to pre-train the synthesizer. Self-supervised learning is mainly divided into context-based methods, generation-based methods, and contrast-based methods. Generation-based self-supervised learning methods are divided into restoration-based and generative adversarial network-based methods. Self-supervised learning based on image restoration is suitable for image synthesis tasks. We pre-train the synthesizer based on the self-supervised learning framework of image restoration. The sub-blocks cropped from random positions in the CT image are passed through a transformation function, and the transformed sub-blocks are used as input to the generator, allowing the model to restore the original sub-blocks. Each sub-block can pass through up to three of the four transformations: nonlinear transformation, local pixel reconstruction, and external or internal occlusion. External occlusion and internal occlusion cannot be used at the same time. The loss function of self-supervised learning is expressed as:

[0145]

[0146] Where G is the generator in the synthesizer, θ G are the parameters of the generator, and f(·) is the transformation function.

[0147] In our experiments, 888 CT scans from the LUNA1 dataset are used to pre-train the generator of the synthesizer.

[0148] In the embodiments of the present disclosure and other possible embodiments, a dual-energy CT is used to pre-train the synthesizer (the second step). Specifically, our synthesizer relies on aligned images. However, plain scan CT and enhanced CT are scanned at different times, and slight movements of the patient (such as breathing) may cause misalignment between plain scan CT and enhanced CT. Misalignment between plain scan CT and enhanced CT is an obstacle to enhanced CT or plain scan CT synthesis. Although registration can alleviate this misalignment. However, directly training the synthesizer using the registered plain scan CT and enhanced CT will result in blurred synthesized CT. Therefore, we use dual-energy CT from data set three to pre-train the synthesizer. Compared with plain scan CT, the virtual plain scan CT and enhanced CT in dual-energy CT are fully aligned. Fully aligned virtual plain scan CT and enhanced CT play a key role in training the synthesizer.

[0149] In the embodiments of the present disclosure and other possible embodiments, the synthesizer is trained through image registration and fine-tuning. Specifically, after pre-training on dual-energy CT, the synthesizer can perform conversion between virtual plain CT and enhanced CT. Due to the differences between virtual plain CT and real plain CT, the synthesizer is fine-tuned using real plain CT and corresponding enhanced CT from Dataset 1.

[0150] Because our synthesizer relies on aligned images, we first need to register the actual plain CT and enhanced CT images. Because plain CT and enhanced CT are scanned at different times, slight movement of the patient may cause misalignment, which can be mitigated by registration. Experiments have shown that different registration methods have a significant impact on the synthesis performance of the synthesizer. We use enhanced CT as the fixed image and plain CT as the moving image. We use affine alignment, Sy, and Elasti to register plain CT and enhanced CT. Affine alignment is often used as the initial stage of image registration, and it helps optimize the registration process for more complex deformable images. Symmetric image normalization (SyN) is currently the best-performing registration algorithm. Elastix is ​​an intensity-based medical image registration toolbox.

[0151] For Synthesizer 1 and Synthesizer 2, 24 pairs of plain CT and enhanced CT were used as the training set, 8 pairs of plain CT and enhanced CT were used as the validation set, and 8 pairs of plain CT and enhanced CT were used as the test set. The number of iterations was set to 300. Adam was used to optimize the network weights. The initial learning rate was 1×10 -3 The cosine annealing strategy is used to adjust the learning rate.

[0152] In an embodiment of the present disclosure, the type of the medical image to be converted is determined; wherein the type is a plain scan medical image or an enhanced medical image.

[0153] In the embodiments of the present disclosure and other possible embodiments, a chest (lung) CT image is used for illustration, and a plain scan medical image or an enhanced medical image is a chest (lung) plain scan CT medical image or a chest (lung) enhanced CT medical image, respectively.

[0154] In an embodiment of the present disclosure, the generator G1 in the synthesizer is configured to perform convolution processing on the medical image to be converted based on the type and setting, thereby completing the conversion of the medical image to be converted from a plain medical image to an enhanced medical image or from an enhanced medical image to a plain medical image.

[0155] In the embodiments of the present disclosure and other possible embodiments, the method of extracting the first set imaging feature of the lung imaging image further includes: obtaining a set imaging selection model; using the imaging selection model, the first label and the second label to select the second set imaging feature from the first set imaging feature; and using a preset classifier and the second set imaging feature to determine whether respiratory distress syndrome is present.

[0156] In the embodiment of the present disclosure and other possible embodiments, the set imaging selection model may be a parameter selection model based on the Lasso algorithm or the GLM model.

[0157] In the embodiments of the present disclosure and other possible embodiments, the least absolute shrinkage and selection operator (Lasso) algorithm compresses variables with larger parameter estimates to smaller ones and compresses variables with smaller parameter estimates to zero. Currently, the Lasso algorithm has become an effective means of feature selection.

[0158] The mathematical expression corresponding to the Lasso algorithm is shown in formula (10).

[0159]

[0160] In formula (1), x ij Represents the independent variable, that is, the first set imaging feature after standardization, yi represents the dependent variable (the first label and the second label corresponding to whether the patient suffers from respiratory distress syndrome), λ represents the penalty parameter (λ≥0), βj represents the regression coefficient, i∈[1,n], j∈[0,p].

[0161] In the embodiments of the present disclosure and other possible embodiments, the generalized linear model (GLM) is an extension of the linear model, which aims to establish the relationship between the mathematical expectation of the response variable and the linear combination of the predictor variables through the link function. Unlike the Lasso algorithm, the GLM can calculate the expected value of each independent variable x ij The R2 value of (feature) is used for feature selection. The mathematical expression of GLM is shown in formula (11).

[0162]

[0163] Among them, the connection function The average With linear predictor Establish a connection; yi represents the dependent variable (whether the patient suffers from respiratory distress syndrome and the corresponding first label and second label); x ij represents the independent variable, that is, the first set imaging feature after standardization; βj represents the regression coefficient; i∈[1,n], j∈[0,p].

[0164] In an embodiment of the present disclosure, the method for determining whether a patient suffers from respiratory distress syndrome further includes: obtaining an imaging fusion model; using the imaging fusion model to fuse the first set imaging feature or the second set imaging feature to obtain a fusion feature; splicing the first set imaging feature or the second set imaging feature and the fusion feature to obtain a spliced ​​feature vector; and using a preset classifier and the spliced ​​feature vector to determine whether a patient suffers from respiratory distress syndrome.

[0165] In the embodiment of the present disclosure and other possible embodiments, the imaging fusion model may be the imaging fusion model based on a PC algorithm or a neural network.

[0166] In the embodiments of the present disclosure and other possible embodiments, principal component analysis (PCA) is used to reduce feature dimensionality through feature mapping and has become a widely used feature dimensionality reduction algorithm. However, the PCA algorithm actually completes the feature fusion task during the feature mapping process.

[0167] The mathematical expressions corresponding to the PCA algorithm are shown in equations (12) to (14). First, the singular value decomposition (SVD) algorithm is used to obtain the eigenvalues ​​(λ1,λ2,λ3,…,λk) corresponding to the m×n-dimensional original feature (the first set imaging feature or the second set imaging feature) matrix Am×n and the eigenvectors (ξ1,ξ2,ξ3,…,ξk) corresponding to the eigenvalues; then, the k-dimensional eigenvectors (ξ1,ξ2,ξ3,…,ξk) are used to form the transformation matrix Pk×n=(ξ1,ξ2,ξ3,…,ξk)k×n; finally, the original feature matrix Am×n is multiplied by the transformation matrix Pk×n to obtain the final dimensionality reduction matrix (fusion feature) Bm×k.

[0168] A T A=(UΣV T ) T UΣV T =VΣ T U T UΣV T =VΣ T ΣV T =VΣ 2 V T (12)

[0169]

[0170] Among them, Am×n=(a1,a2,a3,…,an) represents the original feature matrix (the first set imaging feature or the second set imaging feature); U m×ma and V n×n represent the two orthogonal matrices obtained by the SVD algorithm respectively; ∑m×n=(σ1,σ2,σ3,…,σk) is a diagonal matrix, σi (i=1-k) is the i-th eigenvalue of the matrix ATA, (λ1,λ2,λ3,…,λk) represents the eigenvalue corresponding to the original feature matrix Am×n; (ξ1,ξ2,ξ3,…,ξk) represents the eigenvector corresponding to the eigenvalue (λ1,λ2,λ3,…,λk); Pk×n=(ξ1,ξ2,ξ3,…,ξk)k×n represents the transformation matrix constructed using the k-dimensional eigenvector (ξ1,ξ2,ξ3,…,ξk); Bm×k=(b1,b2,b3,…,bk) represents the final dimensionality reduction matrix (fusion feature).

[0171] In the embodiments of the present disclosure and other possible embodiments, the imaging fusion model corresponding to the neural network includes at least an input layer, a hidden layer, and an output layer. The parameters corresponding to the input layer, hidden layer, and output layer of the neural network are trained using first set imaging features or second set imaging features corresponding to a set number of patients with dyspnea to obtain a trained neural network. Furthermore, based on the trained neural network, the first set imaging features or second set imaging features are fused to obtain a fused feature.

[0172] Step S102 : selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label.

[0173] In an embodiment of the present disclosure, the method for selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first tag, and the second tag includes: obtaining the setting parameter selection model; using the setting parameter selection model to establish the second setting parameter associated with the first tag and the second tag, and completing the selection of the second setting parameter from the first setting parameter.

[0174] In the embodiments of the present disclosure, it is finally determined that the five core parameters of the ventilator, namely tidal volume, respiratory rate, peak airway pressure, positive end-expiratory pressure, and inspiratory oxygen concentration, are tested in two ventilation modes of the ventilator, namely volume-controlled ventilation (VCV) mode and pressure-controlled ventilation (PCV) mode.

[0175] In the embodiment of the present disclosure and other possible embodiments, the setting parameter selection model may be a parameter selection model based on the Lasso algorithm or the GLM model.

[0176] In the embodiments of the present disclosure and other possible embodiments, the least absolute shrinkage and selection operator (Lasso) algorithm compresses variables with larger parameter estimates to smaller ones and compresses variables with smaller parameter estimates to zero. Currently, the Lasso algorithm has become an effective means of feature selection.

[0177] The mathematical expression corresponding to the Lasso algorithm is shown in formula (1).

[0178]

[0179] In formula (1), x ij represents the independent variable, that is, the first set parameter after standardization, yi represents the dependent variable (the first label and the second label corresponding to whether the patient suffers from respiratory distress syndrome), λ represents the penalty parameter (λ≥0), βj represents the regression coefficient, i∈[1,n], j∈[0,p].

[0180] In the embodiments of the present disclosure and other possible embodiments, the generalized linear model (GLM) is an extension of the linear model, which aims to establish the relationship between the mathematical expectation of the response variable and the linear combination of the predictor variables through the link function. Unlike the Lasso algorithm, the GLM can calculate the expected value of each independent variable x ij The R2 value of (feature) is used to select features. The mathematical expression of GLM is shown in formula (2).

[0181]

[0182] Among them, the connection function The average With linear predictor Establish a connection; yi represents the dependent variable (whether the patient suffers from respiratory distress syndrome and the corresponding first label and second label); x ij represents the independent variable, that is, the first setting parameter after standardization; βj represents the regression coefficient; i∈[1,n], j∈[0,p].

[0183] In the embodiments of the present disclosure and other possible embodiments, the second setting parameter selected from the first setting parameter may be one or more of the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory frequency and / or positive end-expiratory pressure.

[0184] Step S103: Acquire the second setting parameters corresponding to the subject with respiratory distress syndrome wearing a ventilator, and complete the ventilator rapid detection.

[0185] In the embodiments of the present disclosure and other possible embodiments, the second setting parameter may be one or more of the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory rate and / or the positive end-expiratory pressure.

[0186] In the embodiments of the present disclosure and other possible embodiments, the sensor for obtaining the second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator includes: a first flow sensor, a second flow sensor, a first pressure sensor and an oxygen concentration sensor; the first flow sensor, the first pressure sensor and the oxygen concentration sensor are arranged at the air supply port of the ventilator, and are respectively used to detect the air supply flow rate of the ventilator / set the first tidal volume, air supply pressure and air supply oxygen concentration of the set lung model or the subject; the second flow sensor is arranged at the collection port of the ventilator, and is used to detect the second tidal volume of the set lung model or the subject.

[0187] In the embodiments of the present disclosure and other possible embodiments, it also includes: a processor; the processor is respectively connected to the first flow sensor, the second flow sensor, the first pressure sensor and the oxygen concentration sensor; the processor is used to determine the respiratory rate based on the first tidal volume and the second tidal volume and / or determine the positive end-expiratory pressure according to the air supply pressure; or, it also includes: a processor and a second pressure sensor; the processor is respectively connected to the first flow sensor, the second flow sensor, the first pressure sensor and the oxygen concentration sensor; the second pressure sensor is configured on the outside of the chest of the set lung model or the subject, and is used to detect the respiratory cycle of the set lung model or the subject; the processor is used to determine the respiratory rate based on the respiratory cycle and / or determine the positive end-expiratory pressure according to the air supply pressure.

[0188] In the embodiment of the present disclosure and other possible embodiments, an analog-to-digital conversion circuit is provided between the processor and the oxygen concentration sensor; the analog-to-digital conversion circuit is configured to convert the analog value of the oxygen concentration of the supplied air into a digital value. Preferably, the first flow sensor, the second flow sensor, and the first pressure sensor are connected to the processor via an IIC bus.

[0189] In the embodiments of the present disclosure and other possible embodiments, the IIC bus includes: a data line and a clock line; the data line and the clock line are respectively connected to one end of a first pull-up resistor and a second pull-up resistor, and the other ends of the first pull-up resistor and the second pull-up resistor are connected to a set power supply.

[0190] In the embodiments of the present disclosure and other possible embodiments, the system further includes: a display mechanism, which is connected to the processor and the oxygen concentration sensor respectively; the display mechanism is used to display the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory frequency and / or the positive end-expiratory pressure.

[0191] In the embodiment of the present disclosure and other possible embodiments, the display device is connected to the processor via a serial port and / or Ethernet. A memory is provided between the display device and the processor; the memory is used to store the ventilator's air flow rate, the first tidal volume of the set lung model or the subject, and / or the second tidal volume of the set lung model or the subject, and / or the respiratory rate and / or positive end-expiratory pressure.

[0192] In the embodiments of the present disclosure and other possible embodiments, it also includes: an atmospheric pressure sensor; the atmospheric pressure sensor is used to detect the atmospheric pressure value to calibrate the air flow rate of the ventilator / the first tidal volume of the set lung model or the subject and the second tidal volume of the set lung model or the subject.

[0193] The ventilator rapid detection method may be executed by a ventilator rapid detection device. For example, the ventilator rapid detection method may be executed by a terminal device, a server, or other processing device, wherein the terminal device may be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the ventilator rapid detection method may be implemented by a processor calling computer-readable instructions stored in a memory.

[0194] Those skilled in the art will understand that in the above-mentioned method of the specific implementation method, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0195] The disclosed embodiment also proposes a ventilator rapid detection device, which includes: an acquisition unit, used to obtain a first label corresponding to a patient with respiratory distress syndrome, a second label corresponding to a patient without respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; a selection unit, used to select a second setting parameter from the first setting parameters based on a setting parameter selection model, the first label, and the second label; and a detection unit, used to obtain the second setting parameter corresponding to a subject with respiratory distress syndrome when wearing a ventilator, to complete the ventilator rapid detection.

[0196] In some embodiments, the functions or modules contained in the device provided by the embodiments of the present disclosure can be used to execute the ventilator rapid detection method described in the above method embodiment. Its specific implementation can refer to the description of the above ventilator rapid detection method embodiment. For the sake of brevity, it will not be repeated here.

[0197] Figure 3 A block diagram of a rapid detection device for a ventilator according to an embodiment of the present disclosure is shown. Figure 3 As shown, the ventilator rapid detection device includes: a first flow sensor 1, a second flow sensor 2, a first pressure sensor 3 and an oxygen concentration sensor 4; the first flow sensor 1, the first pressure sensor 3 and the oxygen concentration sensor 4 are arranged at the air supply port of the ventilator 5, and are respectively used to detect the air supply flow / set lung model or the first tidal volume, air supply pressure and air supply oxygen concentration of the ventilator 5; the second flow sensor 2 is arranged at the collection port of the ventilator 5, and is used to detect the second tidal volume of the set lung model or the subject. It can realize the rapid detection of the set respiratory parameters to solve the current problem of low timeliness and speed of detection. Specifically, the present disclosure combines the clinical needs of epidemic prevention work, and for ventilator (invasive ventilator, emergency ventilator, etc.) products, according to their working principles and application scenarios, confirms the detection methods of their core performance parameters and key safety indicators, and develops corresponding on-site rapid detection devices to solve the problem of rapid confirmation of the safety and effectiveness of the clinical use of ventilators, and effectively improves the emergency response capabilities of medical institutions to prevent and deal with major epidemics and other public health emergencies.

[0198] In the embodiment of the present disclosure and other possible embodiments, the first tidal volume of the lung model or the subject is set as the tidal volume of inspiration; and at the same time, the second tidal volume of the lung model or the subject is set as the tidal volume of expiration.

[0199] In an embodiment of the present disclosure, the detection device further includes a processor 6; the processor 6 is connected to the first flow sensor 1, the second flow sensor 2, the first pressure sensor 3, and the oxygen concentration sensor 4, respectively; the processor 6 is configured to determine the respiratory rate based on the first tidal volume and the second tidal volume and / or determine the positive end-expiratory pressure based on the supplied air pressure. The respiratory waveforms are measured, the respiratory cycle T is read, and the respiratory rate is calculated by dividing the respiratory cycle T by 60.

[0200] In the embodiment of the present disclosure and other possible embodiments, the processor 6 is configured to determine the respiratory frequency based on the first tidal volume and the second tidal volume, specifically including: the processor 6 extracting the first cycle of the first tidal volume and the second cycle of the second tidal volume, calculating the ratio of the first cycle to the second cycle by the processor 6; and configuring the ratio as the respiratory frequency by the processor 6. Alternatively, the processor 6 extracts the first cycle of the first tidal volume or the second cycle of the second tidal volume, and calculating the inverse of the first cycle or the second cycle by the processor 6 to obtain the respiratory frequency.

[0201] In the embodiment of the present disclosure and other possible embodiments, the detection device is characterized by further comprising: a display mechanism 10, the display mechanism 10 being connected to the processor 6 and the oxygen concentration sensor 4 respectively; the display mechanism 10 being used to display in real time the air flow rate / the first tidal volume of the set lung model, the air pressure, and the air oxygen concentration of the ventilator 5. The first cycle of the first tidal volume and the second cycle of the second tidal volume can be displayed on the display mechanism 10, and the processor 6 can then calculate the ratio of the first cycle to the second cycle; the processor 6 configures the ratio as the respiratory rate.

[0202] In the embodiment of the present disclosure and other possible embodiments, the processor 6 is used to determine the positive end-expiratory pressure based on the air supply pressure, specifically including: the processor 6 extracts the pressure value at the end of expiration within the cycle of the air supply pressure, and at the same time, the pressure value at the end of expiration within the cycle of the air supply pressure can be displayed through the display mechanism 10, and the processor 6 can be used to configure the pressure value at the end of expiration.

[0203] Or, in an embodiment of the present disclosure, the detection device further includes: a processor 6 and a second pressure sensor (not shown in the figure); the processor 6 is respectively connected to the first flow sensor 1, the second flow sensor 2, the first pressure sensor 3 and the oxygen concentration sensor 4; the second pressure sensor is arranged on the outside of the chest of the set lung model or the subject, for detecting the respiratory cycle of the set lung model or the subject; the processor 6 is used to determine the respiratory rate based on the respiratory cycle and / or determine the positive end-expiratory pressure according to the air supply pressure.

[0204] In the embodiment of the present disclosure and other possible embodiments, the second pressure sensor may be a pressure sensor configured on a chest strap. Respiratory changes cause changes in the chest strap tension cycle (respiratory cycle), thereby determining the respiratory rate. Specifically, the display mechanism 10 can display the respiratory cycle of the set lung model or the subject, and the respiratory rate can be determined based on the respiratory cycle.

[0205] In the embodiments of the present disclosure and other possible embodiments, the processor 6 can be implemented using an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD / PLC), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor (e.g., a single-chip microcomputer) or other electronic components.

[0206] In the embodiment of the present disclosure and other possible embodiments, the first flow sensor 1, the second flow sensor 2, and the first pressure sensor 3 are respectively ADP810 series produced by Aosong Electronics Co., Ltd., which have digital I 2 C interface, which can be easily connected to the processor 6, with a pressure range of up to ±500Pa (±2inchH_2O / ±5mbar).

[0207] In the embodiment of the present disclosure, an analog / digital conversion circuit 8 is provided between the processor 6 and the oxygen concentration sensor 4. The analog / digital conversion circuit 8 is configured to convert the analog value of the oxygen concentration of the supplied air into a digital value. The analog / digital conversion circuit 8 is conventional in the art and will not be described in detail herein.

[0208] In an embodiment of the present disclosure, the first flow sensor 1 , the second flow sensor 2 , and the first pressure sensor 3 are connected to the processor 6 via an IIC bus 7 .

[0209] In the embodiment of the present disclosure, the IIC bus 7 includes: a data line 71 and a clock line 72; the data line 71 and the clock line 72 are respectively connected to one end of a first pull-up resistor and a second pull-up resistor, and the other end of the first pull-up resistor and the second pull-up resistor are connected to a set power supply VCC. Specifically, the IIC bus 7 (Inter Integrated Circuit, I 2 C) It can realize full-duplex synchronous data transmission, with a bidirectional serial data line SDA (Serial data) and a serial clock line SCL (Serial clock line). 2 The C bus generally implements a one-master-multiple-slave working mode. There are multiple slaves on the bus. The Master sends the slave's device address (Device Address) through the SDA line, and the corresponding slave sends a response signal to establish a connection, realizing a group of Master and Slave working. This arbitration mechanism avoids slave data collision and data error. At the same time, I 2 The C bus uses an open-drain driver internally, and pull-up resistors are required on SDA and SCL to keep their default state high.

[0210] In the embodiment of the present disclosure and other possible embodiments, the first pull-up resistor and the second pull-up resistor can be configured as 10k. Since the SCL frequency is 100kHz and the clock frequency of the processor 6 is 50MHz, a frequency divider is provided between the processor 6 and the clock line 72.

[0211] In an embodiment of the present disclosure, the detection device is characterized in that it includes a display mechanism 10, which is respectively connected to the processor 6 and the oxygen concentration sensor 4; the display mechanism 10 is used to display the air flow rate of the ventilator 5 / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory rate and / or the positive end-expiratory pressure.

[0212] In the embodiment of the present disclosure and other possible embodiments, the display mechanism 10 may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensor may not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0213] In the embodiment of the present disclosure, the display mechanism 10 is connected to the processor 6 via a serial port and / or Ethernet.

[0214] In the embodiments of the present disclosure and other possible embodiments, the display mechanism 10 and the processor 6 are configured with a communication module to achieve wired or wireless communication. The display mechanism 10 and the processor 6 can use the communication module to access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication module receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication module also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0215] In an embodiment of the present disclosure, a memory 11 is provided between the display mechanism 10 and the processor 6; the memory 11 is used to store the air flow of the ventilator 5 / the first tidal volume of the set lung model or the subject and / or the second tidal volume of the set lung model or the subject and / or the respiratory frequency and / or the positive end-expiratory pressure.

[0216] In the embodiments of the present disclosure and other possible embodiments, the memory 11 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0217] In an embodiment of the present disclosure, the detection device further includes: an atmospheric pressure sensor; the atmospheric pressure sensor is used to detect the atmospheric pressure value to calibrate the air supply flow rate / set lung model or the first tidal volume of the subject and the second tidal volume of the set lung model or the subject of the ventilator 5.

[0218] The present disclosure also proposes a ventilator, comprising: the detection device as described above, characterized in that the ventilator 5 is connected to an input mechanism 12; wherein the input mechanism 12 is used to set the ventilation mode of the ventilator 5.

[0219] In an embodiment of the present disclosure, the input mechanism 12 may be a keyboard, a click wheel, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0220] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement the above-mentioned ventilator rapid detection method. The computer-readable storage medium may be a non-volatile computer-readable storage medium.

[0221] The present disclosure also provides an electronic device comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to implement the aforementioned ventilator rapid detection method. The electronic device may be provided as a terminal, server, or other device.

[0222] Figure 4 8 is a block diagram of an electronic device 800 according to an exemplary embodiment. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, or the like.

[0223] Reference Figure 4 , the electronic device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0224] The processing component 802 generally controls the overall operation of the electronic device 800, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 802 may include one or more modules to facilitate interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

[0225] The memory 804 is configured to store various types of data to support operations on the electronic device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0226] The power supply component 806 provides power to the various components of the electronic device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 800.

[0227] The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the electronic device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0228] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), which is configured to receive external audio signals when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0229] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as a keyboard, click wheel, buttons, etc. These buttons may include but are not limited to: a home button, volume buttons, a start button, and a lock button.

[0230] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the electronic device 800. For example, the sensor assembly 814 can detect the open / closed state of the electronic device 800, the relative positioning of components, such as the display and keypad of the electronic device 800. The sensor assembly 814 can also detect changes in the position of the electronic device 800 or a component of the electronic device 800, the presence or absence of user contact with the electronic device 800, the orientation or acceleration / deceleration of the electronic device 800, and temperature changes of the electronic device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0231] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0232] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0233] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 804 including computer program instructions. The computer program instructions can be executed by the processor 820 of the electronic device 800 to perform the above method.

[0234] Figure 5 1 is a block diagram of an electronic device 1900 according to an exemplary embodiment. For example, the electronic device 1900 may be provided as a server. Figure 5The electronic device 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions executable by the processing component 1922, such as an application. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute the instructions to perform the above-described method.

[0235] The electronic device 1900 may further include a power supply component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input / output (I / O) interface 1958. The electronic device 1900 may operate based on an operating system stored in the memory 1932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or the like.

[0236] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions that can be executed by the processing component 1922 of the electronic device 1900 to perform the above method.

[0237] The present disclosure may be a system, method and / or computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0238] A computer-readable storage medium can be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination thereof. As used herein, a computer-readable storage medium is not to be construed as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through an electrical wire.

[0239] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions to be stored in the computer-readable storage medium in each computing / processing device.

[0240] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions. The electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0241] Various aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer-readable program instructions.

[0242] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device is generated that implements the functions / actions specified in one or more blocks in the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium, where these instructions cause the computer, programmable data processing device, and / or other device to operate in a specific manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0243] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0244] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and the part of the module, program segment or instruction contains one or more executable instructions for realizing the prescribed logical function. In some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the prescribed function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0245] While various embodiments of the present disclosure have been described above, the foregoing description is intended to be illustrative, non-exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or technical improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A rapid detection device for a ventilator, characterized in that: include: an acquiring unit, configured to acquire a first tag corresponding to a patient suffering from respiratory distress syndrome, a second tag corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; Determining whether a patient suffers from respiratory distress syndrome includes: acquiring an image conversion model and determining the type corresponding to a lung imaging CT image; if the type corresponding to the lung imaging CT image is configured as a plain scan CT image, enhancing the lung imaging CT image corresponding to the plain scan CT image using the image conversion model to obtain an enhanced CT image; extracting a first set imaging feature of the enhanced CT image; and determining whether a patient suffers from respiratory distress syndrome using a preset classifier and the first set imaging feature; wherein, enhancing the lung imaging CT image using the image conversion model includes: acquiring a set synthesizer; and based on the set synthesizer, The generator in the synthesizer performs convolution processing on the lung imaging CT image to complete the enhancement of the lung imaging CT image; wherein, the training of the setting synthesizer includes: using the generator of the setting synthesizer to perform convolution processing on the lung imaging CT image corresponding to the plain scan CT image after registration with the enhanced CT image to generate a corresponding synthetic enhanced CT image; based on the synthetic enhanced CT image and the lung imaging CT image corresponding to the plain scan CT image, using a preset discriminator to output the synthetic enhanced CT image generated by the generator to reach the maximum probability value of the corresponding enhanced CT image, thereby completing the parameter training of the generator in the setting synthesizer; a selection unit, configured to select a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label; The detection unit is used to obtain the second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator, and complete the rapid detection of the ventilator.

2. The detection device according to claim 1, characterized in that Before extracting the first set imaging feature of the enhanced CT image, the method further includes: Obtaining a lung region segmentation model and an enhanced CT image corresponding to the chest lung imaging CT image to be segmented; The lung region segmentation model is used to perform lung region segmentation on the enhanced CT image corresponding to the chest lung imaging CT image to obtain a lung imaging CT image.

3. The detection device according to any one of claims 1 or 2, characterized in that The method of using a preset classifier and the first set imaging feature to determine whether the patient suffers from respiratory distress syndrome includes: Get the set imaging selection model; Selecting a second set imaging feature from the first set imaging feature using the imaging selection model, the first label, and the second label; Using a preset classifier and the second set imaging feature, it is determined whether the patient suffers from respiratory distress syndrome.

4. The detection device according to claim 3, characterized in that The method of determining whether a patient suffers from respiratory distress syndrome by using a preset classifier and the first set imaging feature further includes: Obtain an imaging fusion model; Using the imaging fusion model, fusing the first set imaging feature or the second set imaging feature to obtain a fusion feature; splicing the first set imaging feature or the second set imaging feature and the fusion feature to obtain a spliced ​​feature vector; Using a preset classifier and the concatenated feature vector, it is determined whether the patient suffers from respiratory distress syndrome.

5. The detection device according to any one of claims 1, 2 and 4, characterized in that: The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

6. The detection device according to claim 3, characterized in that The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

7. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute: Obtain a first label corresponding to respiratory distress syndrome, a second label corresponding to non-respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; determine whether respiratory distress syndrome is present, including: obtaining an image conversion model and determining the type corresponding to a lung imaging CT image; if the type corresponding to the lung imaging CT image is configured as a plain scan CT image, using the image conversion model to enhance the lung imaging CT image corresponding to the plain scan CT image to obtain an enhanced CT image; extracting a first setting imaging feature of the enhanced CT image; using a preset classifier and the first setting imaging feature to determine whether respiratory distress syndrome is present; wherein, using the image conversion model to enhance the lung imaging CT image The method comprises: obtaining a set synthesizer; performing convolution processing on the lung imaging CT image based on a generator in the set synthesizer to complete the enhancement of the lung imaging CT image; wherein the training of the set synthesizer comprises: using the generator of the set synthesizer to perform convolution processing on the lung imaging CT image corresponding to the plain scan CT image registered with the enhanced CT image to generate a corresponding synthesized enhanced CT image; based on the synthesized enhanced CT image and the lung imaging CT image corresponding to the plain scan CT image, using a preset discriminator to output the synthesized enhanced CT image generated by the generator to reach the maximum probability value of the corresponding enhanced CT image, thereby completing the parameter training of the generator in the set synthesizer; selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label; The second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator is obtained to complete the ventilator rapid detection.

8. The electronic device according to claim 7, wherein: Before extracting the first set imaging feature of the enhanced CT image, the method further includes: Obtaining a lung region segmentation model and an enhanced CT image corresponding to the chest lung imaging CT image to be segmented; The lung region segmentation model is used to perform lung region segmentation on the enhanced CT image corresponding to the chest lung imaging CT image to obtain a lung imaging CT image.

9. The electronic device according to any one of claims 7 or 8, characterized in that: The method of using a preset classifier and the first set imaging feature to determine whether the patient suffers from respiratory distress syndrome includes: Get the set imaging selection model; Selecting a second set imaging feature from the first set imaging feature using the imaging selection model, the first label, and the second label; Using a preset classifier and the second set imaging feature, it is determined whether the patient suffers from respiratory distress syndrome.

10. The electronic device according to claim 9, characterized in that The method of determining whether a patient suffers from respiratory distress syndrome by using a preset classifier and the first set imaging feature further includes: Obtain an imaging fusion model; Using the imaging fusion model, fusing the first set imaging feature or the second set imaging feature to obtain a fusion feature; splicing the first set imaging feature or the second set imaging feature and the fusion feature to obtain a spliced ​​feature vector; Using a preset classifier and the concatenated feature vector, it is determined whether the patient suffers from respiratory distress syndrome.

11. The electronic device according to any one of claims 7, 8, and 10, characterized in that: The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

12. The electronic device according to claim 11, wherein: The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

13. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, they: Obtaining a first tag corresponding to a patient suffering from respiratory distress syndrome, a second tag corresponding to a patient not suffering from respiratory distress syndrome, and a first setting parameter of the ventilator in a set ventilation mode; Determining whether a patient suffers from respiratory distress syndrome includes: acquiring an image conversion model and determining the type corresponding to a lung imaging CT image; if the type corresponding to the lung imaging CT image is configured as a plain scan CT image, enhancing the lung imaging CT image corresponding to the plain scan CT image using the image conversion model to obtain an enhanced CT image; extracting a first set imaging feature of the enhanced CT image; and determining whether a patient suffers from respiratory distress syndrome using a preset classifier and the first set imaging feature; wherein, enhancing the lung imaging CT image using the image conversion model includes: acquiring a set synthesizer; and based on the set synthesizer, The generator in the synthesizer performs convolution processing on the lung imaging CT image to complete the enhancement of the lung imaging CT image; wherein, the training of the setting synthesizer includes: using the generator of the setting synthesizer to perform convolution processing on the lung imaging CT image corresponding to the plain scan CT image after registration with the enhanced CT image to generate a corresponding synthetic enhanced CT image; based on the synthetic enhanced CT image and the lung imaging CT image corresponding to the plain scan CT image, using a preset discriminator to output the synthetic enhanced CT image generated by the generator to reach the maximum probability value of the corresponding enhanced CT image, thereby completing the parameter training of the generator in the setting synthesizer; selecting a second setting parameter from the first setting parameter based on a setting parameter selection model, the first label, and the second label; The second setting parameter corresponding to the subject with respiratory distress syndrome wearing a ventilator is obtained to complete the ventilator rapid detection.

14. The computer-readable storage medium according to claim 13, wherein: Before extracting the first set imaging feature of the enhanced CT image, the method further includes: Obtaining a lung region segmentation model and an enhanced CT image corresponding to the chest lung imaging CT image to be segmented; The lung region segmentation model is used to perform lung region segmentation on the enhanced CT image corresponding to the chest lung imaging CT image to obtain a lung imaging CT image.

15. The computer-readable storage medium according to any one of claims 13 or 14, wherein: The method of using a preset classifier and the first set imaging feature to determine whether the patient suffers from respiratory distress syndrome includes: Get the set imaging selection model; Selecting a second set imaging feature from the first set imaging feature using the imaging selection model, the first label, and the second label; Using a preset classifier and the second set imaging feature, it is determined whether the patient suffers from respiratory distress syndrome.

16. The computer-readable storage medium according to claim 15, wherein: The method of determining whether a patient suffers from respiratory distress syndrome by using a preset classifier and the first set imaging feature further includes: Obtain an imaging fusion model; Using the imaging fusion model, fusing the first set imaging feature or the second set imaging feature to obtain a fusion feature; splicing the first set imaging feature or the second set imaging feature and the fusion feature to obtain a spliced ​​feature vector; Using a preset classifier and the concatenated feature vector, it is determined whether the patient suffers from respiratory distress syndrome.

17. The computer-readable storage medium according to any one of claims 13, 14, and 16, wherein: The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

18. The computer-readable storage medium according to claim 15, wherein: The selecting the second setting parameter from the first setting parameter based on the setting parameter selection model, the first label, and the second label includes: Get the setting parameter selection model; The setting parameter selection model is used to establish a second setting parameter associated with the first tag and the second tag, thereby completing the selection of the second setting parameter from the first setting parameter.

Images