Method for evaluating human lung injury caused by explosion based on medical indexes

By combining ultrasonic image processing and deep learning technology and combined with flow-solid coupling numerical calculation, the problem of difficulty in assessing lung damage caused by explosion when wearing personal protective equipment is solved, and rapid and accurate lung injury assessment and diagnosis on the battlefield is achieved , improves the efficiency and accuracy of emergency medical care.

CN118942697APending Publication Date: 2024-11-12BEIJING INST OF TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410980302.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The prior art is difficult to accurately evaluate the lung damage caused by explosions when wearing personal protective equipment, especially due to inaccurate measurement of shock wave pressure due to chest contact with protective equipment.

Method used

Using a method based on medical indicators, using Python scripts to read lung ultrasound images, combined with the LS-DYNA explosion field and the numerical calculation of the flow-solid coupling of human lungs, evaluate lung damage through deep learning neural network, and combine with handheld ultrasound equipment Perform real-time scans and ratings.

Benefits of technology

It has achieved rapid and accurate diagnosis and injury assessment of lung injury in modern battlefields and complex conflict sites, shortening rescue time and improving treatment success rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942697A_ABST
    Figure CN118942697A_ABST
Patent Text Reader

Abstract

The invention provides a method for evaluating human lung injury caused by explosion based on medical indexes. The method comprises the following specific steps: scanning each lung region through a handheld ultrasonic device, inputting an obtained lung video into a deep learning model, carrying out dynamic score estimation on a video stream by the model based on image feature learning, summing the maximum value, and finally outputting the total score of the lung image and the explosion impact injury condition. The deep learning network is based on a human body-explosion flow field numerical model, the overall impact response of the chest is obtained, and a main injury causing mechanism of lung explosion injury is clarified; and different lung injury grades are obtained by changing the explosion equivalent and distance. As an auxiliary emergency medical means, the method can quickly complete lung explosion injury diagnosis and injury condition evaluation of the wounded so as to determine a diagnosis and treatment scheme, the rescue time is shortened, and the success rate is increased.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of medical image computing technology and personnel impact injury, and specifically relates to a method for evaluating explosion-induced human lung injury based on medical indicators. Background Art

[0002] Human beings are the organizers and implementers of tasks such as emergency response, stability maintenance, and combat operations, and their life safety is the key to success or failure. In the process of maintaining stability and emergency response and war, personnel are inevitably faced with threats such as explosions. The lungs are the main target organs of explosion shock waves, and lung injury is also a key factor in the death of explosion injuries. Research on lung explosion injuries has been carried out, but related research is mainly based on biological effect tests. Due to the huge difference between the physiological structure of animals and humans, it is difficult to obtain the true response of various organs in the human chest through such research. A chest model was established, and lung explosion injury was studied based on the finite element method, but the model was too simplified and the injury response obtained was limited. The main methods for evaluating lung explosion injury are the Bowen injury curve, the Axelsson model, and the Stuhmiller model. All three methods require measuring the shock wave pressure acting on the chest, and then obtaining the injury level by comparing it with the injury curve or calculating the injury index. However, when wearing personal protective equipment, the chest is in direct contact with the protective equipment, and the pressure sensor installed on the chest surface will be squeezed, making it difficult to accurately measure the shock wave pressure after attenuation by the protective equipment. Therefore, the above methods cannot be used for injury assessment. The US NIJ 0117 "Public Safety Explosive Ordnance Disposal Clothing Standard" does not assess human injuries when evaluating the protective performance of explosive ordnance disposal clothing, because the standard believes that there is currently no mature and reliable evaluation method.

[0003] Lung ultrasound is the main means of lung imaging examination. Compared with ultrasound, CT is accurate but requires transportation and is radioactive. X-rays are inaccurate and radioactive. MRI is time-consuming and expensive and is not suitable for critically ill patients. Ultrasound is accurate and has a good correlation with CT. It is also portable, convenient, non-radioactive and highly repeatable.

[0004] Ultrasound assessment of lung blast injury is a purposeful, rapid, and effective ultrasound assessment for critically ill patients. Its examination results can provide a basis for doctors to make or adjust clinical diagnosis and treatment decisions. It is impossible for clinicians to examine the intercostal space of each patient one by one. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method for evaluating explosion-induced lung damage in humans based on medical indicators, which can better solve the problems of slow numerical simulation speed, low model accuracy, and difficulty in implementation.

[0006] The present invention is implemented by the following scheme:

[0007] A method for evaluating explosion-induced lung injury in humans based on medical indicators comprises the following steps:

[0008] Step 1: Based on the image features of normal / impact-injured lungs, a large number of lung ultrasound images are read in parallel using Python scripts, and input into the ResNet network model for feature extraction;

[0009] Step 2: Based on the numerical calculation of the fluid-solid coupling between the explosion field and the human lungs using LS-DYNA, the response of the lungs under the explosion load under different working conditions is obtained by changing the explosion parameters and the personnel position information, and the scores are marked in conjunction with the ultrasound images of the damaged lungs obtained in step 1.

[0010] Step 3: using the image set annotated in step 2 for learning and training the deep learning neural network of the method to obtain a scoring module for lung injury;

[0011] Step 4: Use a handheld ultrasound device to scan each lung area separately, and input the obtained lung video into the deep learning model. The model dynamically estimates the video stream based on image feature learning and sums the maximum value, and finally outputs the total lung image score and blast impact damage.

[0012] Beneficial effects:

[0013] This invention innovatively proposes a method for assessing explosion-induced lung injury based on medical indicators. Faced with modern battlefields and complex conflict scene environments, existing conventional lung explosion shock injury treatment methods are limited, and the injured need to be transferred to rear hospitals for further diagnosis and treatment, which may miss the best time for treatment. This method is based on portable ultrasound equipment, which scans the lungs of the injured in real time, and cooperates with the deep learning module to score the lung ultrasound images to finally obtain the lung injury status. Assisted emergency medical means can quickly complete the diagnosis and injury assessment of the injured lung explosion injury to determine the diagnosis and treatment plan, shortening the rescue time and success rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flow chart of the method of the present invention;

[0015] Figure 2 is a schematic diagram of a typical transfer learning model network adopted by the present invention;

[0016] Figure 3 The process of building the network model DETAILED DESCRIPTION

[0017] In order to make the technical problems solved by the method, the technical solutions adopted and the technical effects achieved more detailed, the method is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the present invention rather than to limit the present invention. In addition, it should be noted that for ease of description, only the relevant parts of the method rather than the entire content are shown in the accompanying drawings.

[0018] Figure 1 A flow chart of the method steps of the present invention is given, and the specific method steps are as follows:

[0019] Step 1: Scan each lung area separately using a handheld ultrasound scanner to obtain a lung ultrasound image video stream;

[0020] Specifically, it is divided into the description of local features and the description of global features. The 3D shape descriptor is used to establish the mapping relationship between corresponding points between surfaces. A vector is used to describe the shape characteristics of specified points and areas on the surface of the human body model, and the value of the matching vector is used to describe the correspondence between different surface points.

[0021] Step 2: Input the image data into the deep learning model and add the fluid-structure interaction simulation module to assist;

[0022] The host inputs the acquired ultrasound image video stream to the artificial intelligence model, and the artificial intelligence model performs dynamic scoring based on the input ultrasound image video stream, and records the maximum score of each lung area after scanning each lung area. Each lung area acquires an ultrasound video, which can be separated into multiple ultrasound images and multiple scores. The maximum score of a lung area indicates the most serious damage to that lung area.

[0023] The network model is generally divided into four layers. The first layer is the input feature attributes, including grayscale value, regional morphology, cavity size and number, etc.; the second layer is the input quantity belonging to the cloud distribution, which classifies the first layer feature quantities and calculates the activation degree of the belonging cloud distribution; the third layer is several layers of convolutional neural network layers, which perform rule reasoning and logic establishment, and also includes a finite element simulation field output and input module to assist in establishing the logical relationship between the lung injury level and ultrasound image information; the fourth layer is the output prediction layer, which weights the various output values ​​of the neural network through the previous levels of logic output to obtain the personnel's chest and lung injury score assessment. The higher the score, the more serious the injury.

[0024] Specifically, the present invention uses the ALE fluid-solid coupling algorithm to perform finite element simulation of lung injury. The lung model structure boundary adopts the characteristics of the Lagrangian method to effectively track the material boundary movement; the internal grid is independent of the material entity, which makes up for the shortcomings of the Euler grid. The basic control equations used in this method are:

[0025] (1) Mass conservation equation:

[0026]

[0027] (2) Momentum conservation equation:

[0028]

[0029] (3)Energy conservation equation:

[0030]

[0031] Where ρ is the density, σ ij is the stress tensor, p is the pressure, is the Kronecker function, which contains the relationship between the body force per unit mass and the internal energy per unit mass.

[0032] The model specifications and parameters used in this method are as follows Figure 2 The network model building process is as follows Figure 3 shown.

[0033] Furthermore, the method for constructing a deep learning neural network model includes: annotating the acquired normal / damaged lung images with scores as training data samples and test data samples for training and testing the deep convolutional neural network model. In this example, during the data collection process, all lung ultrasound data annotated with scores of 0-7 will be included in the data set for targeted training of the model. The score table is shown in Table 1.

[0034] Table 1 Scoring values ​​of lung ultrasound images

[0035]

[0036]

[0037] This invention is a projection of AI-enabled medical treatment on handheld ultrasound equipment, and is a lung injury assessment method based on medical indicators. In the lung blast injury assessment scenario, handheld ultrasound with AI technology can quickly and accurately determine the degree of lung blast injury in real time during the ultrasound scan by medical staff. With the assistance of AI technology, even medical staff without rich experience in using ultrasound can quickly interpret lung blast injury images and make corresponding first aid measures, follow-up inspection arrangements and preparations for surgery.

[0038] Through the above steps, the assessment of explosion-induced lung injury in the human body based on medical indicators was completed, the lungs of the injured were scanned in real time, and the lung ultrasound images were scored with the help of the deep learning module to finally determine the lung injury situation. It assisted emergency medical means, and can quickly complete the diagnosis and injury assessment of the injured person's lung explosion injury to determine the diagnosis and treatment plan, shorten the rescue time and success rate, and has good usability, and can be used for personnel impact and post-injury treatment.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating explosion-induced lung damage in humans based on medical indicators, characterized in that: The following steps are involved: The first step is to scan the person's lungs with a handheld ultrasound device to obtain an ultrasound image video stream; The second step is to input the video stream into the deep learning network model for feature comparison learning and training, and then input it into the ResNet network model; The third step is to use LS-DYNA to calculate the fluid-solid coupling between the explosion field and the human lung. By changing the explosion parameters and the personnel position information, the response of the lung under the explosion load under different working conditions is obtained, and the score is marked in conjunction with the ultrasound image of the damaged lung in the first step. The fourth step is to use the image set annotated in step 2 for learning and training the deep learning neural network of this method to obtain a scoring module for lung injury. The network model is generally divided into four layers: The first layer is the input feature attributes, including grayscale value, regional morphology, hole size and number; The second layer is the input quantity belonging to the cloud distribution, which classifies the first layer feature quantity and calculates the activation degree of the belonging cloud distribution; The third layer is a number of convolutional neural network layers for rule reasoning and logic establishment; The fourth layer is the output prediction layer, which weights the output values ​​of the neural network through the logical outputs of the previous levels to obtain a score assessment of the person's chest and lung injuries. The higher the score, the more serious the injury.

2. The method for evaluating explosion-induced lung damage in humans based on medical indicators according to claim 1, characterized in that: The method for constructing the deep learning network model involved includes: obtaining normal / damaged lung medical images, and annotating the obtained medical images with scores as training sets and test sets.

3. The method for evaluating explosion-induced lung damage in humans based on medical indicators according to claim 1, characterized in that: The deep learning network model training process involved incorporates the LS-DYNA-based explosion field and lung numerical simulation method as an aid to the sample set labeling score.

4. The method for evaluating explosion-induced lung damage based on medical indicators according to claim 1, characterized in that: The deep learning model involved includes an input convolution layer, multiple inverse residual convolution layers, a mean pooling layer, a fully connected layer, and a Softmax layer, which are set in sequence.

5. The method for evaluating explosion-induced lung damage based on medical indicators according to claim 3 is characterized in that: The numerical simulation method of auxiliary scoring and annotation involved uses the Axelsson damage model to evaluate the damage caused by explosion impact to the chest and abdomen of the human body. The chest model consists of a mass block, a spring and a damper, and its mathematical expression is: In the formula, M is the equivalent mass, K is the stiffness coefficient, A is e is the equivalent area, p0 is the ambient pressure, V0 is the lung volume at the initial moment, p(t) is the explosion load on the chest, γ is the polytropic index of the gas.

Citation Information

Cited By

  • Block chain-based explosion injury remote diagnosis and treatment data intelligent storage system

    CN122393019A