Lung function, respiratory muscle function measurement method, system and apparatus

Abstract

The application discloses a lung function, respiratory muscle function determination method, system and device, and relates to the technical field of medical detection.The application comprises: MR image data collection, MR image data preprocessing, forming an image data set by MR post-processing image data; dividing the image data set into a training set, a verification set and a test set according to a ratio of 5:2:1; and constructing a UNet segmentation model and training the model by using the image data set.The application innovatively adopts a deep learning method to post-process chest MRI dynamic images, obtains quantitative indexes capable of comprehensively evaluating diaphragm structure and function of hemodialysis patients, and quantitatively describes the diaphragm function condition of the population, thereby making up for the shortcomings that a lung function instrument cannot early identify patients with relatively light respiratory muscle lesions and lung function in a compensatory stage, cannot indicate the position of respiratory muscle lesions, and cannot monitor the lung function of patients with poor lung function due to the inability to complete the lung function instrument examination.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical detection, in particular to a lung function and respiratory muscle function determination method, system and device. BACKGROUND

[0002] The importance of respiratory function to the human body is self-evident. Respiratory movement not only directly affects lung function, but also affects the hemodynamics of the heart, and further affects heart function. At present, the functional state of the respiratory muscle has not been given enough attention due to the limited clinical evaluation methods. The current clinical evaluation method for lung function is a lung function instrument. The evaluation of respiratory muscle strength and function includes maximum inspiratory pressure, transdiaphragmatic pressure determination, electromyography, and ultrasound. The lung function instrument mainly focuses on understanding the overall situation of lung function. The quality of the examination is closely related to the examination cooperation degree of the patient, and the examination result is affected by airway lesions. Maximum inspiratory pressure is a classic method for testing the strength of the inspiratory muscles. It measures the maximum sustained oral pressure for 1s under a closed airway environment. Transdiaphragmatic pressure determination is to place a balloon catheter in the esophagus and lower part of the stomach through the nose to measure the gastric pressure and esophageal pressure. The difference between the two is the transdiaphragmatic pressure, which is currently considered the gold standard for evaluating respiratory muscle strength. It is the transdiaphragmatic pressure generated by electrical or magnetic stimulation of the cervical phrenic nerve, which eliminates the influence of the degree of respiratory effort and the way of effort on the transdiaphragmatic pressure. Electromyography can be used to distinguish between neurological and muscular causes of respiratory muscle weakness. Ultrasound is a commonly used clinical examination method because it is convenient to examine, has good repeatability, is non-invasive, has no radiation, is low in price, and is easy to popularize.

[0003] However, the lung function examination by the lung function instrument cannot early identify patients with mild respiratory muscle lesions and compensatory lung function. For patients with poor lung function, the lung function cannot be monitored and followed up because they cannot cooperate to complete the examination. For patients with a decline in lung function indicated by the lung function instrument examination, the examination cannot indicate the location of respiratory muscle lesions. Therefore, we propose a lung function and respiratory muscle function determination method, system and device. SUMMARY

[0004] The purpose of the present application is to provide a lung function and respiratory muscle function determination method, system and device to solve the problems raised in the background.

[0005] According to the first aspect of the present application, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a lung function and respiratory muscle function determination method, specifically comprising the following steps:

[0006] Receiving MR image data, pre-processing the MR image data, and screening MR image data with high image quality to form an image data set;

[0007] The image dataset is divided into a training set, a validation set and a test set according to 5:2:1;

[0008] A UNet segmentation model is constructed and trained using the image dataset, the UNet segmentation model is based on a UNet network, an ADN module is introduced after each convolutional layer, and multiple residual modules are introduced into the network;

[0009] The original MR image is processed using the trained UNet segmentation model to obtain a nifti format lung label file;

[0010] According to the coordinate information in the label file, the maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter 1, the minimum C-C diameter 1, the maximum C-C diameter 2, the minimum C-C diameter 2, the maximum A-P diameter, the minimum A-P diameter, the diaphragm movement area and the thoracic movement area, and the lung volume are calculated.

[0011] Further, the MR image data is obtained by scanning with a 3.0T superconducting MR scanner, specifically as follows:

[0012] (1) Under normal breathing conditions, the subject is scanned during tidal breathing, and two-dimensional movie MRI scanning is performed on the chest, each subject is scanned along two axial planes, i.e. the upper edge of the aortic arch layer and the right atrial upper edge layer, one sagittal plane, i.e. the right lung cross section along the clavicular midline, and one coronal plane, i.e. the lung apex layer;

[0013] (2) Under deep breathing conditions, deep inspiration and breath holding, full lung scanning is performed; deep expiration and breath holding, full lung scanning is performed.

[0014] Further, the MR image data is preprocessed, specifically as follows:

[0015] (1) All Dicom format image data is converted to Nifti format using MRICron;

[0016] (2) The signal-to-noise ratio of the image is evaluated using Slicer 5.1.0 software:

[0017] The subcutaneous muscle (chest muscle) tissue ROI is selected as the ROI for signal-to-noise ratio evaluation, the ROI is outlined in the Slicer software, and the image enhancement mean and standard deviation of the ROI are extracted, the signal-to-noise ratio SNR is the ratio of the image enhancement mean Mean and the standard deviation SD of the ROI, the calculation formula is: SNR = Mean / SD;

[0018] (3) Use Slicer 5.1.0 software to outline the lung profile in the image multiple times, for patients with pleural effusion, the pleural effusion part is included in the outlined ROI, the average Dice index of the outlined ROI is calculated, and the image with high Dice index is selected as the standard image.

[0019] Further, after dividing the image data set into a training set, a validation set and a test set, it is also necessary to use the RandCropByPosNegLabeld and RandRotate90d functions in monai.transforms to augment the training image.

[0020] Further, the ADN module includes an instance normalization function, a Dropout function and a PReLU activation function.

[0021] The instance normalization function is used to keep the model stable during training, the Dropout function is used to prevent overfitting, and the PReLU function is used to provide nonlinear activation function.

[0022] Further, the residual module is set to multiple, each residual module includes two convolution operations, and part of the residual module residual path is set to identity mapping.

[0023] Further, the up-sampling in the UNet segmentation model is realized by transposed convolution, and the down-sampling is realized by increasing the convolution layer step.

[0024] Further, according to the coordinate information in the Label file, the maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter1, the minimum C-C diameter1, the maximum C-C diameter2, the minimum C-C diameter2, the maximum A-P diamter, the minimum A-P diamter, the diaphragm movement area and the thoracic movement area and the lung volume are calculated:

[0025] (1) Use the nibabel package of Python to read the generated nifti format lung Label file, find the maximum Y coordinate point in the Label, which is the lung vertex, record the Y coordinate minimum value point in the Label from front to back according to X coordinate, record the coordinate array of each point of the diaphragm top lung bottom curve, find the point with the maximum Y value in the diaphragm top lung bottom curve coordinate array, which is the diaphragm top lung bottom highest point, find the minimum X coordinate and the maximum X coordinate with the same Y value in the horizontal direction of the diaphragm top lung bottom highest point, the line connecting the two points is the diaphragm top lung bottom highest point horizontal tangent AP Line1, and the difference between the two X coordinates multiplied by the horizontal length of the voxel is A-Pdiameter.

[0026] (2) Find the minimum point and the maximum point of X value in the diaphragm top lung bottom curve coordinate array, the horizontal distance between the two points is the A-P Line2, and the difference between the X coordinates of the two points multiplied by the horizontal length of the voxel is the A-P diameter2;

[0027] (3) Draw the perpendicular line C-C line of the horizontal tangent line from the lung top point to the highest point of the diaphragm top lung bottom, and the difference between the Y coordinates of the lung top point and the tangent line multiplied by the vertical length of the voxel is the C-C diameter1;

[0028] (4) Calculate the C-C diameter1 of each lung Label in the image data set, record it into the C-C diameter array, observe the fluctuation of the C-C diameter with time, find the maximum value and the minimum value of the array, which are the maximum abdominal deep inspiration phase and the maximum abdominal deep expiration phase respectively, then calculate the number of all voxels in the maximum abdominal deep inspiration phase Label whose coordinates are below the diaphragm top lung bottom curve of the maximum abdominal deep expiration phase Label, and the lung bottom-diaphragm muscle movement area is the product of the obtained voxel number and the voxel area;

[0029] (5) Fit the diaphragm top lung bottom curve of the maximum abdominal deep inspiration phase Label and the minimum abdominal deep inspiration phase Label by using the polynomial fitting method, and calculate the length of the diaphragm top lung bottom curve in the maximum inspiration phase and the maximum expiration phase by using the integral method, which is the maximum lung bottom diaphragm muscle length and the minimum lung bottom diaphragm muscle length:

[0030] (6) Calculate the A-P diameter of each lung Label in the image data set, record it into the A-P diameter array, observe the fluctuation of the A-P diameter with time, find the maximum value and the minimum value of the array, which are the maximum chest deep inspiration phase and the maximum chest deep expiration phase respectively, record the lung front boundary curve coordinates according to the minimum point of the X coordinate of each row of voxels to obtain a coordinate array, calculate the number of all voxels in the maximum chest deep inspiration phase Label whose coordinates are above the diaphragm top lung bottom curve of the maximum chest deep expiration phase Label and in front of the lung front boundary curve, and the thoracic cavity movement area is the product of the obtained voxel number and the voxel area;

[0031] (7) Calculate the total number of voxels with a value of 0 by using the segmented full-lung cross-sectional Label file, and obtain the full-lung volume by multiplying the volume of a single voxel.

[0032] According to the second aspect of the present application, the present application provides a lung function and respiratory muscle function measuring system for implementing the lung function and respiratory muscle function measuring method described above, which specifically comprises the following:

[0033] A preprocessing module is configured to receive MR image data, pre-process the MR image data, and screen MR image data with high image quality to form an image data set.

[0034] a dividing module, configured to divide the image dataset into a training set, a validation set and a test set according to 5:2:1;

[0035] a constructing module, configured to construct a UNet segmentation model and train the UNet segmentation model by using the image dataset, the UNet segmentation model being based on a UNet network, and an ADN module being introduced after each convolutional layer, and a plurality of residual modules being introduced into the network;

[0036] an outputting module, configured to process the original MR image by using the trained UNet segmentation model to obtain a nifti format lung Label file;

[0037] a calculating module, configured to calculate a maximum diaphragm length, a minimum diaphragm length, a maximum C-C diameter 1, a minimum C-C diameter 1, a maximum C-C diameter 2, a minimum C-C diameter 2, a maximum A-P diamter, a minimum A-P diamter, a diaphragm movement area, a thoracic cavity movement area and a lung volume according to coordinate information in the Label file, calculate a mean absolute error of a corresponding parameter according to the above parameters, and determine lung function and respiratory muscle function according to the mean absolute error.

[0038] According to a third aspect of the present application, the present application provides a terminal device comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor, the memory storing a computer program capable of running on the processor, and the processor loading and executing the computer program, the lung function and respiratory muscle function determination method described above being adopted.

[0039] The present application has at least the following beneficial effects:

[0040] 1. The present application innovatively uses deep learning chest MRI dynamic images to comprehensively evaluate the diaphragm structure and function indicators of hemodialysis patients, to quantitatively determine the diaphragm function status of this population, to make up for the shortcomings that the lung function instrument cannot early identify patients with mild respiratory muscle lesions and lung function in the compensatory stage, to make up for the shortcomings that the lung function instrument examination cannot indicate the site of respiratory muscle lesion, to make up for the shortcomings that patients with poor lung function cannot complete the lung function instrument examination due to inability to cooperate, thereby unable to monitor lung function, to make up for the shortcomings that diaphragm ultrasound examination is too dependent on the professional level of the operator and cannot fully reflect the activity of the entire diaphragm, and at the same time, to provide reliable measurement indicators for future end-stage kidney disease patient treatment test results.

[0041] 2.The application can accurately evaluate the lung function, structure and function of respiratory muscles of patients, the examination process is comfortable, and the examination can be completed as long as there is no contraindication of MRI examination, the movement of respiratory muscles and lungs is objectively, comprehensively and visually displayed, the spatiotemporal changes of intercostal muscles, diaphragm and lung tissue respectively occurring with respiratory movement can be distinguished, the application is not only non-invasive and painless, but also further expands the applicable population, so that more patients benefit, and helps clinicians understand the influence of the functional state of respiratory muscles on the lung function of patients.

[0042] Of course, implementing any product of the application does not necessarily require achieving all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 A flowchart of the method according to the first embodiment of the application is shown in the figure;

[0044] Figure 2 An architecture diagram of the UNet segmentation model according to the first embodiment of the application is shown in the figure;

[0045] Figure 3 A lung C-C line and AP line diagram according to the first embodiment of the application is shown in the figure;

[0046] Figure 4 An A-P diameter change over time diagram according to the first embodiment of the application is shown in the figure;

[0047] Figure 5 A diaphragm movement area diagram according to the first embodiment of the application is shown in the figure;

[0048] Figure 6 A fitting diaphragm top lung bottom curve diagram according to the first embodiment of the application is shown in the figure;

[0049] Figure 7 A maximum thoracic movement area diagram according to the first embodiment of the application is shown in the figure;

[0050] Figure 8 A lung volume change and FVC significant correlation diagram when deep breathing according to the first embodiment of the application is shown in the figure;

[0051] Figure 9 A lung volume change and FEV1 significant correlation diagram when deep breathing according to the first embodiment of the application is shown in the figure;

[0052] Figure 10 A lung volume change and MVV significant correlation diagram when deep breathing according to the first embodiment of the application is shown in the figure;

[0053] Figure 11 A lung volume change and IC significant correlation diagram when deep breathing according to the first embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0054] With reference to the drawings of the embodiments of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, but not all the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of the present disclosure.

[0055] The present application breaks through the full lung scan of magnetic resonance imaging (MRI) and makes an MRI dynamic movie, which combines a computer deep learning method based on a convolutional neural network to evaluate the lung function and the structure and function of respiratory muscles. The examination process is comfortable, and the examination can be completed as long as there is no contraindication of MRI examination. The movement of respiratory muscles and lungs is objectively, comprehensively and visually displayed, and the spatiotemporal changes of intercostal muscles, diaphragm and lung tissue respectively occurring with respiratory movement can be distinguished. Not only is it non-invasive and painless, but also it further expands the applicable population, benefits more patients, and helps clinicians understand the influence of the functional state of respiratory muscles on the lung function of patients.

[0056] Deep learning is a complex multi-layer neural network structure that automatically learns the representation of data by converting input information into multiple layers of abstraction. It is the most advanced machine learning method at present. Computer-aided medical image analysis based on deep learning has been widely used in disease diagnosis, feature description, disease staging, treatment response evaluation, disease recurrence monitoring and guidance of interventional therapy, surgery and radiotherapy. The purpose of computer-aided diagnosis (CAD) is to help radiologists accurately and quickly analyze a large number of multi-modal images. In 1998, it was officially approved by FDA as a diagnostic reference for radiologists to diagnose breast molybdenum targets. In recent years, the research results in this field have steadily increased. CAD developers extract image features that can distinguish different states based on image processing and feature extraction techniques based on domain knowledge. Then use the extracted features as the input prediction variables of the classifier, and adjust the weights of the prediction model based on the statistical characteristics of a set of training samples, so as to estimate the probability that the image belongs to one of the states. Existing researches use deep learning methods to reliably identify the contours of diaphragm and lung tissue, and evaluate the changes of thoracic movement and lung area, as well as the length, movement and contraction force of diaphragm. However, full lung scan has not been realized in previous literatures, and only one layer can be scanned, which has limited value. The breakthrough in MRI full lung scan technology will realize more comprehensive evaluation of lung function and spatiotemporal changes of respiratory muscles under different breathing patterns. At the same time, it can also distinguish the influence of intercostal muscles and diaphragm on lung function.

[0057] Embodiment one:

[0058] Please refer toFigures 1-11 The present application provides a technical solution: a lung function and respiratory muscle function determination method, specifically comprising the following steps:

[0059] S1. Receive MR image data, pre-process the MR image data, and screen to obtain MR image data with high image quality to form an image data set;

[0060] S1.1 For the technical solution of the present embodiment, the image data is obtained by scanning on a 3.0T superconducting MR scanner (SIEMENS MAGNETOM Vida-XT-64 Numaris / X VA20A-04N2), and an 18-unit torso phased array coil is used as follows:

[0061] (S1.11) Under normal breathing (non-deep breathing) conditions, scanning is performed when the subject is tidal breathing, two-dimensional movie MRI scanning is performed on the chest, each subject is scanned along two axial planes (the upper edge of the aortic arch layer and the right atrial upper edge layer), one sagittal plane (right lung cross section along the clavicular midline) and one coronal plane (pulmonary apex layer), MRI movie recording is 19 seconds, the average time difference between each frame is 0.11 seconds, real-time movie images are obtained using balanced steady-state free precession sequence and compressed sensing imaging acceleration (total factor = 5); the sequence parameters are repetition time 109.9 ms, echo time 1.3 ms, excitation flip angle 46°, slice thickness 6 mm, field of view 380x340 mm 2 ;

[0062] (S1.12) Under deep breathing conditions, deep inspiration and breath holding, full lung scanning is performed again; deep expiration and breath holding, full lung scanning is performed again. The sequence parameters are repetition time 4.0 msec, double echo time 1.29 msec and 2.52 msec, gradient echo sequence length 48, excitation flip angle 9°, slice thickness 3.5 mm, field of view 360x270 mm 2 ;

[0063] S1.2 For the technical solution of the present embodiment, the MR image data is pre-processed as follows:

[0064] (S1.21) Convert all Dicom format image data to Nifti format using MRICron;

[0065] (S1.22) Evaluate the signal-to-noise ratio of the image using Slicer 5.1.0 software:

[0066] Because the lung contains a large amount of gas, which can cause loss of MR signal, the lung is not suitable as an ROI for evaluating signal-to-noise ratio, and subcutaneous muscle (chest muscle) tissue ROI is selected as the ROI for evaluating signal-to-noise ratio, the ROI is outlined in Slicer software, and the image enhancement mean and standard deviation of the ROI are extracted, the signal-to-noise ratio SNR is the ratio of the image enhancement mean Mean and the standard deviation SD of the ROI, and the calculation formula is: SNR = Mean / SD;

[0067] It should be noted that 30 healthy volunteers are included in this embodiment as an image signal-to-noise ratio evaluation group;

[0068] (S1.23) The Slicer 5.1.0 software is used to outline the lung profile in the image multiple times, for patients with pleural effusion, the pleural effusion part is included in the outlined ROI, the average Dice index of the outlined ROI is calculated, and the image with a high Dice index is selected as the standard image

[0069] S2. The image data set is divided into a training set, a validation set and a test set according to 5:2:1;

[0070] S3. A UNet segmentation model is constructed and trained using an image data set, the UNet segmentation model is based on a UNet network, an ADN module is introduced after each convolutional layer, and multiple residual modules are introduced into the network;

[0071] Specifically, in a Python 3.9 environment, a MONAI package (Version 0.8.1) is used to build a UNet segmentation model framework, the operating system is Win10 22H2, the GPU is NVIDIA 3090, the RandCropByPosNegLabeld and RandRotate90d functions in monai.transforms are used to expand the training image, a total of 25 subjects are included in the training image, 10 subjects are included in the validation set, and 5 subjects are included in the test set, each subject contains 143 sagittal images, 171 coronal images and 143 transverse images, mainly including sagittal movie images for segmentation training;

[0072] The UNet network starts from a convolutional layer that maintains the spatial size (128x128), and extracts features by changing the channel depth, an ADN module is followed after each convolutional layer, which specifically includes an instance normalization (InstanceNorm2d), a Dropout function and a PReLU activation function, wherein the instance normalization helps to keep the model stable during training, the Dropout function is to prevent overfitting, and the PReLU provides a nonlinear activation function;

[0073] The UNet network uses multiple residual units, each containing two convolution operations, which help to address the vanishing gradient problem that may occur during training of deep networks, while preserving important features of the input data;

[0074] The UNet network gradually reduces and restores the image size by alternating down-sampling and up-sampling operations, down-sampling is achieved by increasing the stride of the convolution layer, which helps the model to capture more extensive contextual information, up-sampling is usually achieved by transpose convolution (ConvTranspose2d), the purpose is to restore the original size of the image for detailed pixel-level prediction, in some residual units, the residual path is set to identity mapping, which means that this path only passes the input without any modification, this helps to maintain the integrity of the features and prevent information loss during training;

[0075] S4. Adopt the trained UNet segmentation model to process the original MR image, and get the nifti format lung Label file;

[0076] Specifically, the UNet network uses residual units Adam optimizer, the learning rate is 0.01, the network schematic diagram is shown in Figure 1 , the output probability Figure Two is binarized according to the probability threshold 0.4, the FillHoles function is used to fill the internal gap of ROI, and the largest continuous ROI component is retained, the automatically generated Label file is saved as Nifti format, the final calculation of 2D movie sequence DICE coefficient in Testing group is 0.984, and the average DICE coefficient of whole lung cross-sectional sequence is 0.937;

[0077] S5. According to the coordinate information in the Label file, the maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter1, the minimum C-C diameter1, the maximum C-C diameter2, the minimum C-C diameter2, the maximum A-P diamter, the minimum A-P diamter, the diaphragm movement area and the thoracic movement area, and the lung volume are calculated, the above parameters have strong correlation with the determination results of the lung function instrument, which are used for indirect evaluation of lung function and respiratory muscle function;

[0078] (S5.1) Read the generated lung Label file in nifti format with the nibabel package of Python, find the maximum Y coordinate point in the Label, which is the lung apex, record the Y coordinate minimum point in the Label from front to back according to the X coordinate, record the coordinate array of each point of the diaphragm top lung bottom curve, find the point with the maximum Y value in the diaphragm top lung bottom curve coordinate array, which is the diaphragm top lung bottom highest point, find the minimum X coordinate and the maximum X coordinate with the same Y value according to the diaphragm top lung bottom highest point horizontal direction, the line connecting the two points is the diaphragm top lung bottom highest point horizontal tangent AP Line1 Figure 3 (red line), the difference between the X coordinates of the two points multiplied by the horizontal length of the voxel is the A-P diameter;

[0079] (S5.2) Find the minimum and maximum points of the diaphragm top lung bottom curve coordinate array, the horizontal distance between the two points is A-P Line2 Figure 3 (deep blue line), the difference between the X coordinates of the two points multiplied by the horizontal length of the voxel is the A-P diameter2;

[0080] (S5.3) Draw the perpendicular C-C line Figure 3 (light blue line) from the lung apex to the diaphragm top lung bottom highest point horizontal tangent, the difference between the Y coordinates of the lung apex to the tangent multiplied by the vertical length of the voxel is the C-C diameter1;

[0081] (S5.4) Calculate the C-C diameter1 of each lung Label in the movie sequence, record it into the C-C diameter array, observe the fluctuation of C-C diameter with time Figure 4 ), the horizontal coordinate is the frame number, a total of 143 frames, the peak is the end of inspiration, the trough is the end of expiration, find the maximum and minimum values of the array, which are the maximum abdominal deep inspiration phase and the maximum abdominal deep expiration phase respectively, calculate the number of all voxels in the maximum abdominal deep inspiration phase Label located below the diaphragm top lung bottom curve of the maximum abdominal deep expiration phase Label, the obtained voxel number multiplied by the voxel area is the lung bottom-diaphragm muscle movement area Figure 5 , in the figure, the blue-green area is the maximum abdominal deep expiration phase Label after superimposing the maximum abdominal deep inspiration phase Label and the maximum abdominal deep expiration phase Label, the yellow area below the diaphragm top lung bottom curve of the maximum abdominal deep expiration phase Label is the maximum diaphragm muscle movement area;

[0082] (S5.5) Use the polynomial fitting method to fit the diaphragm top lung bottom curve of the maximum abdominal deep inspiration phase Label and the minimum abdominal deep inspiration phase Label, and use the integral method to calculate the diaphragm top lung bottom length of the two phases, which is the maximum diaphragm length and the minimum diaphragm length, as Figure 6As shown, the abscissa is the X coordinate in the original graph, and the ordinate is the inverted Y coordinate (MaxY-Y). The blue line is the original curve, and the red line is the fitted curve. The diaphragm-pulmonary base curve length of the maximum inspiratory phase and the maximum expiratory phase is calculated, i.e., the maximum pulmonary base diaphragm length and the minimum pulmonary base diaphragm length, which is an inverted image;

[0083] (S5.6) Calculate the A-P diameter of each lung Label in the movie sequence, record it into the A-Pdiameter array, observe the fluctuation of the A-P diameter with time, find the maximum value and the minimum value of the array, which are the maximum chest deep inspiratory phase and the maximum chest deep expiratory phase, respectively. Record the coordinates of the lung front boundary curve according to the minimum value of the X coordinate of each row of voxels to obtain a coordinate array. Calculate the number of voxels whose coordinates in the maximum chest deep inspiratory phase Label are above the diaphragm-pulmonary base curve of the maximum chest deep expiratory phase Label and in front of the lung front boundary curve. Multiply the obtained voxel number by the voxel area to obtain the thoracic movement area, such as Figure 7 As shown, the maximum chest inspiratory phase Label and the maximum chest expiratory phase Label are found out by using the C-Cdiameter array. After superimposing the two phase images, the blue-green area is the maximum expiratory phase Label, and the yellow area above the diaphragm-pulmonary base curve of the maximum expiratory phase Label and in front of the maximum expiratory phase Label is the maximum thoracic movement area.

[0084] (S5.7) Calculate the total number of voxels with a value of 0 by using the segmented full-lung cross-sectional Label file, and multiply it by the volume of a single voxel to obtain the full-lung volume.

[0085] The maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter1, the minimum C-C diameter1, the maximum C-C diameter2, the minimum C-C diameter2, the maximum A-P diameter, the minimum A-P diameter, the diaphragm movement area, the thoracic movement area, and the lung volume calculated automatically are accurately evaluated, and the corresponding parameter MAE (Mean Absolute Error Loss) is calculated.

[0086] Next, the technical solutions of the present application will be further described in combination with specific embodiments:

[0087] 1. Screening of the study population

[0088] Screening 30 cases of healthy volunteers aged 18 years≤age≤75 years, all the enrolled personnel were fully informed of the project before enrollment and signed the informed consent, the inclusion criteria: 1) age 18-75 years; 2) voluntarily signed the informed consent, exclusion criteria: 1) mechanical ventilation history; 2) history of cardiovascular and cerebrovascular diseases; 3) FEV1 < 70% predicted value, and / or using drugs to treat lung disease: such as β2 receptor agonists, M receptor blockers, glucocorticoids, leukotriene receptor modulators, antihistamines; 4) pulmonary fibrosis; 5) patients receiving immunosuppressive therapy; 6) patients with clear diaphragmatic paralysis; 7) history of malignant tumor in the past 5 years; 8) liver failure: Child-Pugh B, C; 9) pregnancy; 10) not suitable for chest MRI after exogenous implantation; 11) mental disorders, cognitive dysfunction and other conditions that cannot cooperate with the examination; 12) participated in other clinical studies in the past 3 months.

[0089] 2, prospectively collect the required clinical data

[0090] Collect baseline data of patients, including: gender, age, height, weight, disease history, medication history, smoking history, pulmonary function within 1 month before chest MRI examination;

[0091] 3, chest MR examination

[0092] All scans were performed on a 3.0T superconducting MR scanner (SIEMENS MAGNETOM Vida-XT-64Numaris / XVA20A-04N2), using an 18-element torso phased array coil;

[0093] ①, in the case of normal breathing (non-deep breathing), scanning was performed when the subject was tidal breathing, two-dimensional movie MRI scanning was performed on the chest, each subject was scanned along two axial planes (the upper edge of the aortic arch and the upper edge of the right atrium), one sagittal plane (right lung cross section along the midclavicular line) and one coronal plane (pulmonary apex), MRI movie recording 19 seconds, the average time difference between each frame was 0.11 seconds, real-time movie images were obtained using balanced steady-state free precession sequence and compressed sensing imaging acceleration (total factor = 5), sequence parameters: repetition time 109.9 ms, echo time 1.3 ms, excitation flip angle 46°, slice thickness 6 mm, field of view 380×340 mm 2 ;

[0094] ②, in the case of deep breathing, full lung scan was performed after deep inspiration and breath holding, and full lung scan was performed after deep expiration and breath holding, sequence parameters: repetition time 4.0 msec, double echo time 1.29 msec and 2.52 msec, gradient echo sequence length 48, excitation flip angle 9°, slice thickness 3.5 mm, field of view 360×270 mm2 ;

[0095] 4. Post-processing of MR data

[0096] Through the above steps S1-S5, the maximum diaphragm length, minimum diaphragm length, maximum C-Cdiameter1, minimum CC diameter1, maximum CC diameter2, minimum CC diameter2, maximum A-Pdiameter, minimum AP diameter, diaphragm movement area, thoracic movement area, and total lung volume (end of inspiration and end of expiration) are calculated.

[0097] 5. Statistical Analysis

[0098] ① Sample size estimation: Use PASS software to estimate the sample size, select Pearson correlation test, assume H1: ρ0≠ρ1, set the test power to 0.9, α=0.05, ρ0=0.0, ρ1=0.7, and calculate that the required sample size is at least 17 cases;

[0099] ② For continuous data, the distribution characteristics are described using the mean, standard deviation, median, 25th and 75th quantiles, maximum and minimum values. Based on the homogeneity of variance between groups, the t-test or t′ test is selected to analyze the differences between groups for continuous data that conform to a normal distribution. The Wilcoxon test or Mann-Whitney U test is used to analyze the differences between groups for continuous variables that are not normally distributed.

[0100] 6. Results

[0101] ① General information of 30 healthy subjects

[0102]

[0103] like Figures 8-11 As shown, the diagrams illustrate the significant correlation between lung volume changes during deep breathing and FVC, lung volume changes during deep breathing and FEV1, lung volume changes during deep breathing and MVV, and lung volume changes during deep inspiration and quiet expiration and IC. These results show a strong correlation between MRI data and pulmonary function tests, indicating that MR whole lung scanning can accurately assess a patient's pulmonary function, respiratory muscle structure, and function.

[0104] In summary, the technical scheme of the embodiment innovatively uses deep learning chest MRI dynamic image to comprehensively evaluate diaphragm structure and function indexes of hemodialysis patients, so as to quantitatively determine the diaphragm function condition of the population. The shortcomings that the lung function instrument cannot early identify the patients with lighter respiratory muscle lesions and the lung function is in the compensatory stage, the lung function instrument examination cannot indicate the position of respiratory muscle lesion, the lung function instrument examination cannot be completed by the patients with poor lung function, the diaphragm ultrasound examination is too dependent on the professional level of the operator and cannot comprehensively reflect the activity of the whole diaphragm are remedied. Meanwhile, reliable measurement indexes are provided for future end-stage kidney disease patient treatment test results.

[0105] Embodiment two:

[0106] The present application provides a lung function and respiratory muscle function measuring system for implementing the lung function and respiratory muscle function measuring method, and specifically comprises the following:

[0107] The preprocessing module is configured to receive MR image data, pre-process the MR image data, and screen MR image data with high image quality to form an image data set.

[0108] The division module is configured to divide the image data set into a training set, a validation set and a test set according to a ratio of 5:2:1.

[0109] The construction module is configured to construct a UNet segmentation model and train the UNet segmentation model using the image data set. The UNet segmentation model is based on a UNet network, and an ADN module is introduced after each convolution layer. Meanwhile, multiple residual modules are introduced into the network.

[0110] The output module is configured to process the original MR image using the trained UNet segmentation model to obtain a lung Label file in nifti format.

[0111] The calculation module is configured to calculate the maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter 1, the minimum C-C diameter 1, the maximum C-C diameter 2, the minimum C-C diameter 2, the maximum A-P diamter, the minimum A-P diamter, the diaphragm movement area, the thoracic movement area and the lung volume according to the coordinate information in the Label file. The mean absolute error of the corresponding parameters is calculated according to the above parameters, and the lung function and respiratory muscle function are determined according to the mean absolute error.

[0112] Specifically, the above-mentioned preprocessing module, division module, construction module, output module and calculation module can be embedded into a computer processing system. The computer calls the above-mentioned modules to complete the task of measuring lung function and respiratory muscle function according to the above-mentioned lung function and respiratory muscle function measuring method. The above-mentioned preprocessing module, division module, construction module, output module and calculation module can perform operations according to the specific steps given by the lung function and respiratory muscle function measuring method.

[0113] It should be noted that the division of each module of the above system is only a logical functional division. In actual implementation, all or part of the modules can be integrated into one physical entity, or can be physically separated. These modules can all be implemented in the form of software through a processing element, or can all be implemented in the form of hardware, or part of the modules can be implemented in the form of software through a processing element, and part of the modules can be implemented in the form of hardware. For example, the preprocessing module can be a separate processing element, or can be integrated into a chip of the above-mentioned device. In addition, the preprocessing module can also be stored in the form of program code in the memory of the above-mentioned device, and the function of the above-mentioned signal processing module can be called and executed by a processing element of the above-mentioned device. The implementation of other modules is similar. In addition, all or part of the modules can be integrated together, or can be implemented independently. The processing element described herein can be an integrated circuit with signal processing capability. In the implementation process, each step of the above-mentioned method or each of the above-mentioned modules can be completed by an integrated logic circuit of hardware or an instruction in the form of software in the processing element.

[0114] For example, the above-mentioned modules can be one or more integrated circuits configured to implement the above-mentioned method, such as one or more application specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs), etc. For another example, when a certain module above is implemented in the form of a processing element scheduling program code, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, the modules can be integrated together to implement in the form of a system on a chip (SOC).

[0115] Embodiment Three

[0116] The application provides a terminal device, comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the memory stores a computer program capable of running on the processor, and the processor loads and executes the computer program to adopt the lung function and respiratory muscle function determination method in Embodiment I.

[0117] It should be noted that the terminal device can adopt a computer device such as a desktop computer, a notebook computer or a cloud server, and the terminal device comprises but is not limited to a processor and a memory, for example, the terminal device can further comprise an input / output device, a network access device and a bus, etc.

[0118] Further, the processor can adopt a central processing unit (CPU), of course, according to the actual use condition, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), ready programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., the general-purpose processor can adopt a microprocessor or any conventional processor, etc., and the application does not limit this.

[0119] It should be noted that in this document, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device.

[0120] For those skilled in the art, the specific meanings of the above terms in the application can be understood according to the specific circumstances. When an element is referred to as "assembled", "mounted", "fixed" or "disposed" on another element, it can be directly on the other element or there can be a middle element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there can be a middle element. The terms "vertical", "horizontal", "up", "down", "left", "right" and similar expressions used herein are only for illustrative purposes and are not the only implementation.

[0121] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous changes, modifications, substitutions, and alterations can be made thereto without departing from the spirit and scope of the application as defined in the appended claims and their equivalents.

[0122] In the description of the specification, reference to "one embodiment", "an example", "a specific example" or the like means that a particular feature, structure, material or characteristic is included in at least one embodiment or example of the disclosure. The appearances of the above expressions in various places in the specification do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics can be combined in any suitable manner in one or more embodiments or examples.

Claims

1. A method of measuring lung function, respiratory muscle function, characterized by, Specifically comprising the following steps: Receiving MR image data, preprocessing the MR image data, and screening to obtain MR image data with high image quality to form an image data set; Dividing the image data set into a training set, a validation set, and a test set according to a ratio of 5:2:1; Constructing a UNet segmentation model and training the UNet segmentation model using the image data set, wherein the UNet segmentation model is based on a UNet network, an ADN module is introduced after each convolutional layer, and multiple residual modules are introduced into the network; Using the trained UNet segmentation model to process the original MR image to obtain a nifti format lung Label file; According to the coordinate information in the Label file, the maximum diaphragm length, the minimum diaphragm length, the maximum C-C diameter1, the minimum C-C diameter1, the maximum C-C diameter2, the minimum C-C diameter2, the maximum A-P diamter, the minimum A-P diamter, the diaphragm movement area, the thoracic movement area, and the lung volume are calculated, and the specific steps are as follows: (1) using the nibabel package of Python to read the generated nifti format lung Label file, finding the maximum Y coordinate point in the Label, which is the lung vertex, recording the coordinate array of each point of the diaphragm top lung bottom curve from front to back according to the X coordinate, finding the point with the maximum Y value in the diaphragm top lung bottom curve coordinate array, which is the diaphragm top lung bottom highest point, finding the minimum X coordinate and the maximum X coordinate with the same Y value in the horizontal direction of the diaphragm top lung bottom highest point, and the line connecting the two points is the diaphragm top lung bottom highest point horizontal tangent AP Line1, and the difference between the two X coordinates multiplied by the horizontal length of the voxel is the A-P diameter; (2) finding the minimum point and the maximum point in the diaphragm top lung bottom curve coordinate array, and the horizontal distance between the two points is A-P Line2, and the difference between the two X coordinates multiplied by the horizontal length of the voxel is A-P diameter2; (3) drawing the perpendicular line C-C line of the horizontal tangent from the lung vertex to the diaphragm top lung bottom highest point, and the difference between the Y coordinates of the lung vertex and the tangent is C-C diameter1; (4) calculating the C-C diameter1 of each lung Label in the image data set, recording it into the C-C diameter array, observing the fluctuation of C-C diameter with time, finding the maximum value and the minimum value of the array, which are the maximum abdominal deep inspiration phase and the maximum abdominal deep expiration phase respectively, then calculating the number of all voxels located below the diaphragm top lung bottom curve in the maximum abdominal deep inspiration phase Label, and multiplying the obtained voxel number by the voxel area to obtain the lung bottom-diaphragm movement area; (5) fitting the diaphragm top lung bottom curve of the maximum abdominal deep inspiration phase Label and the minimum abdominal deep inspiration phase Label using a polynomial fitting method, and calculating the diaphragm top lung bottom curve length of the maximum inspiration phase and the maximum expiration phase using the integral method, which is the maximum lung bottom diaphragm length and the minimum lung bottom diaphragm length: (6) Calculate the A-P diameter of each lung Label in the image dataset, record it into the A-P diameter array, observe the fluctuation of A-P diameter with time, find the maximum and minimum values of the array, which are the maximum chest deep inspiration phase and the maximum chest deep expiration phase respectively, record the coordinates of the lung anterior boundary curve according to the minimum point of the X coordinate of each row of voxels, obtain the coordinate array, calculate the number of voxels in the maximum chest deep inspiration phase Label which are above the diaphragm top and lung bottom curve and in front of the lung anterior boundary curve in the maximum chest deep expiration phase Label, multiply the obtained voxel number by the voxel area to obtain the thoracic cavity movement area; (7) Calculate the total number of voxels with a value of 0 using the segmented full lung cross-sectional Label file, and multiply the volume of a single voxel to obtain the total lung volume; The ADN module includes an instance normalization function, a Dropout function and a PReLU activation function.

2. The method of claim 1, wherein: The acquisition of the MR image data is performed by a 3.0T superconducting MR scanner, and the specific process is as follows: (1) Under normal breathing, the subject is scanned during tidal breathing, and two-dimensional movie MRI scanning is performed on the chest, each subject is scanned along two axial planes, i.e., the upper edge of the aortic arch plane and the upper edge of the right atrium plane, one sagittal plane, i.e., the right lung cross section along the midline of the clavicle, and one coronal plane, i.e., the lung apex plane; (2) Under deep breathing, full lung scanning is performed during deep inspiration and breath holding, and full lung scanning is performed during deep expiration and breath holding.

3. The method of claim 1, wherein: The pre-processing of the MR image data is performed as follows: (1) Convert all Dicom format image data to Nifti format using MRICron; (2) Evaluate the signal-to-noise ratio of the image using Slicer 5.1.0 software: Select the subcutaneous muscle (chest muscle) tissue ROI as the ROI for signal-to-noise ratio evaluation, outline the ROI in the Slicer software, and extract the image enhancement mean and standard deviation of the ROI, the signal-to-noise ratio SNR is the ratio of the image enhancement mean Mean and the standard deviation SD, the calculation formula is: SNR=Mean / SD; (3) Use Slicer 5.1.0 software to outline the lung profile of the full lung cross section in the image multiple times, for patients with pleural effusion, include the pleural effusion part in the outlined ROI, calculate the average Dice index of the outlined ROI, and select the image with high Dice index as the standard image.

4. The method of claim 3, wherein: After dividing the image dataset into a training set, a validation set and a test set, the RandCropByPosNegLabeld and RandRotate90d functions in monai.transforms are used to augment the training images.

5. The lung function, respiratory muscle function determination method according to claim 4, characterized in that: The residual module is set to multiple, each residual module includes two convolution operations, and part of the residual module residual path is set to identity mapping.

6. The method of claim 5, wherein: The up-sampling in the UNet segmentation model is realized by transposed convolution, and the down-sampling is realized by increasing the convolution layer step.

7. A lung function, respiratory muscle function measuring system for implementing the lung function, respiratory muscle function measuring method according to any one of claims 1 to 6, characterized by, Specifically, the following steps are included: The pre-processing module is configured to receive MR image data, pre-process the MR image data, and screen MR image data with high image quality to form an image data set; The division module is configured to divide the image data set into a training set, a verification set, and a test set according to a ratio of 5:2:1; The construction module is configured to construct a UNet segmentation model and train the UNet segmentation model using the image data set, wherein the UNet segmentation model is based on a UNet network, an ADN module is introduced after each convolutional layer, and a plurality of residual modules are introduced into the network; The output module is configured to process original MR images using the trained UNet segmentation model to obtain a nifti format lung Label file; The calculation module is configured to calculate a maximum diaphragm length, a minimum diaphragm length, a maximum C-C diameter 1, a minimum C-C diameter 1, a maximum C-C diameter 2, a minimum C-C diameter 2, a maximum A-P diamter, a minimum A-P diamter, a diaphragm movement area, a thoracic movement area, and a lung volume according to coordinate information in the Label file.

8. A terminal device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, The memory stores a computer program capable of running on the processor, and the processor loads and executes the computer program, and adopts the lung function and respiratory muscle function determination method in any one of claims 1 to 6.

Images