An automatic cardiac magnetic resonance imaging segmentation method based on expanded residual network

Abstract

The application discloses a kind of full-automatic cardiac magnetic resonance imaging segmentation methods based on expansion residual network.The method comprises the following steps: acquiring a cardiac magnetic resonance image;The cardiac magnetic resonance image is input into a trained segmentation network, and the right ventricular region, myocardial region and left ventricular region are segmented out, wherein the segmentation network is constructed based on residual network U-Net, and the bottleneck layer of the residual network is combined with the encoding path and the decoding path using the expansion convolution block with a set expansion rate.The application can accurately segment the right ventricle, left ventricle, myocardium and other regions from the cardiac magnetic resonance image, realize the full-automatic segmentation of the cardiac image, and improve the performance of the cardiac region image segmentation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biomedical engineering, and more particularly to a full-automatic cardiac magnetic resonance imaging segmentation method based on an expanded residual network. BACKGROUND

[0002] Cardiac diseases seriously threaten human life. In order to effectively treat and prevent such diseases, accurate calculation, modeling and analysis of the entire heart structure are crucial for research and application in the medical field. However, during the acquisition of CMRI (cardiac magnetic resonance imaging), the continuous beating of the heart makes it more difficult to obtain clear images, especially for patients with cardiovascular diseases, who are more likely to experience arrhythmia, difficulty in holding their breath, etc. This results in the possibility that the images of the MRI (magnetic resonance imaging) scanner may contain various image artifacts, making it difficult to assess image quality. If the image data is not segmented correctly, clinicians may draw incorrect conclusions from the image data. Existing manual segmentation of images is not only time-consuming, but also difficult to guarantee accuracy. Therefore, it is necessary to realize automatic segmentation of the heart region to solve practical problems in the field of cardiac medicine.

[0003] Cardiac image segmentation refers to dividing a cardiac image into multiple anatomically meaningful regions, based on which quantitative measures such as myocardial mass, wall thickness, left ventricular (LV) and right ventricular (RV) volume can be extracted. Therefore, it is particularly important to design an accurate full-automatic cardiac segmentation algorithm. In recent years, deep convolutional neural networks (DCNNs) have been shown to outperform traditional computer vision methods in segmenting left and right ventricles and myocardium. For example, the U-Net architecture is task-independent and has been applied to various biomedical segmentation tasks, and with only minor or substantial modifications, U-Net is the backbone model of most effective ventricular segmentation algorithms.

[0004] In the prior art, patent application CN202210321078.8 provides a CT image heart segmentation method and system based on artificial intelligence semantic segmentation. This technology reduces the influence of noise by optimizing the class probability, and realizes accurate image segmentation. However, this scheme does not involve the segmentation of the left ventricle, right ventricle, and myocardium in the heart, and cannot obtain independent heart tissue images.

[0005] Patent application CN202110391121.3 describes a cardiac MRI-based cardiac segmentation model and pathological classification model training, cardiac segmentation, and pathological classification method and device. This technology can greatly suppress background interference and promote rapid convergence of neural network training, but does not propose an improved method for image segmentation accuracy and robustness.

[0006] Through analysis, the prior art lacks research on the bottleneck layer of the U-Net, and due to the fact that the background region in the image is much larger than the mask, the pixel degradation, loss of time and space information caused by the deepening of the network layers leads to insufficient extraction ability of the network for sparse features of the image. SUMMARY

[0007] The purpose of the present application is to overcome the defects of the prior art described above, and to provide a full-automatic cardiac magnetic resonance imaging segmentation method based on a dilated residual network. The method comprises the following steps:

[0008] obtaining a cardiac magnetic resonance image;

[0009] inputting the cardiac magnetic resonance image into a trained segmentation network to segment out a right ventricular region, a myocardial region and a left ventricular region;

[0010] wherein the segmentation network is constructed based on a residual network U-Net, and a dilated convolution block with a set dilated rate is used to combine the encoding path and the decoding path of the bottleneck layer of the residual network.

[0011] Compared with the prior art, the full-automatic cardiac MRI (magnetic resonance imaging) segmentation method based on a dilated residual network can accurately segment the right ventricle, left ventricle, myocardium and other regions from the cardiac MRI image, realize full-automatic segmentation of the cardiac image, and improve the performance of the cardiac region image segmentation.

[0012] Other features and advantages of the present application will become apparent from the following detailed description of exemplary embodiments thereof, taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0013] The accompanying drawings incorporated in and forming a part of the specification, illustrate embodiments of the present application and, together with the description, serve to explain the principles of the application.

[0014] Figure 1 is a flowchart of a full-automatic cardiac magnetic resonance imaging segmentation method based on a dilated residual network according to an embodiment of the present application;

[0015] Figure 2 is a process schematic diagram from original magnetic resonance image data to image segmentation according to an embodiment of the present application;

[0016] Figure 3 is an automatic image segmentation architecture diagram based on U-Net according to an embodiment of the present application;

[0017] Figure 4 is an architecture schematic diagram of a dilated residual block according to an embodiment of the present application;

[0018] Figure 5 is a schematic illustration of image segmentation results for ACDC test dataset according to one embodiment of the present application;

[0019] In the drawings, Conv - convolution; Norm - normalization; Maxpool - max pooling; UpConv - up-convolution; Deconvolution - de-convolution; Skip-connection - skip connection; Pixel-wise addition - pixel-wise addition; Kernel - kernel; Concatenation - concatenation; Stride - stride; End Systole - end systole; End Diastole - end diastole. DETAILED DESCRIPTION

[0020] Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. Note that the relative arrangement, numerical expressions, and numerical values of components and steps set forth in these embodiments are not limiting to the scope of the present application unless otherwise specifically stated.

[0021] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way limiting to the scope of the application or its applications or uses.

[0022] Techniques, methods, and apparatus known to those of ordinary skill in the relevant art can not be discussed in detail herein, but should be considered as part of the specification, where appropriate.

[0023] In all of the examples shown and discussed herein, any specific values should be interpreted as illustrative only and not as a limitation. Thus, other examples of exemplary embodiments can have different values.

[0024] Note that like reference numerals and letters in the various drawings herein represent similar items unless otherwise specifically stated, and thus, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0025] The present application develops a fully automatic segmentation method for segmenting right ventricle (RV), myocardium (MYO) and left ventricle (LV) by combining short-axis CMRI (magnetic resonance imaging movie) sequence images. The method significantly increases spatial and temporal information and maintains positioning accuracy by capturing multi-resolution features in U-Net through dilated convolution residual network (DRN).

[0026] In combination Figure 1 and Figure 2 As shown, the provided fully automatic cardiac magnetic resonance imaging segmentation method based on dilated residual network includes the following steps:

[0027] In step S110, the data set is preprocessed to obtain training samples.

[0028] Taking the use of magnetic resonance cine imaging as an example, the size of the three-dimensional image is LxWxH, where L is the length of the image sequence, W is the width of the image, and H is the length of the image. In the data set, the label value of the image is set to four labels in a mapping manner, which are: black background = 0, RV = 1, MYO = 2, and LV = 3.

[0029] It is considered that there is a significant difference between the display space size HxW of the magnetic resonance cine image and the range of intensity distribution. In an embodiment, the training samples are obtained through a data preprocessing process, specifically taking the ACDC (Adverse Conditions Dataset with Correspondences) image as an example. First, the input image is resampled. The ACDC data set has a problem of voxel spacing. Since the convolutional neural network cannot interpret the voxel spacing, all images are resampled to the same voxel spacing of 1.52x1.52x6.35mm.

[0030] The data preprocessing process is to consider that the voxel spacing directly affects the overall voxel size of the image, and also affects the amount of context information extracted from the image patch by the convolutional neural network. In addition, if the voxel spacing is increased significantly, the image size will be reduced to the extent that the details are lost, so it is necessary to ensure a trade-off between the amount of context information contained in the network patch size and the amount of details preserved in the image data to obtain the best performance.

[0031] In an embodiment, for the training data, all images are resampled to a median value of 256x256 pixels. Then, the ACDC data set of magnetic resonance images is obtained using multi-layer magnetic resonance cine images. For example, the 2D-MRI (magnetic resonance imaging) slices of each patient and their related annotations are extracted. And perform normalization slice by slice for each time frame.

[0032] In step S120, the training samples are expanded by data augmentation, and a training set is constructed.

[0033] Due to the limited training data, the model cannot learn the expected invariance and robustness features, resulting in overfitting. Therefore, a variety of data augmentation techniques can be applied to the training data to expand the number of samples. For example, basic image transformation techniques are used, including random rotation, random elastic deformation, scaling, flipping, and gamma correction. When applied to the original training image, this data augmentation technique can effectively generate multiple views of the same image. By using multiple data augmentation methods to expand the training samples, the problem of overfitting and class imbalance can be solved.

[0034] Step S130, constructing a segmentation network based on an expanded residual network.

[0035] In this paper, the U-net network is taken as an example to illustrate the heart segmentation network, as shown in Figure 3 and Figure 4 , wherein Figure 4 corresponds to the expanded residual block architecture of Figure 3 . From the input image to the final output, the segmentation network follows the overall architecture of the encoder-decoder throughout the segmentation process. For example, a 5-block encoding block is used to construct the shrinkage path; each block is composed of 2 convolutional layers with 3x3 kernels and 2x2 max-pooling operations with a step size of 2. Initially, 32 convolutional kernels are selected. After each max-pooling operation, the convolutional kernel will be increased, resulting in 320 convolutional kernels in the bottleneck layer of the U-Net. Similarly, the spatial dimension of the feature map is reduced by 2 times through the downsampling operation. The linear rectifier unit (ReLU) is replaced by the leaky linear rectifier, and instance regularization is used instead of normalization (BN).

[0036] The expanded residual network (DRN) combines the encoding and decoding paths at the U-Net bottleneck layer, which captures global context and restores spatial and temporal information without affecting the resolution of the segmentation map. In addition, the expanded residual network can effectively adjust the depth of the convolutional layer without reducing the performance of the network. For example, by using dilated convolution with different dilation rates (d=1, 3 and 5), the receptive field in the expanded residual network block is expanded. Then, the previously generated features are concatenated with the current features through a residual connection. After each 3x3 convolution in the DRN (expanded residual network) block, a dropout operation with a forgetting rate of 0.5 is performed to prevent overfitting. Therefore, the expanded residual network captures contextual image information, high spatial resolution, and multi-texture features. The process of the decoding path is similar to that of the encoding path, however, the order of operations is reversed. The U-Net architecture provides the advantage of reusing the encoded feature maps from the encoding blocks to their corresponding levels, where the spatial dimensions match. This can be achieved through channel-specific concatenation. A 1x1 kernel projection operation is used at the last level of the decoding path to align the output channel dimensions with the classified classes (left ventricle, myocardium, and right ventricle). Finally, all expanded path outputs are aggregated through upsampling and pixel-wise addition to enhance the training response.

[0037] Generally, natural images contain many objects, whose identities and relative positions are important for understanding the scene. However, segmentation becomes more difficult when the target object is not spatially salient, for example, when the target object is small compared to the background. If the features of the target object are lost in the down-sampling process, it is not easy to recover them in the training. But if the high (large amount) of spatial and temporal information is maintained throughout the network and the output features densely cover the input features, backpropagation can learn important features from smaller and less salient objects. Therefore, the present application adopts a dilated convolutional network to predict small and dense image features by increasing the receptive field to extract more spatial information. The discrete dilated convolution is as follows:

[0038]

[0039] wherein, is the input and output discrete function, k is a discrete kernel with size (2d+1) 2 , and l is a dilated convolution, which satisfies s+lt=p in the summation process, s represents the dilation stride, l represents the scaling factor, p represents the receptive field, and t represents an integer sequence, i.e., t=1, 2, 3...n.

[0040] A dilated residual network can better expand the receptive field to achieve a promising result and avoid image information loss at the bottleneck of the U-Net. The dilated convolution introduces a new parameter called "dilation rate" to the convolutional layer, which defines the spacing of each value when the convolution kernel processes the data, and expands the receptive field by adding holes. The dilated convolution layer is based on the regular convolution with a dilation factor (d=1, 3 and 5). For example, a 1x1 kernel is selected for the ordinary convolution layer, and a 3x3 kernel is selected for the dilated convolution.

[0041]

[0042] wherein, y ij represents a dilated convolution with input x ij , which is a convolution kernel with length M and width N, m and n are input variables of the dilated convolution. w(i, j) is the corresponding weight value, i represents the image length index, j represents the image width index, and d represents the dilation rate.

[0043] In step S140, the segmentation network is trained using the set loss function.

[0044] The purpose of segmentation is to detect the target object and draw a contour around it. The automatic segmentation contour Cp (predicted) is compared with the corresponding labeled image to measure the accuracy of the proposed method. In this paper, the pixels enclosed by the contour are referred to as A p and the pixels not enclosed by the contour are referred to as Ag .

[0045] Various loss functions can be used in training segmentation networks. For example, dice similarity coefficient, Hausdorff distance, or other loss function types.

[0046] For example, the Dice Similarity Coefficient (DSC) is the ratio between the predicted profile and the ground truth profile, representing the DSC score, typically expressed as a percentage between 0 and 1. A high Dice value indicates a good match.

[0047]

[0048] Among them, A p A represents the pixels enclosed by the predicted contour. g Represents the pixels enclosed by the actual outline.

[0049] Hausdorff distance (HD) is a symmetrical distance between the predicted and actual contours, providing spatial resolution for magnetic resonance imaging. A lower HD value indicates better segmentation matching performance.

[0050]

[0051] Among them, C p For the predicted automatic segmentation contour, C g For the corresponding ground truth contour, d(i,j) represents the distance between the ground truth and the predicted contour, where i represents the pixel value of the predicted contour and j represents the pixel value of the ground truth contour. Considering the significant class imbalance between the region of interest (ROI) and the background, different loss functions, including dice loss and weighted cross-entropy loss, were tested to address this issue.

[0052] In a preferred embodiment, a dual loss function comprising dice loss and cross-entropy loss is used to train the segmentation network. Specifically, the cross-entropy loss is defined as follows:

[0053]

[0054] Where C represents the total number of categories; c represents the category indicator, and W = (w1, w2, w3...w... n ) is a series of learnable weights, w n It is the weight matrix of the nth layer; p(Y) i |X i W) represents a predicted pixel X i The probability of misclassification of a pixel relative to the ground truth label; Y(c,x) represents the target label corresponding to input x; represents the activation function value of the predicted class c corresponding to the input x. For example, for the class represented by c, black background = 0, RV = 1, MYO = 2, LV = 3.

[0055] The training of the model was performed for 500 iterations, in each iteration of the training set, 250 images were randomly sampled from the dataset until all image data were traversed. To improve the generalization ability, patches were randomly cropped from the training images and the network was evaluated after each iteration on the validation set. For example, the dice loss multiclass variant of the following formula was used to train the segmentation network.

[0056]

[0057] where u and v are the one-hot encoded vector and the image segmentation label value corresponding to the class identifier of the activation function softmax output, i represents the image length index, k represents the image width index; c∈C is the class identifier, i.e. the left ventricle, right ventricle, myocardium and background of the heart; ε is a small constant. After each traversal, the learning rate lr is recalculated according to the following formula. Finally, the best model is selected to evaluate the test set to ensure that the verification of RV (right ventricle), MYO (myocardium) and LV (left ventricle) reaches the highest DSC (dice similarity coefficient). The network provides consistent and stable performance in all folds.

[0058]

[0059] where initial_learning_rate is the initial learning rate, currentepoch is the current number of iterations, and totalepoch is the total number of iterations.

[0060] Step S150, for the collected target magnetic resonance image, the right ventricle, myocardium and left ventricle and other regions are identified by using the trained segmentation network.

[0061] After the training of the segmentation network is completed, the optimized model parameters can be obtained, and then by using the trained segmentation network, the right ventricle, myocardium and left ventricle and other regions of interest, as well as the complete heart contour, can be accurately distinguished, and then based on these regions, quantitative metrics such as myocardial mass, left ventricular and right ventricular volume can be extracted.

[0062] To further verify the effect of the present application, experiments have been carried out on a plurality of patient cardiac magnetic resonance images. See Figure 5The schematic diagram of different slice positions of the heart is shown. The experimental results show that the application obtains higher segmentation accuracy and segmentation speed of the left ventricle, the right ventricle and the myocardium, obtains an overall dice similarity coefficient of 0.92±0.02 and an average Hausdorff distance of 8.06±0.05mm. And the application improves the speed of image segmentation, for example, it takes an average of 0.28 seconds to process a 2D magnetic resonance image. In addition, the network design of the application is used to predict a single magnetic resonance image to segment the ventricular region, and the automatic segmentation of the cardiac image is successfully realized.

[0063] In summary, compared with the prior art, the application has the following technical effects:

[0064] 1) The application introduces an expanded convolution residual network, enhances the performance of the U-Net bottleneck layer, and realizes the full-automatic precise segmentation of the cardiac MRI (magnetic resonance imaging) image, solves the limitation of the U-Net bottleneck layer, and significantly enhances the spatial and temporal information, while maintaining the spatial consistency and improving the accuracy.

[0065] 2) The application designs an extended residual network (DRN) block to replace the original bottleneck layer of the U-Net. And a variety of loss functions are used to better utilize the cardiac features to train the model during the segmentation of the cardiac image, and improve the accuracy.

[0066] 3) The application has higher computing speed and robustness, and can be applied to diversified cardiac CMRI (magnetic resonance movie imaging) data sets. For example, the data processed are magnetic resonance images of patients under two different magnetic intensities, and the processed data can obtain the complete cardiac contour, left and right ventricular and myocardial contour images of the patient at the same time.

[0067] The application can be a system, a method and / or a computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions loaded thereon for causing a processor to implement various aspects of the application.

[0068] Computer readable storage media can be tangible storage media which can retain and store instructions for use by an instruction execution device. Computer readable storage media can be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer readable storage media include the following: a portable computer diskette, 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 mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.

[0069] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network can comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0070] Computer readable program instructions for carrying out operations of the present application can be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate array (FPGA), or programmable logic array (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present application.

[0071] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0072] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0073] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0074] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0075] Embodiments of the present application have been described above, and the description is intended to be illustrative, and not restrictive, of the various embodiments of the present application. Many modifications and variations of the described embodiments of the present application are possible, given the benefit of the present disclosure, without departing from the scope and spirit of the described embodiments of the present application. The scope of the present application is defined by the appended claims.

Claims

1. An automatic cardiac magnetic resonance imaging segmentation method based on dilated residual network, comprising the following steps: obtaining a cardiac magnetic resonance image; inputting the cardiac magnetic resonance image into a trained segmentation network to segment a right ventricular region, a myocardial region and a left ventricular region; wherein the segmentation network is constructed based on a residual network U-Net, and a bottleneck layer of the residual network adopts a dilated convolution block with a set dilated rate to combine an encoding path and a decoding path; wherein training the segmentation network with a set loss function comprises: training the segmentation network with a cross-entropy loss function within a set range of iteration times, and extracting a set number of sample images from a training set for each iteration; randomly cropping slices from the training images, and evaluating the segmentation network after each iteration on a validation set, and training the segmentation network using a dice loss multi-class variant; recomputing a learning rate after each iteration; selecting a segmentation network meeting a set performance requirement as the trained segmentation network; wherein the dice loss is represented as: wherein, and is a one-hot encoding vector of the softmax output of the activation function and indicates the label class corresponding to the image segmentation label value, i denotes an image length index, k denotes an image width index, is a set constant, denotes the total number of label classes.

2. The method of claim 1, wherein, the segmentation network is trained according to the following steps: constructing a training set containing a plurality of sample data, each sample data being a magnetic resonance image with a labeled category for distinguishing the right ventricular region, the myocardial region and the left ventricular region; performing image enhancement on the training set to generate a plurality of views for the same magnetic resonance image, the image enhancement including one or more of random rotation, random elastic deformation, scaling, flipping and gamma correction; training the segmentation network with a set loss function using the image-enhanced training set to obtain optimized parameters.

3. The method of claim 1, wherein, the cross-entropy loss is represented as: wherein, denotes the total number of labeled classes; c denotes a labeled class indicator, is a series of weights to be learned, is a weight matrix of the nth layer, denotes a predicted pixel relatively to ground truth labeled pixels a probability of classification error, denotes a target label corresponding to an input ; denotes an activation function value of a predicted class c corresponding to an input .

4. The method of claim 1, wherein, updating the learning rate according to the following formula: wherein, is the initial learning rate, is the current iteration number, is the total number of iterations.

5. The method of claim 1, wherein, the last stage of the decoding path of the segmentation network uses a 1x1 kernel projection operation.

6. The method of claim 1, wherein, the dilated rate is set to d = 1, 3 or 5.

7. A computer readable storage medium having stored thereon a computer program, wherein, The computer program is executed by a processor to implement the steps of the method according to any one of claims 1 to 6.

8. A computer device comprising a memory and a processor, having stored on the memory a computer program capable of running on the processor, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

Images