Guide wire bending energy calculation method, interventional robot and storage medium

Abstract

The application provides a guide wire bending energy calculation method, an interventional robot and a storage medium. The method comprises the following steps: acquiring a guide wire tip image and inputting the guide wire tip image into a convolution module to obtain a first feature image; performing image segmentation on the first feature image according to an image segmentation algorithm to obtain a plurality of feature image blocks; mapping the plurality of feature image blocks to an embedding layer to obtain a second feature image; inputting the second feature image into a Transformer module to obtain a third feature image; performing image connection on the first feature image and the third feature image to obtain a target feature image, and inputting the target feature image into a segmentation head module for image segmentation processing to obtain a guide wire segmentation image; and obtaining a guide wire bending energy calculation result according to imaging state information of the guide wire segmentation image and the guide wire tip image. The application calculates the guide wire bending energy in real time, and replaces the doctor in the operation in the radiation environment through the interventional robot, so that the operation stability of the doctor is ensured and the risk of the operation is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of guidewire intervention control, in particular to a guidewire bending energy calculation method, an intervention robot and a storage medium. BACKGROUND

[0002] Percutaneous coronary intervention surgery is one of the most effective ways to treat coronary artery disease due to its small trauma and high precision controllability. A guidewire is a main interventional instrument in the surgery process. The guidewire is used to reach a lesion area and establish an interventional channel under the control of a physician by passing through small blood vessel branches. In the process of entering the blood vessel, if an obstruction or a branch is encountered, the guidewire will be bent and deformed due to the resistance of the blood vessel wall, and elastic potential energy, i.e., guidewire bending energy, will be accumulated in the structure of the guidewire. In the case of excessive guidewire bending energy, the guidewire may puncture the blood vessel wall, causing a risk of surgery.

[0003] In the existing percutaneous coronary intervention surgery process, an interventional physician needs to wear a lead suit and stand to control the movement of the guidewire in the blood vessel by observing an X-ray image screen outside the natural line of sight. Based on the original X-ray image, the interventional physician can only observe whether the guidewire encounters an obstruction or a branch in the process of entering the blood vessel, and cannot determine the bending energy of the guidewire. There is a certain risk of surgery. In addition, due to the long surgery time and the heavy lead suit, the interventional physician may suffer from muscle fatigue, central nervous fatigue and mental fatigue, thereby affecting the decision-making and operation stability of the interventional physician during the surgery. SUMMARY

[0004] The embodiments of the present application provide a guidewire bending energy calculation method, an intervention robot and a storage medium. The guidewire bending energy in the percutaneous coronary intervention surgery process is calculated in real time by introducing an intervention robot, and the intervention robot replaces the interventional physician to make corresponding surgical actions in a radiation environment, so as to ensure the decision-making and operation stability of the physician during the surgery and reduce the risk of percutaneous coronary intervention surgery.

[0005] In a first aspect, the embodiments of the present application provide a guidewire bending energy calculation method applied to an intervention robot, and the guidewire bending energy calculation method comprises the following steps.

[0006] obtaining a DSA image sequence, the DSA image sequence comprising a guidewire tip image of a target object in an intervention surgery process;

[0007] obtaining a pre-trained guidewire segmentation model, the guidewire segmentation model comprising an encoder, a decoder and a segmentation head module, the encoder comprising a convolution module, an embedding layer and a Transformer module, and the decoder comprising an up-sampling layer;

[0008] The guidewire tip image is input to the convolution module for convolution processing to obtain a first feature image;

[0009] The first feature image is subjected to image segmentation processing according to a preset image segmentation algorithm to obtain a plurality of feature image blocks;

[0010] The plurality of feature image blocks are mapped to the embedding layer to obtain a second feature image;

[0011] The second feature image is input to the Transformer module for image processing to obtain a third feature image;

[0012] The first feature image and the third feature image are subjected to image connection processing based on the up-sampling layer to obtain a target feature image;

[0013] The target feature image is input to the segmentation head module for image segmentation processing to obtain a guidewire segmentation image;

[0014] The guidewire segmentation image and imaging state information of the guidewire tip image are input to a preset guidewire bending energy calculation module to obtain a guidewire bending energy calculation result.

[0015] In some embodiments, the Transformer module includes a multi-head self-attention module and a multi-layer perception module, and the inputting of the second feature image to the Transformer module for image processing to obtain a third feature image includes:

[0016] The second feature image is subjected to normalization processing according to a preset normalization algorithm to obtain a first intermediate feature image;

[0017] The first intermediate feature image is input to the multi-head self-attention module for feature extraction to obtain a second intermediate feature image;

[0018] The second intermediate feature image is input to the multi-layer perception module for image processing to obtain the third feature image.

[0019] In some embodiments, the inputting of the first intermediate feature image to the multi-head self-attention module for feature extraction to obtain a second intermediate feature image is obtained according to the following formula:

[0020] Z' ζ =MSA(LN(Z ζ-1 ))+Z ζ-1 ;

[0021] wherein, Z ζ-1 is the second feature image, and Z ζZ' is the second intermediate feature image, LN is the normalization algorithm, and MSA is the multi-head self-attention module.

[0022] In some embodiments, the second intermediate feature image is input into the multi-layer perception module for image processing to obtain the third feature image, according to the following formula:

[0023] Z ζ =MLP(LN(Z ζ ))+Z' ζ ;

[0024] wherein Z ζ is the third feature image, and MLP is the multi-layer perception module.

[0025] In some embodiments, the first feature image blocks are mapped to the embedding layer to obtain a second feature image, according to the following formula:

[0026]

[0027] wherein Z0 is the second feature image, is each feature image block, N is the number of feature image blocks, and the expression of N is:

[0028] N=HW / p 2 ;

[0029] wherein HW is the resolution of the first feature image, p 2 is the size of each feature image block, E is the image block embedding projection, and E pos is the embedding position information of the feature image block.

[0030] In some embodiments, the guidewire tip image is obtained by an X-ray machine, the X-ray machine is in communication connection with the interventional robot, and after the guidewire segmentation image, imaging state information, and inter-frame prior information are input into a preset guidewire bending energy calculation module to obtain a guidewire bending energy calculation result, the method further comprises:

[0031] determining guidewire tip length information, guidewire length information, and imaging parameter information of the X-ray machine;

[0032] determining target guidewire curve information according to the guidewire bending energy calculation result, the guidewire tip length information, the guidewire length information, the imaging parameter information, and inter-frame prior information of the guidewire tip image;

[0033] when the target guidewire curve information represents that the guidewire tip corresponding to the guidewire tip image presents a ring structure, determining that the guidewire bending energy calculation result is substandard.

[0034] In some embodiments, after the guide wire segmentation image, imaging state information and inter-frame prior information are input into the preset guide wire bending energy calculation module to obtain a guide wire bending energy calculation result, the method further comprises:

[0035] When the guide wire bending energy value corresponding to the guide wire bending energy calculation result exceeds a preset energy threshold, it is determined that the guide wire bending energy calculation result is substandard.

[0036] In some embodiments, before the guide wire tip image is input into the convolution module for convolution processing to obtain a first feature image, the method further comprises:

[0037] According to a preset image preprocessing rule, the guide wire tip image is preprocessed to obtain a preprocessed guide wire tip image.

[0038] In a second aspect, the embodiments of the present application provide an interventional robot, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the guide wire bending energy calculation method according to the first aspect.

[0039] In a third aspect, the embodiments of the present application further provide a computer readable storage medium storing computer executable instructions for executing the guide wire bending energy calculation method according to the first aspect.

[0040] The embodiment of the present application provides a guide wire bending energy calculation method, an interventional robot and a storage medium, and the method comprises the following steps: acquiring a DSA image sequence, wherein the DSA image sequence comprises a guide wire tip image of a target object in an interventional operation process; acquiring a pre-trained guide wire segmentation model, wherein the guide wire segmentation model comprises an encoder, a decoder and a segmentation head module, the encoder comprises a convolution module, an embedding layer and a Transformer module, and the decoder comprises an up-sampling layer; inputting the guide wire tip image into the convolution module for convolution processing to obtain a first feature image; performing image segmentation processing on the first feature image according to a preset image segmentation algorithm to obtain a plurality of feature image blocks; mapping the plurality of feature image blocks to the embedding layer to obtain a second feature image; inputting the second feature image into the Transformer module for image processing to obtain a third feature image; performing image connection processing on the first feature image and the third feature image to obtain a target feature image; inputting the target feature image into the segmentation head module for image segmentation processing to obtain a guide wire segmentation image; and inputting the guide wire segmentation image and imaging state information of the guide wire tip image into a preset guide wire bending energy calculation module to obtain a guide wire bending energy calculation result. The embodiment of the present application introduces an interventional robot to calculate the guide wire bending energy in the percutaneous coronary intervention operation process in real time, and the interventional robot replaces the interventional physician to make corresponding operation actions in the radiation environment, so that the decision and operation stability of the physician in the operation are ensured, and the risk of the percutaneous coronary intervention operation is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 It is a step flow chart of the guide wire bending energy calculation method provided by the embodiment of the present application;

[0042] Figure 2 It is a step flow chart for obtaining the third feature image provided by another embodiment of the present application;

[0043] Figure 3 It is a step flow chart for judging whether the guide wire bending energy calculation result meets the standard provided by another embodiment of the present application;

[0044] Figure 4 It is a step flow chart for judging whether the guide wire bending energy calculation result meets the standard provided by another embodiment of the present application;

[0045] Figure 5 It is a step flow chart for performing image preprocessing on the guide wire tip image provided by another embodiment of the present application;

[0046] Figure 6 It is a hardware schematic diagram of the interventional robot provided by another embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not used to limit the present application.

[0048] It can be understood that, although the functional modules are divided in the device schematic diagram, and the logical order is shown in the flowchart, in some cases, the steps shown or described can be performed in a manner different from the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims or above-described drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence.

[0049] Percutaneous coronary intervention is one of the most effective ways to treat coronary artery disease due to its small trauma and high precision control. The guide wire is the main interventional instrument in the operation process. The function of the guide wire is to reach the lesion area through the small blood vessel branches under the control of the physician and establish an interventional channel. In the process of entering the blood vessel, if the guide wire encounters an obstruction or branch, the guide wire will bend and deform under the resistance of the blood vessel wall, accumulating elastic potential energy in its structure, i.e. guide wire bending energy. In the case of excessive guide wire bending energy, the guide wire may puncture the blood vessel wall, causing surgical risks.

[0050] In the existing percutaneous coronary intervention process, the interventional physician needs to wear a lead suit and stand to control the movement of the guide wire in the blood vessel by looking at the X-ray image screen outside the natural line of sight. Based on the original X-ray image, the interventional physician can only observe whether the guide wire encounters an obstruction or branch in the process of entering the blood vessel, and cannot determine the bending energy of the guide wire, which poses certain surgical risks. In addition, due to the long operation time and heavy lead suit, the interventional physician may suffer from muscle fatigue, central nervous fatigue and mental fatigue, which may affect the decision-making and operation stability of the interventional physician during the operation.

[0051] To solve the above problems, the embodiment of the present application provides a guide wire bending energy calculation method, an interventional robot and a storage medium. The method comprises the following steps: acquiring a DSA image sequence, wherein the DSA image sequence comprises a guide wire tip image of a target object in an interventional operation process; acquiring a pre-trained guide wire segmentation model, wherein the guide wire segmentation model comprises an encoder, a decoder and a segmentation head module, the encoder comprises a convolution module, an embedding layer and a Transformer module, and the decoder comprises an up-sampling layer; inputting the guide wire tip image into the convolution module for convolution processing to obtain a first feature image; performing image segmentation processing on the first feature image according to a preset image segmentation algorithm to obtain a plurality of feature image blocks; mapping the plurality of feature image blocks to the embedding layer to obtain a second feature image; inputting the second feature image into the Transformer module for image processing to obtain a third feature image; performing image connection processing on the first feature image and the third feature image to obtain a target feature image; inputting the target feature image into the segmentation head module for image segmentation processing to obtain a guide wire segmentation image; and inputting the guide wire segmentation image and imaging state information of the guide wire tip image into a preset guide wire bending energy calculation module to obtain a guide wire bending energy calculation result. The embodiment of the present application introduces an interventional robot to calculate the guide wire bending energy in the percutaneous coronary intervention operation process in real time, and the interventional robot replaces the interventional physician to make corresponding operation actions in the radiation environment, so as to ensure the decision and operation stability of the physician in the operation and reduce the risk of the percutaneous coronary intervention operation.

[0052] The embodiment of the present application will be further described below with reference to the drawings.

[0053] As Figure 1 shown, Figure 1 is a step flowchart of a guide wire bending energy calculation method provided by an embodiment of the present application. The embodiment of the present application provides a guide wire bending energy calculation method, which is applied to an interventional robot. The method comprises the following steps but is not limited to the following steps:

[0054] In step S110, a DSA image sequence is acquired, wherein the DSA image sequence comprises a guide wire tip image of a target object in an interventional operation process.

[0055] In step S120, a pre-trained guide wire segmentation model is acquired, wherein the guide wire segmentation model comprises an encoder, a decoder and a segmentation head module, the encoder comprises a convolution module, an embedding layer and a Transformer module, and the decoder comprises an up-sampling layer.

[0056] In step S130, the guide wire tip image is input into the convolution module for convolution processing to obtain a first feature image.

[0057] Step S140, performing image segmentation processing on the first feature image according to a preset image segmentation algorithm to obtain a plurality of feature image blocks;

[0058] Step S150, mapping the plurality of feature image blocks to an embedding layer to obtain a second feature image;

[0059] Step S160, inputting the second feature image into a Transformer module for image processing to obtain a third feature image;

[0060] Step S170, performing image connection processing on the first feature image and the third feature image based on an upsampling layer to obtain a target feature image;

[0061] Step S180, inputting the target feature image into a segmentation head module for image segmentation processing to obtain a guide wire segmentation image;

[0062] Step S190, inputting the guide wire segmentation image and the imaging state information of the guide wire tip image into a preset guide wire bending energy calculation module to obtain a guide wire bending energy calculation result.

[0063] It can be understood that in the embodiments of the present application, the guide wire segmentation model pre-trained by the intervention robot can be composed of an encoder, a decoder and a segmentation head module. The encoder is a hybrid CNN-Transformer module composed of a convolution module (which can be a ResNet-50 convolution module) and a Transformer module. The hybrid CNN-Transformer module further includes an embedding layer. The decoder includes an upsampling layer and mainly realizes convolution operation and series connection operation related to the convolution module of the encoder. Since the Transformer module needs to take an image sequence as input, in the present embodiment, the guide wire tip image is first input into the convolution module for convolution processing to obtain a first feature image. The first feature image is subjected to image segmentation processing according to a preset image segmentation algorithm to obtain a plurality of feature image blocks. Each feature image block can be represented as , wherein p 2 is the size of each feature image block, and N is the number of the feature image blocks. The expression of N is:

[0064] N = HW / p 2 ;

[0065] HW, wherein HW is the resolution of the first feature image; then, the plurality of feature image blocks are mapped to the embedding layer, and the feature image blocks can be embedded into the embedding layer based on a preset convolution operation to obtain a second feature image; the second feature image is input into a Transformer module for image processing to obtain a third feature image; based on an up-sampling layer, the first feature image and the third feature image are subjected to image connection processing to obtain a target feature image; the target feature image is input into a segmentation head module for image segmentation processing to obtain a guide wire segmentation image; and the imaging state information of the guide wire segmentation image and the guide wire tip image is input into a preset guide wire bending energy calculation module to obtain a guide wire bending energy calculation result. Based on the guide wire segmentation model preset in the interventional robot, the present application can accurately obtain the guide wire segmentation image, provide accurate and effective data basis for calculating the guide wire bending energy, and thus improve the accuracy of the guide wire bending energy. Moreover, in the related art, an interventional physician needs to wear a heavy lead suit and stand to watch an X-ray image screen outside the natural line of sight to control the movement of the guide wire in the blood vessel. Based on this, the interventional physician can only observe whether the guide wire encounters an obstruction or a branch during the process of entering the blood vessel, and cannot determine the bending energy of the guide wire, which has a certain surgical risk. The embodiments of the present application introduce an interventional robot to calculate the guide wire bending energy in the percutaneous coronary intervention process in real time, and the interventional robot replaces the interventional physician to make corresponding surgical actions in the radiation environment, thereby ensuring the decision-making and operation stability of the physician during the operation and reducing the risk of percutaneous coronary intervention.

[0066] It should be noted that the embodiments of the present application do not limit the specific content of the imaging state information of the guide wire tip image, which can include the position offset information of the device for obtaining the guide wire tip image and the data integrity information (whether frame loss) of the guide wire tip image, and no more limitation is made here.

[0067] It should be noted that in the embodiments of the present application, the plurality of first feature image blocks are mapped to the embedding layer to obtain the second feature image, which can be obtained according to the following formula:

[0068]

[0069] wherein Z0 is the second feature image, E is the image block embedding projection, E pos is the embedding position information of the feature image block.

[0070] It should be noted that the convolution module of the embodiments of the present application can be a ResNet-50 convolution module, and those skilled in the art can select the specific structure of the convolution module according to actual needs, and no more limitation is made here.

[0071] It should be noted that the training data set for training the guide wire segmentation model in the embodiment of the present application can be 55 X-ray image sequences, a total of 7131 frames of images, of which 44 sequences are used as a training set and 11 sequences are used as a test set. Each frame of image is subjected to data enhancement processing such as random flipping, rotation, cropping and scaling before being input into the guide wire segmentation model, and the size of each frame of image is set to 512*512 and the batch size is set to 4. For the guide wire segmentation model, a stochastic gradient descent optimizer is used for optimization, and the learning rate is set to 0.01, the momentum is set to 0.9, and the weight decay is set to 1e -4 .

[0072] In addition, in an embodiment, the Transformer module includes a multi-head self-attention module and a multi-layer perception module, and the step S160 in the embodiment shown in Figure 2 , Figure 1 The step S160 in the embodiment shown in

[0073] Step S210, normalizing the second feature image according to a preset normalization algorithm to obtain a first intermediate feature image;

[0074] Step S220, inputting the first intermediate feature image into the multi-head self-attention module for feature extraction to obtain a second intermediate feature image;

[0075] Step S230, inputting the second intermediate feature image into the multi-layer perception module for image processing to obtain a third feature image.

[0076] It can be understood that, Figure 1 The step S160 in the embodiment shown in , i.e., inputting the second feature image into the Transformer module for image processing to obtain a third feature image, can be obtained according to the following steps: normalizing the second feature image according to a preset normalization algorithm to obtain a first intermediate feature image, inputting the first intermediate feature image into the multi-head self-attention module for feature extraction to obtain a second intermediate feature image, and inputting the second intermediate feature image into the multi-layer perception module for image processing to obtain a third feature image, which can provide an effective data basis for obtaining the target feature image.

[0077] It should be noted that the Transformer module of the embodiment of the present application can include a multi-head self-attention module and a multi-layer perception module, and the first intermediate feature image is input into the multi-head self-attention module for feature extraction to obtain a second intermediate feature image, which is obtained according to the following formula:

[0078] Z' ζ = MSA(LN(Z ζ-1 ))+Z ζ-1 ;

[0079] wherein, Z ζ-1 is the second feature image, Z ζ is the second intermediate feature image, LN is the normalization algorithm, and MSA is the multi-head self-attention module.

[0080] It should be noted that in the embodiments of the present application, the second intermediate feature image is input into the multi-layer perception module for image processing to obtain a third feature image, which is obtained according to the following formula:

[0081] Z ζ = MLP(LN(Z ζ ))+Z' ζ ;

[0082] wherein, Z ζ is the third feature image, and MLP is the multi-layer perception module.

[0083] It should be noted that the normalization algorithm in the embodiments of the present application can be an LN normalization algorithm, or a BN normalization algorithm or an IN normalization algorithm, which is not limited here.

[0084] In addition, in an embodiment, the guidewire tip image is obtained by an X-ray machine, the X-ray machine is in communication connection with an interventional robot, and reference is made to Figure 3 , after step S190 in the embodiment shown in Figure 1 , the guidewire bending energy calculation method further includes but is not limited to the following steps:

[0085] Step S310, determining guidewire tip length information, guidewire length information, and imaging parameter information of the X-ray machine;

[0086] Step S320, determining target guidewire curve information according to the guidewire bending energy calculation result, the guidewire tip length information, the guidewire length information, the imaging parameter information, and inter-frame prior information of the guidewire tip image;

[0087] Step S330, when the target guidewire curve information represents that the guidewire tip corresponding to the guidewire tip image presents a ring structure, determining that the guidewire bending energy calculation result is substandard.

[0088] In addition, in an embodiment, reference is made to Figure 4 , after step S190 in the embodiment shown in Figure 1 , the guidewire bending energy calculation method further includes but is not limited to the following steps:

[0089] Step S410, when the guidewire bending energy value corresponding to the guidewire bending energy calculation result exceeds a preset energy threshold, determining that the guidewire bending energy calculation result is substandard.

[0090] It can be understood that the purpose of obtaining the guide wire bending energy calculation result is to detect the risk of guide wire backfolding in the interventional surgery process. By determining the guide wire tip length information, the guide wire length information and the imaging parameter information of the X-ray machine, and determining the target guide wire curve information according to the guide wire bending energy calculation result, the guide wire tip length information, the guide wire length information, the imaging parameter information and the inter-frame prior information of the guide wire tip image, an effective data basis can be provided for determining the evaluation result of the guide wire backfolding risk. The target guide wire curve information can be used to represent the structure of the target guide wire, and can be represented by a continuous mapping γ from a simple open curve c to a plane space point set X:

[0091] γ: c→X;

[0092] Wherein, c is a simple open curve, c=[0, 1], X is a two-dimensional point set of the guide wire tip image plane, and the expression of X is:

[0093] X={x i ;I i =1};

[0094] Wherein, I is a guide wire segmentation image corresponding to the guide wire tip image, I∈{0,1} H×W . It can be seen that the process of obtaining the target guide wire curve information can be understood as obtaining a point set conforming to the continuous mapping γ. In this embodiment, a graph structure is constructed based on the point set X, and the continuous point set is obtained based on the graph structure, and the specific steps are as follows:

[0095] First, an adjacency matrix A∈{0,1} N×N is established to describe the connection relationship of each point in the point set X, where N is the number of points in the point set. A i,j =1 indicates that points i and j are connected, otherwise A i,j =0. Whether two points in the point set X are connected can be determined by two-dimensional neighborhood, and 8-neighborhood is usually used.

[0096] Then, the points with a connection degree of 1, i.e. the points connected to only one other point, can be determined from the point set X based on the characteristics of the adjacency matrix A. The target guide wire segmentation image of this embodiment has and only has two points with a connection degree of 1, which are the head and tail points of the guide wire. However, due to the influence of noise and segmentation errors, there are usually multiple points with a connection degree of 1, and the abnormal points need to be removed by a filtering algorithm. Assuming that there are only two points with a connection degree of 1, the two endpoints γ(0) and γ(1) of the curve can be determined, and the graph traversal method can be used to traverse the point set X from the point γ(0) and record the path p reaching the point γ(1). When there is a loop in the graph structure constructed based on the point set X, there are multiple paths, and a filtering method is needed to obtain the target path from the multiple paths; assuming that there is only one path p={x1,x2,L,x PIf so, the target guidewire curve information can be obtained as γ: c→ p.

[0097] It can be understood that, since the guidewire tip is a curved spring, Hooke's law (linear relationship between spring stress and strain) can be used to calculate the bending energy of the guidewire. Based on the description of the above embodiment, the point set p is taken as the spring node, the adjacent two points can be represented by a rigid straight rod, and the included angle θ between the adjacent two straight rods is the spring strain. The bending potential energy U of the entire curved spring, i.e. the calculation result of the guidewire bending energy, can be represented as the sum of the bending potential energy of all straight rods:

[0098]

[0099] Wherein, a is the spring constant related to the stiffness of the straight rod.

[0100] For the case where multiple guidewire endpoint candidates exist in the point set X and multiple traversal paths exist, all possibilities need to be enumerated, and the target guidewire endpoint and the target path are selected from them based on the physical characteristics of the guidewire tip. For M candidate guidewire endpoints, the combination of the first and last points of the guidewire curve is M×(M-1) / 2, and for each combination, there are a number of candidate paths related to the number of loops of the graph. The embodiments of the present application can determine the target guidewire curve information through a preset objective function, and the expression of the objective function is as follows:

[0101]

[0102] Wherein, SID is the source-to-detector distance of the C-arm of the X-ray machine, SOD is the source-to-object distance of the C-arm of the X-ray machine, L' is the guidewire tip length information, and α is the balance coefficient.

[0103] It can be understood that the guidewire folding phenomenon is that the guidewire tip bends towards the proximal end, forming a larger bending angle or even overlapping. The following two ways can be used to monitor whether the target guidewire is folded or whether there is a risk of guidewire folding: one is that the graph structure of the point set X forms a loop structure, and the other is that the guidewire bending energy value corresponding to the calculation result of the guidewire bending energy exceeds the preset threshold value U th When the target guidewire curve information represents that the guidewire tip image corresponding to the guidewire tip is in a loop structure, or when the guidewire bending energy value corresponding to the calculation result of the guidewire bending energy exceeds the preset energy threshold value, it is determined that the calculation result of the guidewire bending energy is not up to standard, i.e. the target guidewire is folded or the target guidewire has a risk of guidewire folding.

[0104] For the graph structure constructed based on the point set X, the graph structure includes an arbitrary graph and a loop-free graph. The adjacency matrix A corresponding to the arbitrary graph is L=D-A, where D is the degree matrix of A. For the loop-free graph, there is the following identity relationship:

[0105] 0.5 × trace(L) = rank(L);

[0106] Here, trace and rank are the trace and rank of the matrix, respectively. Therefore, the presence of a cycle in the graph structure can be determined by checking whether the above equation holds. Combining the above two cases, the determination of guidewire retraction can be expressed by the following equation:

[0107] b=(0.5×trace(L)≠rank(L))||(U≥U th );

[0108] Where b is a logical variable, || is a logical OR operation, and when b = True, it indicates that the guide wire has folded back.

[0109] In addition, the technical solution provided in this embodiment may also include real-time monitoring of the b signal by the interventional robot. When b = True is detected, the interventional robot is controlled to perform a guidewire retraction operation on the target guidewire to avoid guidewire folding, thereby providing a safety mechanism for interventional surgery and further reducing the risk of interventional surgery.

[0110] It should be noted that the guidewire tip image can include multiple temporally consecutive image frames. Inter-frame prior information can be understood as contextual information between adjacent image frames. For example, obtaining inter-frame prior information helps accelerate the extraction of the target guidewire curve information. If the start and end points of the guidewire in the previous image frame are known, the start and end points of the guidewire in the current image frame can be determined based on the inter-frame prior information without enumeration calculations, thus speeding up the computation process. During interventional surgery, guidewire advancement is continuous and slow; therefore, the movement of the guidewire between adjacent image frames is finite. Based on inter-frame prior information, we can directly determine the start and end points from candidate combinations based on distance information. For example, if the start and end points γ of the guidewire in the image frame corresponding to time t-1 are known... t-1 (0) and γ t-1 (1) Assume the guide wire point set X of the current image frame t The set of points with a connectivity of 1 is Q = {x} t,1 ,x t,2 ,L,x t , q}, where q is the capacity of the point set, calculate {γ} t-1 (0),γ t-1 (1)} and the Euclidean distance matrix of Q The Hungary algorithm is used to determine the start and end points γ of the guidewire in the current frame. t (0) and γ t (1).

[0111] Additionally, refer to Figure 5 In one embodiment, during execution Figure 1Before step S130 in the illustrated embodiment, the guide wire bending energy calculation method further includes but is not limited to the following steps:

[0112] In step S510, the guide wire tip image is preprocessed according to a preset image preprocessing rule to obtain a preprocessed guide wire tip image.

[0113] It can be understood that the main purpose of preprocessing the guide wire tip image is to eliminate irrelevant information in the guide wire tip image, restore useful real information, enhance the detectability of relevant information, and maximize data simplification, thereby improving the reliability of subsequent guide wire tip image applications.

[0114] It should be noted that the present application does not limit the specific method of preprocessing the guide wire tip image, which can be image random number rotation processing, normalization processing, or image binarization processing, etc. Those skilled in the art can select the specific way of image preprocessing according to the actual situation.

[0115] In addition, with reference to Figure 6 , Figure 6 is a hardware schematic diagram of an interventional robot provided by another embodiment of the present application. One embodiment of the present application further provides an interventional robot 600, which comprises a memory 610, a processor 620, and a computer program stored in the memory 610 and executable on the processor 620.

[0116] The processor 620 and the memory 610 can be connected through a bus or other means.

[0117] The non-transitory software program and instructions required for the guide wire bending energy calculation method of the above-mentioned embodiments are stored in the memory 610, and when executed by the processor 620, the guide wire bending energy calculation method in the above-mentioned embodiments is executed, for example, the method steps S110 to S190 in Figure 1 , the method steps S210 to S230 in Figure 2 , the method steps S310 to S330 in Figure 3 , the method steps S410 in Figure 4 , and the method step S510 in Figure 5 .

[0118] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separate, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment scheme.

[0119] Furthermore, one embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions that are executed by a processor or controller, for example, by a processor in the above-described interventional robot 600 embodiment, causing the processor to execute the guide wire bending energy calculation method in the above-described embodiment, for example, to perform the above-described... Figure 1 Method steps S110 to S190, Figure 2 Method steps S210 to S230, Figure 3 Method steps S310 to S330, Figure 4 Method steps S410 and Figure 5 Method step S510. Those skilled in the art will understand that all or some of the steps in the methods disclosed above, and the system, can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method for calculating guide wire bending energy, characterized in that, The method for calculating guidewire bending energy, applied to interventional robots, includes: Acquire a DSA image sequence, which includes images of the guidewire tip of the target object during the interventional procedure; Obtain a pre-trained guidewire segmentation model, which includes an encoder, a decoder, and a segmentation head module. The encoder includes a convolutional module, an embedding layer, and a Transformer module. The decoder includes an upsampling layer. The guidewire tip image is input into the convolution module for convolution processing to obtain the first feature image; The first feature image is segmented according to a preset image segmentation algorithm to obtain multiple feature image blocks; Multiple feature image blocks are mapped to the embedding layer to obtain a second feature image; The second feature image is input into the Transformer module for image processing to obtain the third feature image; Based on the upsampling layer, the first feature image and the third feature image are processed by image concatenation to obtain the target feature image; The target feature image is input into the segmentation head module for image segmentation processing to obtain a guide wire segmentation image; The imaging state information of the guidewire segmentation image and the guidewire tip image is input into a preset guidewire bending energy calculation module to obtain the guidewire bending energy calculation result; The guidewire bending energy calculation result is obtained according to the following formula: in, a The spring constant is related to the stiffness of the straight rod. U The results of the guidewire bending energy calculation are shown. The guidewire tip is a curved spring, and adjacent points in the point set corresponding to the spring nodes are represented by a rigid straight rod. θ The angle between two adjacent rigid straight bars; When there are multiple candidate guidewire endpoints in the point set corresponding to the guidewire structure, the target guidewire curve information is determined by an objective function, the expression of which is as follows: ； in, f Let SID be the distance from the light source of the C-arm of the X-ray machine to the detector of the X-ray machine, SOD be the distance from the light source of the C-arm of the X-ray machine to the bed, and L′ be the guidewire tip length information. α This is the balance coefficient.

2. The method for calculating guide wire bending energy according to claim 1, characterized in that, The Transformer module includes a multi-head self-attention module and a multi-layer perception module. The step of inputting the second feature image into the Transformer module for image processing to obtain the third feature image includes: The second feature image is normalized according to a preset normalization algorithm to obtain the first intermediate feature image; The first intermediate feature image is input into the multi-head self-attention module for feature extraction to obtain the second intermediate feature image; The second intermediate feature image is input into the multilayer sensing module for image processing to obtain the third feature image.

3. The method for calculating guide wire bending energy according to claim 2, characterized in that, The first intermediate feature image is input into the multi-head self-attention module for feature extraction to obtain the second intermediate feature image, which is obtained according to the following formula: ； in, The second feature image, This is the second intermediate feature image. The normalization algorithm is as follows: This refers to the multi-head self-attention module.

4. The method for calculating guide wire bending energy according to claim 3, characterized in that, The second intermediate feature image is input into the multilayer perception module for image processing to obtain the third feature image, which is obtained according to the following formula: ； in, The third feature image, This refers to the multi-layer sensing module.

5. The method for calculating guide wire bending energy according to claim 1, characterized in that, The second feature image is obtained by mapping multiple first feature image blocks to the embedding layer, according to the following formula: ； in, The second feature image, For each of the aforementioned feature image blocks, The number of the feature image blocks. The expression is: ； in, The resolution of the first feature image is [resolution value]. The size of each of the aforementioned feature image blocks, Embed projections for image patches, This refers to the embedding location information of the feature image block.

6. The method for calculating guide wire bending energy according to claim 1, characterized in that, The guidewire tip image is acquired by an X-ray machine, which is communicatively connected to the interventional robot. After inputting the segmented guidewire image, imaging status information, and inter-frame prior information into a preset guidewire bending energy calculation module to obtain the guidewire bending energy calculation result, the method further includes: Determine the guidewire tip length information, guidewire length information, and imaging parameter information of the X-ray machine; The target guidewire curve information is determined based on the guidewire bending energy calculation result, the guidewire tip length information, the guidewire length information, the imaging parameter information, and the inter-frame prior information of the guidewire tip image. When the target guidewire curve information indicates that the guidewire tip corresponding to the guidewire tip image has a ring-shaped structure, it is determined that the guidewire bending energy calculation result does not meet the standard.

7. The method for calculating guide wire bending energy according to claim 1, characterized in that, After inputting the guidewire segmentation image, imaging state information, and inter-frame prior information into a preset guidewire bending energy calculation module to obtain the guidewire bending energy calculation result, the method further includes: If the guidewire bending energy value corresponding to the calculated guidewire bending energy exceeds a preset energy threshold, the calculated guidewire bending energy is determined to be substandard.

8. The method for calculating guide wire bending energy according to claim 1, characterized in that, Before inputting the guidewire tip image into the convolution module for convolution processing to obtain the first feature image, the method further includes: The guidewire tip image is preprocessed according to preset image preprocessing rules to obtain a preprocessed guidewire tip image.

9. An interventional robot, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the guide wire bending energy calculation method as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing computer-executable instructions for performing the guide wire bending energy calculation method as described in any one of claims 1 to 8.

Images