Implantable electrode detection method, detection device and computer readable storage medium

Abstract

The present invention provides a detection method and detection device for an implantable electrode and a computer-readable storage medium, wherein the detection method comprises the following steps: obtaining image information of the implantable electrode; locating the region where the metal electrode sheet is located in the image information; extracting texture information of the region where the metal electrode sheet is located; and obtaining direction information of the metal electrode sheet based on the extracted texture information. The present invention can directly detect the direction of the metal electrode sheet using the diffraction-like characteristic information of the metal electrode sheet on the image information, has high versatility, is convenient for doctors to check the postoperative diagnosis and treatment effect, and improves the diagnostic efficiency and accuracy.

Description

Technical Field

[0001] The present invention relates to the field of electrical stimulation medical technology, and in particular to a detection method and a detection device for an implantable electrode and a computer-readable storage medium. Background Art

[0002] Deep Brain Stimulation (DBS) can be used to treat hereditary dystonia (such as Parkinson's disease, etc.). Its mechanism is to implant DBS electrodes into the patient's brain and use a pulse generator to stimulate certain nerve nuclei deep in the patient's brain to correct abnormal brain electrical circuits, thereby alleviating these neurological symptoms.

[0003] Figure 1 The structure of a DBS electrode 100 commonly used at present is shown. The DBS electrode 100 includes an electrode carrier 1 and a plurality of metal electrode sheets 2 arranged on the electrode carrier 1. In DBS surgery, the metal electrode sheet 2 will be fixed on the electrode carrier 1 and implanted in the basal nerve nucleus of the brain. The metal electrode sheets 2 in different directions will form magnetic fields in different directions, stimulate different nerve nuclei, and have different therapeutic effects. For the DBS electrode 100, the direction of the metal electrode sheet 2 that the doctor needs to determine refers to the position of the projection of the metal electrode sheet 2 on the horizontal plane of the human body (the anatomical plane perpendicular to the vertical axis of the human body) relative to the vertical axis of the human body, and the direction of the magnetic field is determined based on this. The orientation is related to the rotation angle of the DBS electrode 100, but has nothing to do with its translation position. At present, most of them are based on the state of the doctor implanting the DBS electrode 100 during surgery to determine the direction of the metal electrode sheet 2, but as time passes after surgery, the DBS electrode 100 will rotate due to swelling and peristalsis of the brain tissue, so that the metal electrode sheet 2 is greatly different from the direction just after the surgery, and it is necessary to re-acquire the direction of the metal electrode sheet 2 to assist the doctor in subsequent treatment.

[0004] At present, since multiple metal electrode sheets 2 are usually symmetrical about the central axis of the electrode carrier 1, it is impossible to determine the direction of the metal electrode sheet 2 through X-ray images after surgery, and it is usually necessary to add additional marking points on the electrode carrier 1. For example, the prior art discloses a DBS electrode positioning method with indicator marks, which artificially adds marking points that can be presented in medical images to the DBS electrode 100 to obtain the rotation direction of the DBS electrode 100 relative to the implantation position through X-ray images, that is, to obtain the direction information of the metal electrode sheet 2. However, this method requires additional marking points to be added to the DBS electrode 100 to form a specific electrode, which needs to be customized and has low versatility. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide an implantable electrode detection method and detection device and a computer-readable storage medium, which can directly identify the direction of the metal electrode sheet by utilizing the diffraction-like features of the metal electrode sheet in the image, without the need for customized electrodes and with high versatility.

[0006] In order to achieve the above object, according to a first aspect of the present invention, a detection method for an implantable electrode is provided, wherein the implantable electrode comprises an electrode carrier and a metal electrode sheet disposed on the electrode carrier, and the detection method comprises the following steps:

[0007] Acquiring image information of the implantable electrode;

[0008] Locating the area where the metal electrode sheet is located in the image information;

[0009] Extracting texture information of the region where the metal electrode sheet is located; and

[0010] According to the extracted texture information, direction information of the metal electrode sheet is obtained.

[0011] Optionally, after extracting the texture information of the area where the metal electrode sheet is located, the detection method further includes inputting the extracted texture information into a prediction model, the prediction model is used to output the direction information of the metal electrode sheet, and the prediction model is obtained by the following steps:

[0012] Acquire multiple examples of image information of a metal electrode sheet with known direction information;

[0013] For each image information, extract the texture information of the area where the metal electrode sheet is located, and use the extracted texture information and the known direction information of the metal electrode sheet as a sample. For multiple image information, a total of multiple samples are obtained; and

[0014] The prediction model is obtained based on a plurality of sample trainings.

[0015] Optionally, the prediction model is trained and obtained using a support vector machine or a gradient descent tree.

[0016] Optionally, the support vector machine uses a radial basis function, a Laplace function or a Sigmoid function as a kernel function.

[0017] Optionally, the step of locating the area where the metal electrode sheet is located in the image information includes:

[0018] Acquiring the position of the electrode carrier in the image information; and

[0019] According to the relative positional relationship between the electrode carrier and the metal electrode sheet, the region where the metal electrode sheet is located on the electrode carrier is located.

[0020] Optionally, the step of acquiring the position of the electrode carrier in the image information includes:

[0021] Set the region of interest;

[0022] In the region of interest, detecting the endpoints of the electrode carrier using threshold analysis; and

[0023] Based on the endpoints of the electrode carrier and the shape of the electrode carrier, the position of the electrode carrier is iteratively detected by a region growing method.

[0024] Optionally, within the region of interest, the step of detecting the endpoint of the electrode carrier using threshold analysis includes:

[0025] Selecting a direction slice and slicing layer by layer to perform threshold segmentation on the region of interest until the electrode carrier is found for the first time, and taking the position of the electrode carrier found for the first time as a candidate endpoint of the electrode carrier;

[0026] Taking the candidate endpoint as a starting point, performing multi-layer slice detection again, and taking the position of the electrode carrier of each slice as a verification endpoint; and

[0027] The difference between the positions of the candidate endpoint and the verification endpoint is calculated respectively, and whether the candidate endpoint is valid is determined based on the difference. If invalid, the candidate endpoint is reacquired until a valid candidate endpoint is obtained as the endpoint of the electrode carrier.

[0028] Optionally, based on the endpoints of the electrode carrier and the shape of the electrode carrier, the step of iteratively detecting the position of the electrode carrier by a region growing method includes:

[0029] Selecting a direction slice and taking the slice where the end point of the electrode carrier is located as the current slice, and predicting the predicted position of the electrode carrier in the next slice based on the position of the end point of the electrode carrier;

[0030] Binarizing the region of interest of the next slice, and performing morphological calculation on the region corresponding to the predicted position of the electrode carrier in the next slice, to obtain the actual position of the electrode carrier in the next slice; and

[0031] A threshold is set based on the shape of the electrode carrier, and the next slice is used as the current slice for iterative detection until the difference between the position of the electrode carrier in the current slice and the next slice exceeds the threshold, thereby completing the detection of the position of the electrode carrier.

[0032] Optionally, if the electrode carrier is rotationally symmetric about the central axis and has an end face perpendicular to the central axis, then according to the relative positional relationship between the electrode carrier and the metal electrode sheet, the step of locating the region where the metal electrode sheet is located on the electrode carrier includes:

[0033] According to the relative distance between the metal electrode sheet and the end surface of the electrode carrier, the metal electrode sheet is positioned in an annular surface area on the electrode carrier, and the annular surface area is rotationally symmetric about the central axis of the electrode carrier.

[0034] Optionally, the detection method further includes:

[0035] Determining the position of the metal electrode sheet relative to the central axis of the electrode carrier according to the acquired direction information of the metal electrode sheet; and

[0036] The position of the metal electrode sheet in the annular surface area is determined according to the orientation of the metal electrode sheet relative to the central axis of the electrode carrier.

[0037] Optionally, the texture information is extracted through a filter, and the filter is a Gabor filter.

[0038] Optionally, the standard deviation of the Gaussian factor of the Gabor filter and the sine wave frequency satisfy the following relationship:

[0039] σf=0.56

[0040] Where σ is the standard deviation of the Gaussian factor, and f is the frequency of the sine wave;

[0041] When extracting the texture information, it also includes:

[0042] A sine wave frequency is selected and reduced in multiples to obtain four sine wave frequencies, so that the standard deviations of four different Gaussian factors are set according to the above relationship, four scales are constructed, and 360 degrees of the two-dimensional plane are evenly divided into eight directions, eight directions are constructed, and a total of 32 Gabor filters are obtained;

[0043] The texture information is extracted through 32 Gabor filters to obtain a feature vector, and the feature vector is input into the prediction model to obtain the direction information of the metal electrode sheet.

[0044] Optionally, the image information includes a computed tomography image.

[0045] Optionally, the implantable electrode includes a DBS electrode.

[0046] According to a second aspect of the present invention, there is provided a detection device for an implantable electrode, the detection device comprising:

[0047] An image information acquisition unit, used to acquire image information of the implantable electrode;

[0048] A metal electrode sheet positioning unit, used for locating the area where the metal electrode sheet is located in the image information;

[0049] A texture information extraction unit, used to extract texture information of the area where the metal electrode sheet is located; and

[0050] The metal electrode sheet direction information acquisition unit is used to acquire the direction information of the metal electrode sheet according to the extracted texture information.

[0051] Optionally, the metal electrode sheet direction information acquisition unit includes a prediction model training module, which is used to repeatedly extract texture information of the area where a metal electrode sheet with known direction information is located, and take the texture information extracted once and the known direction information of the metal electrode sheet as a sample to obtain multiple samples; and obtain the prediction model based on the training of the multiple samples.

[0052] Optionally, the prediction model training module includes a support vector machine submodule, which is used to train the prediction model using a support vector machine.

[0053] Optionally, the metal electrode sheet positioning unit includes:

[0054] an electrode carrier position acquisition module, used to acquire the position of the electrode carrier in the image information; and

[0055] The relative position positioning module is used to locate the area where the metal electrode sheet is located on the electrode carrier according to the relative position relationship between the electrode carrier and the metal electrode sheet.

[0056] Optionally, the electrode carrier position acquisition module includes:

[0057] An area of ​​interest setting submodule is used to set an area of ​​interest;

[0058] A threshold analysis submodule, configured to detect the endpoints of the electrode carrier within the region of interest using threshold analysis; and

[0059] The region-growing iterative detection submodule is used to iteratively detect the position of the electrode carrier by using a region-growing method based on the endpoints of the electrode carrier and the shape of the electrode carrier.

[0060] Optionally, the texture information extraction unit includes a Gabor filter module, which is used to extract the texture information through a Gabor filter.

[0061] According to a third aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the method for detecting an implantable electrode as described in any technical solution of the present invention.

[0062] The detection method and detection device of the implantable electrode of the present invention and the computer-readable storage medium utilize the metal property of the metal electrode sheet that can form diffraction-like stripes on the image to directly extract the texture information of the surrounding area of ​​the metal electrode sheet. The direction information of the metal electrode sheet can be obtained based on the extracted texture information, which is particularly suitable for determining the direction of the metal electrode sheet of the DBS electrode. For DBS electrodes and other similar electrodes, based on the acquired direction information of the metal electrode sheet and the known relative position relationship between the metal electrode sheet and the electrode carrier, the present invention can further determine the position of the metal electrode sheet on the electrode carrier, which can be used to assist in establishing a three-dimensional model of the metal electrode sheet in the body. The present invention utilizes the metal property of the metal electrode sheet itself for detection, does not require additional customized electrodes, has high versatility, can better assist doctors in determining the direction of the postoperative metal electrode sheet, and improves the accuracy of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a schematic diagram of the structure of an existing DBS electrode;

[0064] Figure 2 A schematic diagram of a flow chart of a method for detecting an implantable electrode according to an embodiment of the present invention;

[0065] Figure 3 A schematic diagram of the bottom end of an electrode carrier of a DBS electrode obtained according to an embodiment of the present invention;

[0066] Figure 4 for Figure 3 A schematic cross-sectional view of an electrode carrier of a DBS electrode of the illustrated embodiment when sliced ​​at the 170th layer;

[0067] Figure 5 for Figure 3 A schematic diagram of the top of the electrode carrier of the DBS electrode obtained in the illustrated embodiment;

[0068] Figure 6 for Figure 3 A schematic diagram of three-dimensional modeling of an electrode carrier of a DBS electrode of the illustrated embodiment;

[0069] Figure 7 A CT image of a cross section of a DBS electrode with a metal electrode sheet implanted in the brain;

[0070] Figure 8 This is a CT image of a cross section of a DBS electrode implanted in the brain, away from the metal electrode sheet.

[0071] Description of reference numerals:

[0072] 100-DBS electrode; 1-electrode carrier; 2-metal electrode sheet; 101-left electrode; 102-right electrode; 200-metal bracket; 103-three-dimensional coordinate axis; S1-step S1; S2-step S2; S3-step S3; S4-step S4. DETAILED DESCRIPTION

[0073] To make the content of the present invention more clear and understandable, the following Figure 1-8 To illustrate the detection method of the implantable electrode of the present invention. However, it is understood that the present invention is not limited to the specific embodiments described below, and general replacements known to those skilled in the art are also covered within the scope of protection of the present invention. It should be noted that the drawings are all in a very simplified form and in an inaccurate scale, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0074] like Figure 2 As shown, the present invention provides a detection method for an implantable electrode, wherein the implantable electrode comprises an electrode carrier and a metal electrode sheet disposed on the electrode carrier. The present invention also provides a detection device for an implantable electrode, which is used to perform the detection method, and the detection device comprises:

[0075] An image information acquisition unit, used to execute step S1, i.e., to acquire image information of the implantable electrode;

[0076] A metal electrode sheet positioning unit, used to execute step S2, i.e., to locate the area where the metal electrode sheet is located in the image information;

[0077] A texture information extraction unit, used to execute step S3, i.e., to extract texture information of the region where the metal electrode sheet is located; and

[0078] The metal electrode sheet direction information acquisition unit is used to execute step S4, that is, to acquire the direction information of the metal electrode sheet according to the extracted texture information.

[0079] The image information may be a computed tomography image (hereinafter referred to as a CT image), or other images that satisfy the requirement that the metal electrode sheet can form diffraction-like fringes. Figure 7A pair of DBS electrodes 100 implanted in the brain has a cross-sectional CT image of a metal electrode sheet 2. The peripheral areas of the pair of DBS electrodes 100 show obvious diffraction-like fringes (i.e., the quasi-diffraction fringes have different grayscale fluctuations and belong to texture information). Figure 8 In the CT image of the cross section of the pair of DBS electrodes 100 implanted in the brain away from the metal electrode sheet 2, only a small area has diffraction-like fringes. The inventors found that metal electrode sheets 2 in different directions will lead to different diffraction-like fringes. Therefore, the direction information of the metal electrode sheet 2 can be obtained based on the texture information of the area where the diffraction-like fringes are extracted, that is, the area where the metal electrode sheet 2 is located. The present invention uses the diffraction-like features of the metal electrode sheet on the image information to obtain the direction information of the metal electrode sheet, which has high versatility and can assist doctors in improving the accuracy of diagnosis.

[0080] Furthermore, the metal electrode sheet direction information acquisition unit includes a prediction model training module, which is used to obtain multiple image information of a metal electrode sheet with known direction information; for each image information, extract the texture information of the area where the metal electrode sheet is located, and use the extracted texture information and the known direction information of the metal electrode sheet as a sample. For multiple image information, a total of multiple samples are obtained; and a prediction model is obtained by training based on multiple samples. After extracting the texture information of the area where the metal electrode sheet is located, the detection method also includes inputting the extracted texture information into the prediction model, and the prediction model outputs the direction information of the metal electrode sheet. The prediction model is obtained by training using a nonlinear model, including models such as a support vector machine or a gradient descent tree. The support vector machine uses a nonlinear function as a kernel function, including a radial basis function, a Laplace function, or a Sigmoid function.

[0081] Furthermore, the prediction model training module includes a support vector machine submodule, which is used to execute the step of training the prediction model using a support vector machine.

[0082] Furthermore, the metal electrode sheet positioning unit includes:

[0083] an electrode carrier position acquisition module, configured to execute the step of acquiring the position of the electrode carrier in the image information; and

[0084] The relative position positioning module is used to execute the step of positioning the area where the metal electrode sheet is located on the electrode carrier according to the relative position relationship between the electrode carrier and the metal electrode sheet.

[0085] Furthermore, the electrode carrier position acquisition module includes:

[0086] A region of interest setting submodule is used to execute the step of setting a region of interest;

[0087] A threshold analysis submodule is used to execute the step of detecting the endpoints of the electrode carrier in the region of interest using threshold analysis; and

[0088] The region-growing-like iterative detection submodule is used to execute the step of iteratively detecting the position of the electrode carrier by a region-growing-like method based on the endpoints of the electrode carrier and the shape of the electrode carrier.

[0089] Among them, when the image information is a CT image, due to the large difference in grayscale between the electrode carrier and the background, threshold analysis can be directly used to detect the endpoints of the electrode carrier. For other images, the image can be processed in advance, such as using an enhancement filter to enhance the appearance of the implanted electrode and remove the background of the image, and then threshold analysis can be used to detect the endpoints of the electrode carrier.

[0090] The quasi-region growing method is a region growing method that specifies a region growing direction. When the endpoints of the electrode carrier are detected and the shape of the electrode carrier is known, a region growing direction can be specified and the direction can be iteratively detected to detect the position of the electrode carrier. The quasi-region growing method generally sets a threshold value according to the shape of the electrode carrier as the maximum range of the coordinates or cross-sectional area of ​​the electrode carrier. When the detected coordinates or cross-sectional area exceeds the threshold value, it is determined that other interferences such as the stent are detected, so that the positions of the electrode carrier and the stent can be distinguished.

[0091] Especially when the bracket is located at the other end of the electrode carrier away from the endpoint, the position of the electrode carrier can be iteratively detected step by step from the endpoint to the other end using a region growing method. When the bracket is detected, the position detection of the electrode carrier is considered to be completed.

[0092] For example, in one embodiment, based on the detected positions of the endpoints of the electrode carrier and the shape of the electrode carrier, the step of iteratively detecting the position of the electrode carrier by a region growing method includes:

[0093] Selecting a direction slice and taking the slice where the end point of the electrode carrier is located as the current slice, and predicting the predicted position of the electrode carrier in the next slice based on the position of the end point of the electrode carrier;

[0094] Binarizing the region of interest of the next slice, and performing morphological calculation on the region corresponding to the predicted position of the electrode carrier in the next slice, to obtain the actual position of the electrode carrier in the next slice;

[0095] A threshold is set based on the shape of the electrode carrier, and the next slice is used as the current slice for iterative detection until the difference between the position of the electrode carrier in the current slice and the next slice exceeds the threshold, thereby completing the detection of the position of the electrode carrier.

[0096] Furthermore, the texture information extraction unit includes a Gabor filter module for extracting the texture information through a Gabor filter. The Gabor filter is very similar to the visual stimulus response of simple cells in the human visual system and has good direction selection and scale selection characteristics. Therefore, it is preferred to use a Gabor filter to extract the texture information of the surrounding area of ​​the area where the metal electrode sheet is located.

[0097] In step S3, for example, the standard deviation of the Gaussian factor of the Gabor filter and the sine wave frequency satisfy the following relationship (Formula 1):

[0098] σf=0.56 (Formula 1)

[0099] Where σ is the standard deviation of the Gaussian factor, and f is the frequency of the sine wave;

[0100] Select a sine wave frequency and press The multiple reduction results in 4 sine wave frequencies, so that the standard deviations of 4 different Gaussian factors are set according to the above relationship, 4 scales are constructed, and 360 degrees of the two-dimensional plane are evenly divided into 8 directions, 8 directions are constructed, and a total of 32 Gabor filters are obtained;

[0101] The texture information is extracted through 32 Gabor filters to obtain a feature vector, and the feature vector is input into the prediction model to obtain the direction information of the metal electrode sheet.

[0102] According to the density of the diffraction stripes of the metal electrode sheet, the sine wave frequency preferably ranges from 0.1 Hz to 0.8 Hz to obtain a better texture extraction effect. The above-mentioned Gabor filter is used to extract the texture information of the area where the metal electrode sheet is located, which has a better effect and can more accurately obtain the direction information of the metal electrode sheet.

[0103] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for detecting an implantable electrode as described in any of the above technical solutions are performed.

[0104] Next, the detection method and device of the implantable electrode and the computer-readable storage medium are further described in conjunction with the following preferred embodiments.

[0105] In this embodiment, the implantable electrode is a DBS electrode 100. For other electrodes (such as nerve stimulation electrodes), if they have a structure similar to the DBS electrode 100, that is, an electrode carrier 1 and a metal electrode sheet 2 arranged on the electrode carrier 1, the detection method, detection device and computer-readable storage medium of the present invention may also be applicable.

[0106] In step S1, image information of the implantable electrode is obtained.

[0107] by Figure 1 Taking the DBS electrode 100 shown as an example, the DBS electrode 100 includes an electrode carrier 1 and a metal electrode sheet 2. The electrode carrier 1 is generally cylindrical, and three metal electrode sheets 2 are arranged on the outer ring surface of the cylinder near the end surface, which are used to release current to stimulate the brain nerve nuclei. During the operation, the electrode carrier 1 and the metal electrode sheet 2 are implanted into the brain together, and the DBS electrodes 100 generally need to be implanted in pairs. The DBS electrode 100 is generally implanted into the brain with its central axis roughly parallel to the vertical axis of the human body, and has a relative top end (i.e., the end close to the top of the head) and a bottom end (i.e., the end away from the top of the head), that is, the electrode carrier 1 has a relative top end and bottom end. In order to distinguish below, the left electrode 101 and the right electrode 102 (see Figure 2-Figure 5 ) represents a pair of DBS electrodes 100 .

[0108] After the DBS electrode 100 is implanted in the brain, the doctor will usually take an image of the brain to check the condition of the implanted DBS electrode 100, that is, to obtain an image of the implanted electrode. The image used in this embodiment is a special CT image, namely a CT angiography image (CT angiography, abbreviated as CTA), which can obtain a three-dimensional data image of the human brain. The three-dimensional data size of the CTA image is 512*512*160 (in pixels). Of course, CTA images of other sizes can also be used. The length, width, and height directions are respectively set to the x-axis (the direction of the human body's coronal axis), the y-axis (the direction of the human body's sagittal axis), and the z-axis (the direction of the human body's vertical axis), so as to establish a three-dimensional coordinate axis 103 (see Figure 6 ).

[0109] In step S2, the region where the metal electrode sheet 2 is located is located in the image information.

[0110] First, the position of the electrode carrier 1 is obtained in the image information, including the following steps:

[0111] (1) Set the region of interest

[0112] The electrode carrier 1 will show a high gray value on the CTA image, and the bottom of the electrode carrier 1 is basically concentrated in the basal cranial nerve position. According to manual experience, the region of interest of a single slice is set to [height / 2-80, height / 2+80, width / 2-50, width / 2+50], where height is the height of the CTA image 512, and width is the width of the CTA image 512. Different CTA images have different widths and heights;

[0113] (2) In the region of interest, the endpoints of the electrode carrier 1 are detected using threshold analysis.

[0114] Taking the Z axis as the direction, the region of interest is segmented layer by layer, that is, the CTA image is binarized and the threshold is set to 1500 until the bottom of the electrode carrier 1 is first found. The results are as follows: Figure 3 As shown. Figure 3 , the bottom of the electrode carrier 1 of the left electrode 101 and the right electrode 102 is morphologically elliptical, and the centroid coordinates and area of ​​the elliptical area are calculated (in a binary image, the centroid coordinates are the average of the coordinate values ​​of all points in the area, and the area is the number of all pixels in the area), and the left electrode 101 information [x1_left, y1_left, z1_left, area1_left] (representing x, y, z coordinate values ​​and area values ​​(area) respectively, the digital subscripts of x, y, z are used to represent a certain coordinate value, left is used to represent the left electrode 101, and the following right is used to represent the right electrode 102) and the right electrode 102 information [x1_right, y1_right, z1_right, area1_right] are obtained respectively, and they are used as candidate endpoints of the electrode carrier 1 of the left electrode 101 and the right electrode 102; the candidate endpoints are Perform three-layer slice detection again as the starting point, and obtain the three left electrode 101 information [x2_left, y2_left, z2_left, area2_left], [x3_left, y3_left, z3_left, area3_left] and [x4_left, y4_left, z4_left, area4_left], and the three right electrode 102 information [x2_right, y2_right, z2_right, area2_right], [x3_right, y3_right, z3_right, area3_right] and [x4_right, y4_right, z4_right, area4_right], which are used as the verification endpoints of the electrode carrier 1 of the left electrode 101 and the right electrode 102 respectively;

[0115] The candidate endpoints are verified, and the difference between the positions of the candidate endpoints and the verification endpoints is calculated respectively, and the validity of the candidate endpoints is determined according to the difference. Taking the left electrode 101 as an example, the difference calculation is performed on a total of 4 values, x1_left, x2_left, x3_left, and x4_left. When the difference values ​​are all less than 2, it is considered to meet the conditions. Similarly, the y coordinate value and the area value are verified respectively. When the difference of the y coordinate satisfies less than 2 and the difference of the area value satisfies less than 20, it is considered that the endpoint of the electrode carrier 1 of the left electrode 101 has been determined. Similarly, the x, y, and area of ​​the right electrode 102 are differentially calculated respectively. If the difference of the x and y coordinates satisfies less than 2 and the difference of the area satisfies less than 20, it is considered that the endpoint of the electrode carrier 1 of the right electrode 102 has been determined. If the conditions are not met, the candidate endpoints are cleared and the candidate endpoints are re-acquired until valid candidate endpoints are obtained as the endpoints of the electrode carrier 1 of the left electrode 101 and the right electrode 102 respectively.

[0116] (3) Based on the endpoints of the electrode carrier 1 and the shape of the electrode carrier 1, the position of the electrode carrier 1 is iteratively detected by a similar region growing method

[0117] The endpoint information of the electrode carrier 1 of the left electrode 101 and the right electrode 102 is obtained by the above steps, and the position information of the electrode carrier 1 of the left electrode 101 in the z layer is set to (x_l, y_l, z1, area_l), and the position information of the electrode carrier 1 of the right electrode 102 is set to (x_r, y_r, z1, area_r). Along the z-axis direction, the position of the electrode carrier 1 in the next layer of slices is predicted based on the position of the electrode carrier 1 in the current layer of slices. Taking the left electrode 101 as an example, the predicted coordinates of the electrode carrier 1 of the left electrode 101 of the z+1 layer are (x_l, y_l, z+1). The region of interest of the z+1 layer is binarized, and the region corresponding to the point (x_l, y_l) is morphologically calculated to calculate its centroid and area, and the actual coordinate information (x_l_actual, y_l_actual, z+1, area_l_actual) of the electrode carrier 1 of the left electrode 101 of the z+1 layer is obtained. Then, the difference of the corresponding values ​​of the z layer and the z+1 layer is calculated respectively. When the difference of the x, y and z coordinates is less than 2 or the area difference is less than 20, continue to detect the upper layer. Figure 4 As shown, a cross-sectional schematic diagram of the electrode carrier 1 of the left electrode 101 and the right electrode 102 obtained at the 170th slice layer, that is, a schematic diagram of a cross section located between the top and the bottom of the electrode carrier 1, according to this step, the position of the electrode carrier 1 in each cross section can be gradually obtained until the top of the electrode carrier 1 is reached.

[0118] like Figure 5As shown, a schematic diagram of the top of the acquired DBS electrode 100 is shown, wherein the electrode carrier 1 is circled to distinguish the electrode carrier 1 from the metal bracket 200. Because the top of the DBS electrode 100 implanted in the brain is connected to the metal bracket 200 fixed outside the skull, the grayscale value of the electrode carrier 1 and the metal bracket 200 are consistent on the CTA image, and adhesion appears on the CTA image. When the difference value of the x, y and z coordinates is greater than 2 or the area difference is greater than 20, the detection of the position of the electrode carrier 1 is completed, and the three-dimensional model of the detected electrode carrier 1 on the three-dimensional coordinate axis 103 is as shown in FIG. Figure 6 shown.

[0119] Next, according to the relative positional relationship between the electrode carrier 1 and the metal electrode sheet 2, the region where the metal electrode sheet 2 is located on the electrode carrier 1 is located:

[0120] Generally, the metal electrode sheet 2 is 1 cm above the end point of the electrode carrier 1. The proportion information is obtained from the CTA image of the human brain data to locate the slice layer where the metal electrode sheet 1 is located. Specifically, there are 3 spacing values ​​obtained from the CTA image, which respectively represent the proportional relationship between 1 pixel and millimeter on x, y and z. For example, if the spacing in the z direction is 0.5, it means that a slice layer is 0.5 mm thick. The metal electrode sheet 2 is 20 layers upward from the end point slice layer of the electrode carrier 1, that is, on an annular surface area on the cylindrical electrode carrier 1. The annular surface area is symmetrical about the central axis of the electrode carrier 1. However, the specific position of the metal electrode sheet 2 in the annular surface area cannot be determined at present, and the direction information of the metal electrode sheet 2 needs to be obtained through the following steps.

[0121] In step S3, the texture information of the region where the metal electrode sheet 2 is located is extracted.

[0122] like Figure 8 As shown in FIG. 1 , the cross section of the electrode carrier 1 without the metal electrode sheet 2 has fewer diffraction-like fringes, while Figure 7 As shown, the cross-section of the area with the metal electrode sheet 2 on the electrode carrier 1 will show diffraction-like stripes in the CTA image, and for the DBS electrode 100, the three metal electrode sheets 2 are located on the same cross-section of the electrode carrier 1, and the diffraction-like stripe areas generated by each other overlap with each other, and the positions of the three metal electrode sheets 2 are relatively fixed. Therefore, the texture information of the entire three metal electrode sheets 2 can be extracted as a whole, and the direction information of the three metal electrode sheets 2 can be obtained thereby. For the sake of simplicity, the texture information and direction information of the metal electrode sheets 2 are still represented below.

[0123] In this embodiment, the texture information of the surrounding area where the metal electrode sheet 2 is located is extracted by window Fourier transform (i.e., Gabor transform, which is formed by modulating a Gaussian kernel and a two-dimensional sine wave), as follows:

[0124] As shown in formula 1, the value of the sine wave frequency f of the Gabor filter is selected, and the standard deviation σ of 4 different Gaussian factors is set to construct 4 different scales (according to the density of the diffraction fringes, f = 0.22 is selected as the frequency of the sine wave, and the values ​​are adjusted in turn according to the density of the diffraction fringes. The 360-degree angle of the two-dimensional plane is divided into 8 directions, and a total of 32 Gabor filters are obtained. Among them, since the direction information that the DBS electrode needs to obtain is the relative position of the projection of the metal electrode sheet 2 on the horizontal plane of the human body and the vertical axis of the human body (related to the direction of the magnetic field), it is preferred to divide the plane composed of the x-axis and y-axis of the coordinate axis (i.e., the horizontal plane of the human body) into 360 degrees, construct multiple directions (for example, divided into 8 directions), and obtain the Gabor filter.

[0125] σf=0.56 (Formula 1)

[0126] Taking the left electrode 101 as an example, the endpoint coordinates of the electrode carrier 1 of the left electrode 101 are [x1_left, y1_left, z1_left, area1_left]. It is considered that the layers z1_left+20 to z1_left+40 are slices containing the metal electrode sheet 2. The 30 pixel value range around the centroid coordinate point is the main diffraction range, that is, the area where the metal electrode sheet 2 is located. The 32 Gabor filters are convolved layer by layer and the mean and standard deviation of the convolved image are calculated to obtain a feature vector composed of 32*20*2 values, which implicitly contains the relative orientation information of the projection of the metal electrode sheet 2 on the horizontal plane of the human body and the vertical axis of the human body.

[0127] In step S4, the direction information of the metal electrode sheet 2 is obtained according to the extracted texture information. In this embodiment, specifically: after extracting the texture information of the area where the metal electrode sheet is located, the texture information is input into the prediction model, and the prediction model outputs the direction information of the metal electrode sheet.

[0128] The prediction model is obtained based on the DBS electrode 100 with known directional information of the metal electrode sheet 2, and is used to calculate the direction of the metal electrode sheet 2 based on the relationship between the regularity of the diffraction-like fringes generated by the metal electrode sheet 2 and the direction of the metal electrode sheet 2. The prediction model is obtained by the following steps: obtaining multiple image information (such as CT images) of a metal electrode sheet 2 with known directional information; for each image information, extracting the texture information of the area where the metal electrode sheet 2 is located, and taking the extracted texture information and the known directional information of the metal electrode sheet 2 as a sample, and for multiple image information, obtaining multiple samples in total; and training based on multiple samples to obtain the prediction model. In this embodiment, a Gabor filter can also be used to extract the texture information of the area where a metal electrode sheet 2 with known directional information is located. The Gabor filter can also be set in the above step S3, and a total of 32 Gabor filters are set. Similarly, since the directional information that the DBS electrode needs to obtain is the relative position of the projection of the metal electrode sheet 2 on the horizontal plane of the human body and the vertical axis of the human body (determining the direction of the magnetic field), in this embodiment, it is preferred to divide the plane composed of the x-axis and y-axis of the coordinate axis (i.e., the horizontal plane of the human body) into 360 degrees, construct multiple directions (for example, equally divided into 8 directions), and obtain the Gabor filter. The texture information extracted by the Gabor filter is also presented in the form of a feature vector, which implicitly contains the information of the relative position of the projection of the metal electrode sheet 2 on the horizontal plane of the human body and the vertical axis of the human body.

[0129] In this embodiment, a total of 200 samples are obtained, and then a support vector machine (SVM) based on radial basis function (RBF) (using RBF function instead of nonlinear mapping to high-dimensional space, RBF function is shown in Formula 2) is used for training 200 times, the number of iterations is 1000 times, the termination condition is less than 0.001, the penalty parameter range is [2^(-8), 2^(8)], the gamma parameter is 1 / 8 (the default is the inverse of the number of categories, here the Gabor filter is set to 8 directions, that is, the number of categories is 8), and 3-fold cross validation is performed to obtain the prediction model.

[0130]

[0131] Where x and y are vectors of a single sample and σ is the width parameter.

[0132] A feature vector consisting of 32*20*2 values ​​obtained in step S3 is input into the prediction model, and the prediction model outputs a direction angle value. The direction angle value is the relative position of the projection of the metal electrode sheet 2 on the plane composed of the x-axis and y-axis of the coordinate axis (the horizontal plane of the human body) and the z-axis (the vertical axis of the human body), that is, the direction information of the metal electrode sheet 2 is obtained. According to the direction information, the direction of the magnetic field generated by the metal electrode sheet 2 can be determined to assist doctors in treatment and diagnosis.

[0133] In this embodiment, for the DBS electrode, it is only necessary to obtain the direction information of the metal electrode sheet 2. However, if it is necessary to mark the position of the electrode in the image information (such as three-dimensional modeling), it is also preferred that the position of the metal electrode sheet 2 in the annular surface area where the metal electrode sheet 2 is positioned on the electrode carrier 1 and the obtained direction information of the metal electrode sheet 2 can be determined, that is, the specific position of the metal electrode sheet 2 is obtained.

[0134] In summary, the present invention provides an implantable electrode detection method, detection device and computer-readable storage medium, which can directly utilize the diffraction-like characteristic information of the metal electrode sheet on the image to detect the direction information of the metal electrode sheet. It has high versatility, is convenient for doctors to check the postoperative diagnosis and treatment effects, and improves the diagnostic efficiency and accuracy.

[0135] The above description is only a description of some embodiments of the present invention, and is not intended to limit the scope of the present invention. Any changes or modifications made by a person skilled in the art in the field of the present invention based on the above disclosure shall fall within the protection scope of the present invention.

Claims

1. A method for detecting an implantable electrode, wherein the implantable electrode comprises an electrode carrier and a metal electrode sheet disposed on the electrode carrier, characterized in that: The steps include: Acquiring image information of the implantable electrode; Locating the area where the metal electrode sheet is located in the image information; Extracting texture information of the area where the metal electrode sheet is located; as well as Acquiring direction information of the metal electrode sheet according to the extracted texture information; After extracting the texture information of the area where the metal electrode sheet is located, the method further includes inputting the extracted texture information into a prediction model, wherein the prediction model is used to output the direction information of the metal electrode sheet, and the prediction model is obtained by the following steps: obtaining multiple examples of image information of a metal electrode sheet with known direction information; for each example of image information, extracting the texture information of the area where the metal electrode sheet is located, and taking the extracted texture information and the known direction information of the metal electrode sheet as a sample, and obtaining multiple samples for multiple examples of image information; and obtaining the prediction model based on a plurality of the sample training; The step of locating the region where the metal electrode sheet is located in the image information includes: acquiring the position of the electrode carrier in the image information; And according to the relative positional relationship between the electrode carrier and the metal electrode sheet, the region where the metal electrode sheet is located on the electrode carrier is located.

2. The method for detecting an implantable electrode according to claim 1, wherein: The prediction model is trained and obtained by using a support vector machine or a gradient descent tree.

3. The method for detecting an implantable electrode according to claim 2, wherein: The support vector machine uses radial basis function, Laplace function or Sigmoid function as kernel function.

4. The method for detecting an implantable electrode according to claim 1, wherein: The step of obtaining the position of the electrode carrier in the image information comprises: Set the region of interest; In the region of interest, detecting the endpoints of the electrode carrier using threshold analysis; and Based on the endpoints of the electrode carrier and the shape of the electrode carrier, the position of the electrode carrier is iteratively detected by a region growing method.

5. The method for detecting an implantable electrode according to claim 4, wherein: In the region of interest, the step of detecting the endpoint of the electrode carrier using threshold analysis comprises: Selecting a direction slice and slicing layer by layer to perform threshold segmentation on the region of interest until the electrode carrier is found for the first time, and taking the position of the electrode carrier found for the first time as a candidate endpoint of the electrode carrier; Taking the candidate endpoint as a starting point, performing multi-layer slice detection again, and taking the position of the electrode carrier of each slice as a verification endpoint; and The difference between the positions of the candidate endpoint and the verification endpoint is calculated respectively, and whether the candidate endpoint is valid is determined based on the difference. If invalid, the candidate endpoint is reacquired until a valid candidate endpoint is obtained as the endpoint of the electrode carrier.

6. The method for detecting an implantable electrode according to claim 5, wherein: Based on the endpoints of the electrode carrier and the shape of the electrode carrier, the step of iteratively detecting the position of the electrode carrier by a region growing method comprises: Selecting a direction slice and taking the slice where the end point of the electrode carrier is located as the current slice, and predicting the predicted position of the electrode carrier in the next slice based on the position of the end point of the electrode carrier; Binarizing the region of interest of the next slice, and performing morphological calculation on the region corresponding to the predicted position of the electrode carrier in the next slice, to obtain the actual position of the electrode carrier in the next slice; and A threshold is set based on the shape of the electrode carrier, and the next slice is used as the current slice for iterative detection until the difference between the position of the electrode carrier in the current slice and the next slice exceeds the threshold, thereby completing the detection of the position of the electrode carrier.

7. The method for detecting an implantable electrode according to claim 1, wherein: If the electrode carrier is rotationally symmetric about the central axis and has an end face perpendicular to the central axis, then according to the relative positional relationship between the electrode carrier and the metal electrode sheet, the step of locating the area where the metal electrode sheet is located on the electrode carrier includes: According to the relative distance between the metal electrode sheet and the end surface of the electrode carrier, the metal electrode sheet is positioned in an annular surface area on the electrode carrier, and the annular surface area is rotationally symmetric about the central axis of the electrode carrier.

8. The method for detecting an implantable electrode according to claim 7, wherein: The detection method further comprises: Determining the position of the metal electrode sheet relative to the central axis of the electrode carrier according to the acquired direction information of the metal electrode sheet; and The position of the metal electrode sheet in the annular surface area is determined according to the orientation of the metal electrode sheet relative to the central axis of the electrode carrier.

9. The method for detecting an implantable electrode according to claim 1, wherein: The texture information is extracted through a filter, and the filter is a Gabor filter.

10. The method for detecting an implantable electrode according to claim 9, characterized in that: The standard deviation of the Gaussian factor of the Gabor filter and the sine wave frequency satisfy the following relationship: ； Where σ is the standard deviation of the Gaussian factor, and f is the frequency of the sine wave; When extracting the texture information, it also includes: Select a sine wave frequency and press The multiple reduction results in 4 sine wave frequencies, so that the standard deviations of 4 different Gaussian factors are set according to the above relationship, 4 scales are constructed, and 360 degrees of the two-dimensional plane are evenly divided into 8 directions, 8 directions are constructed, and a total of 32 Gabor filters are obtained; The texture information is extracted through 32 Gabor filters to obtain a feature vector, and the feature vector is input into the prediction model to obtain the direction information of the metal electrode sheet.

11. The method for detecting an implantable electrode according to any one of claims 1 to 10, characterized in that: The image information includes a computed tomography image.

12. The method for detecting an implantable electrode according to any one of claims 1 to 10, characterized in that: The implantable electrode includes a DBS electrode.

13. A detection device for an implantable electrode for performing the detection method described in any one of 1 to 12, characterized in that: include: An image information acquisition unit, used to acquire image information of the implantable electrode; A metal electrode sheet positioning unit, used for locating the area where the metal electrode sheet is located in the image information; A texture information extraction unit, used to extract texture information of the area where the metal electrode sheet is located; as well as The metal electrode sheet direction information acquisition unit is used to acquire the direction information of the metal electrode sheet according to the extracted texture information.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting an implantable electrode according to any one of claims 1 to 12 are performed.

Images