Magnetometer depth adjustment method for a wearable magnetoencephalograph

Abstract

The application discloses a magnetometer depth adjustment method of a wearable magnetoencephalograph, and comprises the following steps: obtaining two-part three-dimensional information of a subject's helmet and face based on an optical scanner, and obtaining the position and direction of the magnetometer relative to the scalp surface of an MRI image by two-step registration; calculating the minimum distance of the bottom surface of the magnetometer moving along the probe axial direction to the scalp based on the position and direction of the magnetometer and the scalp in the MRI coordinate system; and based on the distance, using a motor push rod or other control method to make the magnetometer travel a corresponding distance along the axial direction. The magnetometer depth adjustment method has the characteristics of automation, high precision and high efficiency, can automatically calculate the optimal distance of magnetometer depth adjustment, assist in helmet design, improve the magnetoencephalic source positioning precision and the helmet scalp adhesion, simplify the depth adjustment process, facilitate the use of medical staff, and further promote the medical application of magnetoencephalography.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of biomedical engineering, and particularly relates to a magnetometer depth adjustment method of a wearable magnetoencephalograph. BACKGROUND

[0002] Magnetoencephalography is a non-invasive and non-contact brain magnetic field signal measuring instrument, which is a new type of brain medical device still in the experimental development stage and not yet used for large-scale clinical detection. It has ultra-high sensitivity and higher time and spatial resolution than the current existing brain signal measuring instruments such as electroencephalogram and near-infrared.

[0003] Among them, the magnetoencephalograph is divided into SQUID magnetoencephalograph and OPM magnetoencephalograph. The initial SQUID magnetoencephalograph needs to be in a superconducting environment under liquid nitrogen to measure the magnetoencephalogram.

[0004] However, in recent years, the emergence of SERF (Spin-Exchange Relaxation-Free) atomic magnetometer, i.e. OPM (optical pumping magnetometer), makes it possible to invent a new type of OPM magnetoencephalograph. Its miniaturization, low cost, and measurement position closer to the scalp make it possible to measure the high signal-to-noise ratio of the helmetized brain magnetic field.

[0005] In the technical process of magnetoencephalography, after the registration between the magnetometer and the MRI coordinate system, depth adjustment is needed to obtain the most accurate source positioning result. In the previous depth adjustment problem, people usually use manual adjustment, but this process is inefficient and has low accuracy, which is prone to large manual errors and cannot quickly and accurately achieve automatic depth adjustment, limiting the popularization and application of OPM magnetoencephalograph. SUMMARY

[0006] In view of the depth adjustment requirement of the OPM atomic magnetometer, the traditional depth measurement is based on vernier caliper measurement, which requires complex manual operation and has large errors. The application provides a magnetometer depth adjustment method of a wearable magnetoencephalograph, which can automatically display the optimal distance of magnetometer depth adjustment and quickly and accurately complete the entire registration depth adjustment process.

[0007] To achieve the above purpose, the application provides the following scheme:

[0008] A magnetometer depth adjustment method of a wearable magnetoencephalograph, comprising the following steps:

[0009] Step 1: The point cloud scanned by the optical scanner is regionally grown and segmented into helmet point cloud and face point cloud, and the helmet coordinate system with the magnetometer is registered with the head coordinate system in terms of relative position and direction to obtain a transformation matrix T1;

[0010] Step 2: Register the head coordinate system and the MRI coordinate system in relative position and direction to obtain a transformation matrix T2; based on the transformation matrix T1 and the transformation matrix T2, obtain the relative position and direction of the probe relative to the MRI, that is, the overall transformation matrix operation T=T1*T2, and express the magnetometer and the scalp in the same coordinate system; wherein the MRI is a head nuclear magnetic resonance image;

[0011] Step 3: Based on the relative position and direction of the probe relative to the MRI, calculate the surface normal direction of the scalp point cloud;

[0012] Step 4: Based on the surface normal direction of the scalp point cloud, use the distance range to screen the tangent point candidate points of the magnetometer, and realize the preliminary screening of the tangent point position;

[0013] Step 5: In the corresponding normal of the preliminary screening point, calculate the normal closest to the axial direction of the magnetometer, and define the normal corresponding starting point as the tangent point of the magnetometer and the scalp;

[0014] Step 6: Calculate the distance between the tangent surface and the midpoint of the magnetometer bottom surface as the adjustment distance of the magnetometer from the reference position to the scalp tangent surface, and based on the adjustment distance, use the motor push rod or other control method to make the magnetometer travel a corresponding distance along the axial direction, to realize the automatic depth adjustment of the magnetometer.

[0015] Preferably, in the step 1, the point cloud scanned by the optical scanner is segmented into a helmet point cloud and a face point cloud by region growing, and the coordinate system of the helmet equipped with the magnetometer is registered with the head coordinate system in relative position and direction to obtain the transformation matrix T1, which comprises:

[0016] Using color threshold method to extract the positions of the six digitized reference points on the helmet

[0017] According to the prior known six reference points on the helmet in the MEG system coordinate system

[0018] Calculate the centroids of the two groups of reference points and , and calculate the intercovariance matrix H of the centralized reference point data, and perform eigenvalue decomposition H=USV on H;

[0019] According to the decomposition result, obtain the rotation matrix R and the translation matrix T that align the two groups of reference points, and R and T constitute the transformation matrix T1 of the helmet and the human head coordinate system.

[0020] Preferably, in the step 2, the head coordinate system and the MRI coordinate system are registered in relative position and direction to obtain a transformation matrix T2, which comprises:

[0021] Collecting the transformation matrix T1 weighted MRI data of the tester, segmenting the scalp data by using FreeSurfer software, and extracting the human head point cloud model containing a plurality of vertices;

[0022] Based on the human head point cloud model, the reference point alignment of the head coordinate system and the MRI coordinate system is realized by using the corresponding point method, and a preliminary conversion matrix T2 is obtained.

[0023] For the MRI point cloud and the optical scanner point cloud with different quantities, the ICP algorithm based on KD tree is adopted for further matching to obtain a rotation matrix R and a translation matrix T, and finally the transformation matrix T2 is obtained.

[0024] Preferably, in step 3, the surface normal direction of the scalp point cloud is perpendicular to the surface of the scalp and the bottom surface of the probe, that is, the axial direction of the movement of the magnetometer.

[0025] Preferably, in step 4, based on the surface normal direction of the scalp point cloud, the tangent point candidate points of the magnetometer are screened by using the distance range, and the method for realizing the preliminary screening of the tangent point position comprises:

[0026] The position range of each magnetometer and the tangent point of the scalp is circled, that is, the neighborhood point of the intersection of the axis of the magnetometer and the scalp is taken as the center of the circle, the length of half the bottom edge of the magnetometer is taken as the radius to draw a circle, and the scalp points with a distance less than the radius from the axis in the circle are taken as the candidate points of the tangent point of the magnetometer and the scalp, so that the preliminary screening of the tangent point position is realized.

[0027] Preferably, in step 5, in the corresponding normal of the preliminary screening point, the normal closest to the axial direction of the magnetometer is calculated, and the starting point corresponding to the normal is defined as the tangent point of the magnetometer and the scalp.

[0028] The matching tangent point of each magnetometer is found, that is, the included angle between the normal direction of each point and the direction of the magnetometer is compared in the preliminary screening range, and the point with the smallest included angle is selected as the tangent point of the magnetometer and the scalp.

[0029] Preferably, in step 6, the distance between the tangent surface and the midpoint of the bottom surface of the magnetometer is calculated as the adjustment distance of the magnetometer from the reference position to the tangent surface, and based on the adjustment distance, the magnetometer is moved along the axial direction by a corresponding distance by using a motor push rod or other control method, so that the automatic depth adjustment of the magnetometer is realized.

[0030] The Euclidean distance between the tangent point and the magnetometer and the vertical distance from the tangent point to the axis are calculated, and then the final depth adjustment distance is calculated by using the Pythagorean theorem, and based on the final depth adjustment distance that each probe should move, the magnetometer is moved along the axial direction close to the scalp by a corresponding distance by using a motor push rod or other control method.

[0031] Compared with the prior art, the present application has the following advantages:

[0032] The application realizes a sensor depth adjustment method for a wearable magnetoencephalograph, can greatly improve the depth adjustment efficiency without manual operation, is very friendly to medical personnel without engineering background, has an average running time of 10s, can automatically display the optimal distance of the magnetometer depth adjustment, has a position error of the magnetometer of 0.25±0.03mm and a direction error of 0.27±0.04°, and can assist in the design of the helmet, improve the magnetoencephalography source positioning accuracy and helmet scalp adhesion, and has higher accuracy than manual adjustment, so that the algorithm has the outstanding characteristics of high efficiency, high accuracy and automation. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the present application, the following briefly introduces the drawings needed to be used in the embodiments, obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0034] Figure 1 The flow chart of the sensor depth adjustment method for the wearable magnetoencephalograph in the present application;

[0035] Figure 2 The flow chart of the sensor depth adjustment method for the wearable magnetoencephalograph in the present application;

[0036] Figure 3 The depth adjustment distance effect diagram in the present application. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application, obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0038] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0039] Embodiment one

[0040] The present application provides a magnetometer depth adjustment method for a wearable magnetoencephalograph, comprising the following steps:

[0041] Step 1: The point cloud scanned by the optical scanner is segmented into helmet point cloud and face point cloud by region growing, and the helmet coordinate system with the magnetometer is registered with the head coordinate system in relative position and direction to obtain a transformation matrix T1;

[0042] Step 2: Register the head coordinate system with the MRI coordinate system in terms of relative position and orientation to obtain the transformation matrix T2; based on the transformation matrix T1 and the transformation matrix T2, obtain the relative position and orientation of the probe with respect to the MRI, that is, the overall transformation matrix operation T = T1 * T2, and represent the magnetometer and the scalp in the same coordinate system; where MRI is the head magnetic resonance imaging.

[0043] Step 3: Calculate the surface normal direction of the scalp point cloud based on the relative position and orientation of the probe with respect to the MRI.

[0044] Step 4: Based on the surface normal direction of the scalp point cloud, filter the candidate tangent points of the magnetometer using the distance range to achieve preliminary screening of the tangent point positions;

[0045] Step 5: Calculate the normal line that is closest to the axial direction of the magnetometer in the corresponding normal line of the initial screening point, and set the starting point of the normal line as the tangent point between the magnetometer and the scalp;

[0046] Step 6: Calculate the distance between the cross-section and the midpoint of the bottom surface of the magnetometer. Use this distance as the adjustment distance of the magnetometer from the reference position to the scalp cross-section. Based on the adjustment distance, use a motor push rod or other control methods to move the magnetometer axially a corresponding distance to achieve automatic depth adjustment of the magnetometer.

[0047] In this embodiment, in step 1, a point cloud matching method with corresponding relationships can be used to align the reference points on the helmet coordinate system with the reference points of the optical scan point cloud, thereby realizing MEG-Head (helmet-head) conversion.

[0048] Specifically, in step 1, the method of dividing the point cloud scanned by the optical scanner into helmet point cloud and face point cloud by region growing and segmenting, and registering the relative position and orientation of the helmet coordinate system equipped with the magnetometer with the head coordinate system to obtain the transformation matrix T1 includes:

[0049] The positions of six digitized reference points on the helmet were extracted using a color thresholding method.

[0050] Based on the prior known information of six reference points on the helmet in the MEG system coordinate system

[0051] Calculate two sets of reference points and The centroid is determined, and the cross-covariance matrix H is calculated for the centered reference point data. Then, the eigenvalue decomposition H = USV is performed on H.

[0052] Based on the decomposition results, the rotation matrix R and translation matrix T are obtained to align the two sets of reference points. R and T together form the transformation matrix T1 between the helmet and the head coordinate system.

[0053] In the step 2, in the embodiment, firstly, four points selected from the three-dimensional scanning point cloud data are aligned with four corresponding points in the MRI point cloud data, to realize rough matching; then KD trees between the two groups of point cloud data are established, and the ICP algorithm is used to precisely match the two groups of point cloud data, to realize Head-MRI conversion. Firstly, T1 weighted MRI data of the subject is obtained by using nuclear magnetic resonance scanning, and then the scalp surface is obtained by using the Freesurfer software for tissue segmentation. Since the helmet used is pre-designed, the position and direction of the magnetometer in the helmet coordinate system are prior information before registration, and finally the pose information of the magnetometer can be represented in the MRI coordinate system.

[0054] Specifically, in the step 2, the method for registering the relative position and direction of the head coordinate system and the MRI coordinate system to obtain the transformation matrix T2 comprises:

[0055] The transformation matrix T1 weighted MRI data of the subject is collected, the scalp data is segmented by using the FreeSurfer software, and the human head point cloud model containing a plurality of vertices is extracted;

[0056] Based on the human head point cloud model, the reference point alignment of the head coordinate system and the MRI coordinate system is realized by using the corresponding point method, to obtain the preliminary conversion matrix T2;

[0057] For the MRI point cloud and the optical scanner point cloud with different quantities, the ICP algorithm based on the KD tree is used for further matching to obtain the rotation matrix R and the translation matrix T, and finally the transformation matrix T2 is obtained.

[0058] In the step 3, in the embodiment, the surface normal direction of the scalp point cloud is the direction perpendicular to the scalp and the probe bottom surface, that is, the axial direction of the magnetometer movement. In order to calculate the position of the tangent point of the probe and the scalp, based on the scalp MRI file, the normal of the scalp is calculated, the normal direction C of all head point clouds is calculated, and is stored in an array corresponding to the serial number of the point cloud point a.

[0059] In the step 4, in the embodiment, based on the surface normal direction of the scalp point cloud, the tangent point candidate points of the magnetometer are screened by using the distance range, to realize preliminary screening of the tangent point position. The method comprises:

[0060] The position range of each magnetometer and the scalp tangent point is circled, that is, the neighborhood point of the intersection of the axis of the magnetometer and the scalp is taken as the center, and the length of half of the bottom side of the magnetometer (about 10 mm) is taken as the radius to draw a circle, and the scalp points with a distance less than the radius from the axis in the circle are taken as the candidate points of the magnetometer tangent to the scalp, to realize preliminary screening of the tangent point position.

[0061] In the embodiment, in step 5, the method for determining the normal line closest to the axial direction of the magnetometer in the corresponding normal line at the initial screening point and setting the starting point corresponding to the normal line as the tangent point between the magnetometer and the scalp includes:

[0062] The tangent point matched with each magnetometer is searched, that is, the included angle between the normal line direction of each point and the magnetometer direction is compared in the initial screening range, and the point with the smallest included angle, that is, the point with the most coincident direction, is selected as the tangent point between the magnetometer and the scalp.

[0063] In the embodiment, in step 6, the distance between the tangent surface and the midpoint of the magnetometer bottom surface is calculated as the adjustment distance of the magnetometer from the reference position to the tangent surface of the scalp, and based on the adjustment distance, the magnetometer is driven along the axial direction by a corresponding distance by using a motor push rod or other control method, thereby realizing automatic depth adjustment of the magnetometer.

[0064] The Euclidean distance between the tangent point and the magnetometer and the vertical distance from the tangent point to the axis are calculated, and then the final depth adjustment distance is calculated by using the Pythagorean theorem, and based on the final depth adjustment distance that each probe should move, the magnetometer is driven along the axial direction close to the scalp by a corresponding distance by using a motor push rod or other control method.

[0065] Embodiment two

[0066] As shown in Figure 1 , a sensor depth adjustment method for a wearable magnetoencephalograph is provided, and the entire process is realized based on a PCL library in a Windows and Linux operating system, and the specific steps are as follows:

[0067] Step 1: After the point cloud scanned by an optical scanner is segmented into a helmet point cloud and a face point cloud by region growing, the helmet system coordinate system with a magnetometer is registered with the head coordinate system in terms of relative position and direction by using a corresponding reference point method, to obtain a transformation matrix T1, and the specific implementation is as follows:

[0068] (1) The color threshold method is used to extract the positions of the six digitized reference points on the helmet

[0069] (2) According to the prior known information of the six reference points on the helmet in the helmet coordinate system (for example, the Figure 2 first coordinate system), the centroids c A and c B of the two sets of feature points are calculated A and c B ,

[0070]

[0071] (3) The intercovariance matrix H of the centralized feature point data is calculated

[0072]

[0073] and perform eigenvalue decomposition on H, H = USV;

[0074] where S is the eigenvalue matrix of the cross-covariance matrix, U and V are the left and right multiplication matrices of the eigenvalue S;

[0075] (4) Calculate the rotation matrix R of the two sets of feature points:

[0076] R = UV T ,

[0077] and get the translation matrix T of the two sets of feature points:

[0078] T = -c A R + c B

[0079] Thus, the transformation matrix T1 (i.e. conversion 1 in Figure 2 ) is obtained.

[0080] Step 2: Register the relative position and direction of the head-mounted helmet coordinate system (such as Figure 2 the second coordinate system) under the optical scanner with the head model coordinate system (such as Figure 2 the third coordinate system) under the MRI, and obtain the transformation matrix T2 (i.e. conversion 2 in Figure 2 ); thus, the relative position and direction of the sensor with respect to the MRI are obtained, that is, the overall transformation matrix operation T = T1*T2, and are drawn in the same coordinate system, which is implemented as follows:

[0081] (1) Collect the T1-weighted MRI data of the testee, segment the scalp using FreeSurfer software, and extract the data of the layer into 10,000 vertices;

[0082] (2) First, use the corresponding point method to realize the coarse registration of the head coordinate system and the MRI coordinate system, that is, mark the information of the four reference points on the face in the head coordinate system, and mark the four points in the corresponding MRI image. Similarly, calculate the centroid of the two sets of feature points as in step 1, and calculate the cross-covariance matrix H of the centralized feature point data. Perform eigenvalue decomposition on H, H = USV, to obtain the rotation matrix R: R = UV T and the translation matrix T: T = -c A R + c B , which is the initial transformation matrix T2;

[0083] (3) For the optical scanning point cloud P A and the MRI point cloud P B :

[0084] establish a correspondence between P BKD tree data structure is targeted for the head point cloud three-dimensional data to be processed, first calculate the variance of each dimension data, and then select the dimension with the largest variance as the initial partition dimension; and using the median as the node, form the partition section of this dimension, divide the data into two branches; continue to select the larger variance of the remaining two dimensions as the partition axis (when all dimensions are partitioned, start from the initial dimension iteration), until there is no partition node, that is, the two three-dimensional data binary tree construction is completed, and the nearest neighbor of the target point is searched along the KD tree branching path, and then the mapping point set P B A is extracted from the point cloud P B′ ;

[0085] According to the mapping, the corresponding reference point method in step 1 is used for conversion to obtain the converted new point cloud P A-new

[0086] P A-new = P A R+T,

[0087] fit P B′ , then calculate the matching error between P A-new and P B′ :

[0088]

[0089] When ε> precision requirement, continue iteration of steps a.b., until ε≤ precision requirement, as the final transformation T2;

[0090] (4) Two-step registration design steps are as follows Figure 2 conversion, the results are as follows Figure 2 Registration result diagram.

[0091] Step 3: To calculate the position of the probe tangent to the scalp, find the normal direction c of all head point clouds and store it in an array corresponding to the serial number of the point cloud point a.

[0092] Step 4: Since the magnetometer can only move fixedly along the axial direction on the rigid helmet, first, circumscribe the position range of each magnetometer and the scalp tangent point, that is, take the axis of the magnetometer as the center of the cylinder, and the length of half the base of the magnetometer as the radius of the cylinder, and the head skin points with a distance from the axis less than the radius as the candidate points for the magnetometer and the scalp tangent, thereby realizing the preliminary screening of the tangent point position, which is implemented as follows:

[0093] (1) Screening tangent point range: for each magnetometer, according to the magnetometer axis direction m and the point cloud point a, the magnetometer position p, and the vector ap formed, calculate the distance D of all representative points to the axis,

[0094] ​

[0095] (2) Select all representative points with a distance less than half the length of the magnetometer bottom side (about 10mm) from the axis as the candidate points of the intersection of the magnetometer and the scalp corresponding to the axis.

[0096] Step 5: Calculate the angle between the normal of all candidate points and the axial direction of the magnetometer, and take the point corresponding to the minimum angle as the best fitting tangent point, which is implemented as follows:

[0097] (1) For each candidate point, calculate the angle between its normal direction c and the axial direction m of the magnetometer (all acute angles),

[0098]

[0099] Store it in an array corresponding to the serial number;

[0100] (2) Keep the smallest angle, and take the starting point of the corresponding normal as the optimal fitting tangent point t of the magnetometer

[0101] Step 6: Calculate the distance between the tangent plane and the midpoint of the magnetometer bottom surface as the adjustment distance of the magnetometer from the reference position to the scalp tangent plane, and take the distance as the execution standard of the magnetometer depth adjustment, write it into the automatic adjustment program of the motor push rod on the magnetocephalic helmet, and realize the automatic depth adjustment of the magnetometer.

[0102] Figure 3 Show the final depth adjustment result, the three-dimensional scatter plot of the adjustment depth corresponding to each magnetometer, and the gray level of the ball represents the size of the adjustment distance.

[0103] The above-described embodiments are only descriptions of the preferred modes of the present application and do not limit the scope of the present application. Without departing from the design spirit of the present application, various modifications and improvements to the technical solutions of the present application made by those skilled in the art shall fall within the protection scope determined by the claims of the present application.

Claims

1. A magnetometer depth adjustment method for a wearable magnetoencephalograph, characterized by, The method comprises the following steps: Step 1: Region growing segmentation of the point cloud scanned by the optical scanner into a helmet point cloud and a face point cloud, and registration of the coordinate system of the helmet equipped with a magnetometer with the head coordinate system in terms of relative position and direction to obtain a transformation matrix T 1 ; Step 2: register the relative position and direction of the head coordinate system and the MRI coordinate system to obtain a transformation matrix T 2 ; based on the transformation matrix T 1 and the transformation matrix T 2 , obtain the relative position and direction of the probe relative to the MRI, that is, the overall transformation matrix operation , and represent the magnetometer and the scalp in the same coordinate system; wherein the MRI is a head magnetic resonance image; Step 3: calculating the surface normal direction of the scalp point cloud based on the relative position and direction of the probe relative to the MRI; Step 4: screening the tangent point candidate points of the magnetometer based on the surface normal direction of the scalp point cloud with a distance range to realize the preliminary screening of the tangent point position; Step 5: in the normal of the preliminary screening point, the normal closest to the axial direction of the magnetometer is calculated, and the starting point corresponding to the normal is defined as the tangent point of the magnetometer and the scalp; Step 6: calculating the distance between the tangent surface and the midpoint of the bottom surface of the magnetometer as the adjustment distance of the magnetometer from the reference position to the tangent surface of the scalp, and based on the adjustment distance, the magnetometer is moved along the axial direction by a corresponding distance by using a motor push rod or other control method to realize the automatic depth adjustment of the magnetometer; In step 6, the distance between the tangent surface and the midpoint of the bottom surface of the magnetometer is calculated as the adjustment distance of the magnetometer from the reference position to the tangent surface of the scalp, and based on the adjustment distance, the magnetometer is moved along the axial direction by a corresponding distance by using a motor push rod or other control method to realize the automatic depth adjustment of the magnetometer, and the method comprises: calculating the Euclidean distance of the tangent point and the magnetometer and the perpendicular distance of the tangent point to the axis, and then calculating the final depth adjustment distance by using the Pythagorean theorem, and based on the final depth adjustment distance that each probe should move, the magnetometer is moved along the axial direction close to the scalp by a corresponding distance by using a motor push rod or other control method.

2. The magnetometer depth adjustment method of a wearable magnetoencephalograph according to claim 1, wherein, In the step 1, the point cloud scanned by the optical scanner is segmented into a helmet point cloud and a face point cloud by region growing, and a coordinate system of the helmet equipped with a magnetometer is registered with the head coordinate system in terms of relative position and direction to obtain a transformation matrix T 1 The method comprises: Extracting digitized six reference point positions on a helmet using color thresholding ; According to the prior known six reference point information on the helmet in the MEG system coordinate system ; The centroids of the two sets of reference points are computed and The cross-covariance matrix is computed for the centered reference point data H and H Eigenvalue decomposition is performed on , S is the matrix of eigenvalues of the cross-covariance matrix, U and V are the left and right multiplication matrices of the matrix of eigenvalues S . According to the decomposition result, a rotation matrix aligning the two sets of reference points is obtained R and a translation matrix T , R with T a transformation matrix of the helmet and the human head coordinate system T 1 .

3. The magnetometer depth adjustment method of a wearable magnetoencephalograph according to claim 1, wherein, In step 2, the head coordinate system is registered with the MRI coordinate system in terms of relative position and direction, and a transformation matrix is obtained T 2 The method comprises: Transform matrix of the test subject T 1 The weighted MRI data is segmented into scalp data by using FreeSurfer software, and a human head point cloud model containing a plurality of vertices is extracted. Based on the head point cloud model, the reference point alignment of the head coordinate system and the MRI coordinate system is realized by the corresponding point method, and a preliminary conversion matrix is obtained T 2 ; For the MRI point cloud and the optical scanner point cloud with different quantities, the ICP algorithm based on KD tree is used for further matching to obtain a rotation matrix R and a translation matrix T , and finally obtain a transformation matrix T 2 .

4. The magnetometer depth adjustment method of a wearable magnetoencephalograph according to claim 1, wherein, In step 3, the surface normal direction of the scalp point cloud is the direction perpendicular to the scalp and the bottom surface of the probe, that is, the axial direction of the movement of the magnetometer.

5. The magnetometer depth adjustment method of a wearable magnetoencephalograph according to claim 1, wherein, In step 4, the method for screening the tangent point candidate points of the magnetometer based on the surface normal direction of the scalp point cloud to realize the preliminary screening of the tangent point position comprises: circumscribing the position range of each magnetometer and the tangent point of the scalp, that is, taking the neighborhood point of the intersection of the axis of the magnetometer and the scalp as the center and the length of half the bottom edge of the magnetometer as the radius to draw a circle, and taking the scalp points inside the circle and having a distance smaller than the radius from the axis as the candidate points of the tangent point of the magnetometer and the scalp to realize the preliminary screening of the tangent point position.

6. The magnetometer depth adjustment method of a wearable magnetoencephalograph according to claim 1, wherein, In step 5, in the normal of the preliminary screening point, the normal closest to the axial direction of the magnetometer is calculated, and the starting point corresponding to the normal is defined as the tangent point of the magnetometer and the scalp, and the method comprises: finding the matching tangent point of each magnetometer, that is, comparing the included angle between the normal direction of each point and the direction of the magnetometer in the preliminary screening range, and selecting the point with the smallest included angle as the point with the most coincident direction as the tangent point of the magnetometer and the scalp.

Images