A magnetoencephalography apparatus and a method of registering a magnetoencephalography apparatus with an MRI

Abstract

The application relates to a magnetoencephalography device and a method for registering the magnetoencephalography device and MRI, and belongs to the field of magnetoencephalography research. The method comprises the following steps: acquiring a head image of a subject wearing a magnetoencephalography device; point cloudizing the head image of the subject to obtain a scene point cloud; performing corresponding point clustering processing on the scene point cloud and a marker module model point cloud, identifying the marker module, and determining the position coordinates of the marker module; determining the sensor probe position according to the position coordinates and the relative position relationship between the marker module and the sensor array; acquiring a three-dimensional MRI image of the head of the subject, and separating the three-dimensional MRI image into a scalp three-dimensional image and a cerebral cortex anatomical structure; extracting an MRI facial image according to the scalp three-dimensional image; and registering the MRI facial image and the head image of the subject based on the sensor probe position, so that the sensor array and the cerebral cortex anatomical structure are in a unified coordinate system. The method improves the accuracy of registration.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of magnetoencephalography, and in particular to a magnetoencephalography measuring device and a method for registering a magnetoencephalography measuring device with an MRI. BACKGROUND

[0002] Magnetoencephalography (MEG) is a powerful functional neuroimaging technique that detects the tiny extracranial magnetic fields generated by electrical currents in the neurons of the brain to realize the detection of brain function. Magnetoencephalography can provide a non-invasive window for studying brain activity and diagnosing brain diseases. Since magnetoencephalography is a direct inference of physiological phenomena in the brain, it has good temporal accuracy. In addition, because the magnetic field is relatively unaffected by the uneven conductivity of the head, magnetoencephalography has higher spatial resolution than electroencephalography (EEG). These advantages make magnetoencephalography a powerful tool for studying brain activity and function, and have important significance for cutting-edge neuroscience, clinical applications, etc. At present, magnetoencephalography technology has been applied to language, vision, hearing, somatosensory evoked, etc. Brain wave signal research, preoperative brain function localization in neurosurgery, diagnosis of intracranial diseases, etc.

[0003] In recent years, the maturity of weak magnetic sensors represented by atomic magnetometers has greatly promoted the development of a new generation of magnetoencephalography. Compared with traditional magnetoencephalography measurement systems based on superconducting quantum interference devices (SQUIDs), the new generation of magnetoencephalography systems no longer require a large head device, and only need the subject to wear a small magnetoencephalography helmet in a magnetic shielding environment for measurement, with the advantages of small size, light weight, and convenient measurement. However, in the new generation of magnetoencephalography systems, there are problems such as non-fixed sensor array, non-fixed insertion depth of the helmet, sensor line obstruction, and long time-consuming scanning of the sensor array, which cause certain difficulties in scanning the probe position. In addition, the large number of sensor probes and cables on the helmet make it difficult for the magnetoencephalography system to measure the subject in a supine experimental scenario.

[0004] In magnetoencephalography research, correctly positioning the sensor array is an essential step, but traditional registration methods are time-consuming and laborious, and are difficult to support practical applications. For example, traditional registration methods such as using color markers to position sensors and using coils to measure sensor positions are not only tedious and time-consuming, but also introduce a large amount of human error during operation, making them unsuitable for practical use. How to quickly and accurately scan and position the sensor array on the magnetoencephalography helmet has become an important problem for new magnetoencephalography systems.

[0005] In addition, how to accurately register the scanned image and the cerebral cortex in the same coordinate system is crucial to the accuracy of the subsequent signal source tracing of the magnetoencephalogram. The traditional method uses the iterative closest point algorithm (ICP) to register the three-dimensional scanned image and the MRI face image, but the iterative closest point algorithm (ICP) has a high requirement for the initial position of the point cloud to be registered. If the initial position selected is unreasonable, the algorithm will fall into local optimization, resulting in inaccurate registration. SUMMARY

[0006] The purpose of the present application is to provide a magnetoencephalography device and a method for registering the magnetoencephalography device with MRI, so as to solve the problem of inaccurate registration result of the traditional registration method.

[0007] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0008] A magnetoencephalography device, comprising a magnetoencephalography helmet main body, an identification module and a sensor array; the sensor array and the identification module are arranged on the upper surface of the magnetoencephalography helmet main body.

[0009] The magnetoencephalography helmet main body is used for being worn on the head of a subject.

[0010] The sensor array is used for detecting the extracranial magnetic field generated by the current in the cerebral neuron.

[0011] The identification module is used for positioning the sensor array.

[0012] Optionally, the identification module is a cross-shaped cylinder, a T-shaped cuboid or a cone.

[0013] A method for registering a magnetoencephalography device with MRI, comprising:

[0014] Obtaining the head image of a subject wearing the above-mentioned magnetoencephalography device;

[0015] Point cloudizing the head image of the subject to obtain a scene point cloud;

[0016] Performing corresponding point clustering processing on the scene point cloud and the model point cloud of the constructed identification module model, identifying the identification module of the magnetoencephalography device and determining the position coordinates of the identification module;

[0017] Determining the sensor array probe position coordinates according to the position coordinates of the identification module and the relative position relationship between the identification module and the sensor array;

[0018] Obtaining the three-dimensional MRI image of the head of the subject, and separating the three-dimensional MRI image of the head of the subject into a scalp three-dimensional image and a cerebral cortex anatomical structure;

[0019] Extracting an MRI face image according to the scalp three-dimensional image;

[0020] based on the sensor array probe position coordinates, registering the MRI face image with the subject head image, so that the sensor array and the cerebral cortex anatomical structure are in a unified coordinate system.

[0021] Optionally, the corresponding point clustering processing of the scene point cloud and the model point cloud of the constructed identification module model is performed to identify the identification module of the wearable neuromagnetic measurement device and determine the position coordinates of the identification module, specifically comprising:

[0022] The scene point cloud and the model point cloud are respectively down-sampled to extract scene key points and model point cloud key points;

[0023] According to the scene key point feature descriptor and the model point cloud key point feature descriptor, the corresponding points of the identification module model in the scene are calculated, and a corresponding point pair set between the scene descriptor point cloud and the model descriptor point cloud is determined; the scene descriptor point cloud comprises scene key point feature descriptors; the model descriptor point cloud comprises model point cloud key point feature descriptors;

[0024] According to the scene key point feature descriptor and the model point cloud key point feature descriptor, the corresponding points of the identification module model in the scene are calculated, and a corresponding point pair set between the scene descriptor point cloud and the model descriptor point cloud is determined; the scene descriptor point cloud comprises scene key point feature descriptors; the model descriptor point cloud comprises model point cloud key point feature descriptors;

[0025] The corresponding point pair set is subjected to corresponding point clustering processing, and the corresponding point cluster in the scene point cloud matched with the model point cloud is identified to obtain the identification module of the wearable neuromagnetic measurement device and determine the position coordinates of the identification module.

[0026] Optionally, based on the sensor array probe position coordinates, the MRI face image is registered with the subject head image, so that the sensor array and the cerebral cortex anatomical structure are in a unified coordinate system, specifically comprising:

[0027] The feature descriptors of the subject head image and the MRI face image are respectively calculated to obtain head image feature descriptors and face image feature descriptors;

[0028] According to the head image feature descriptors and the face image feature descriptors, a sample consistency registration algorithm is used to coarsely register the subject head image and the MRI face image to obtain a first coordinate transformation matrix;

[0029] The first coordinate transformation matrix is subjected to normal distribution transformation to obtain a second coordinate transformation matrix;

[0030] The first coordinate transformation matrix and the second coordinate transformation matrix are applied to the sensor array probe position coordinates to complete registration of the MRI face image and the subject head image, so that the sensor array and the cerebral cortex anatomical structure are aligned in the same coordinate system.

[0031] Optionally, the corresponding point clustering method is a Hough voting algorithm or a geometric consistency clustering algorithm.

[0032] Optionally, according to the scene point cloud key points and the model point cloud key points, scene key point feature descriptors and model point cloud key point feature descriptors are calculated based on a direction histogram feature, a point histogram feature, an angle histogram feature, a normal alignment radial feature, an inertia moment feature or an eccentricity feature.

[0033] According to the specific embodiments of the present application, the following technical effects are disclosed:

[0034] The present application provides a magnetoencephalography device and a method for registering a magnetoencephalography device and an MRI image. The magnetoencephalography device comprises an identification module. The head image of a subject wearing the magnetoencephalography device is obtained, the identification module in the image is identified, the position of the identification module and the relative position relationship between the identification module and the sensor array are used to determine the coordinates of the sensor array probe position. In the present application, the identification module is used to position the sensor array, without the need to scan all sensor probes, thereby greatly saving the scanning time and improving the overall efficiency. The subject head image and the MRI image are registered by using the sample consistency registration and the normal distribution transformation, so that the sensor array and the cerebral cortex anatomical structure are placed in the same coordinate system. The use of the nearest point iteration algorithm (ICP) for registration has a high requirement for the initial position of the point cloud to be registered. If the selected initial position is unreasonable, the nearest point iteration algorithm will fall into local optimization, thereby causing the problem of inaccurate registration. The registration method of the present application can improve the registration accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0036] Figure 1 A structural schematic diagram of the magnetoencephalography device provided by the present application is shown in the figure.

[0037] Figure 2 A schematic diagram of the identification module provided by the present application is shown in the figure.

[0038] Figure 3 A flow chart of a method for registering a magnetoencephalography device with an MRI image is provided in the present application.

[0039] Figure 4 A flow chart of a registration method in practical application of the present application.

[0040] Symbol explanation: 1, main body of magnetoencephalography helmet; 2, identification module; 3, magnetic sensor. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0042] The purpose of the present application is to provide a magnetoencephalography device and a method for registering the magnetoencephalography device with an MRI, so as to solve the problem of inaccurate registration result of the conventional registration method.

[0043] The present application uses a corresponding point clustering algorithm to position the sensor array through the identification module on the magnetoencephalography device, and performs registration of the magnetoencephalography device with the MRI image of the subject through a sampling consistency algorithm (SAC-IA) and a normal distribution transformation (NDT), so as to reduce the requirement for coarse registration and improve the rapidity and accuracy of the entire scanning registration process.

[0044] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0045] Figure 1 A structural schematic diagram of a magnetoencephalography device provided in the present application is shown in FIG. 1, which comprises a magnetoencephalography helmet main body 1, an identification module 2 and a sensor array. Figure 1 The sensor array and the identification module 2 are both arranged on the upper surface of the magnetoencephalography helmet main body 1. The sensor array comprises a plurality of magnetic sensors 3. The magnetoencephalography device is a wearable magnetoencephalography device.

[0046] The magnetoencephalography helmet main body 1 is used for wearing on the head of a subject; the sensor array is used for detecting the extracranial magnetic field generated by the current in the brain neurons; and the identification module 2 is used for positioning the sensor array. The transmission information between the brain neurons relies on the potential difference generated after the neurons receive stimulation, so there is a local current on the brain neurons.

[0047] Further, the identification module 2 is arranged at a preset position on the upper surface of the brain magnetic helmet body 1, and the identification module 2 at the preset position is not blocked by the cable of the magnetic sensor 3.

[0048] Further, the identification module 2 is a cross-shaped cylinder, a T-shaped cuboid or a cone. In actual application, the identification module 2 is a cross-shaped cylinder, as shown in Figure 2

[0049] Figure 3 A method flow chart for registering a brain magnetic measurement device with MRI provided by the present application is shown in Figure 4 A registration method flow chart in actual application of the present application is shown in Figure 3 and Figure 4 The method comprises the following steps:

[0050] Step 301: Obtain the head image of a subject wearing the brain magnetic measurement device.

[0051] Step 302: Point cloud the head image of the subject to obtain a scene point cloud.

[0052] In actual application, a three-dimensional structured light scanner is used to scan the head of a subject wearing a brain magnetic measurement device to obtain a head three-dimensional point cloud as a scene point cloud, and the point cloud of the identification module model is taken as a model point cloud. The scene point cloud comprises an identification module point cloud and a face point cloud, and the identification module model is known. The model point cloud can be directly obtained by point cloud conversion of the known identification module model. The known identification module model refers to the shape of the known identification module model, and the complete shape point cloud thereof can be directly obtained.

[0053] Step 303: Perform corresponding point clustering processing on the scene point cloud and the model point cloud of the constructed identification module model to identify the identification module of the brain magnetic measurement device and determine the position coordinates of the identification module. In actual application, the key points of the above-mentioned scene point cloud and model point cloud are obtained, all corresponding point clusters in the scene point cloud that match the identification module model are identified, and the corresponding point clusters are the identification module to be identified on the brain magnetic measurement device.

[0054] Further, the step 303 specifically comprises the following steps:

[0055] Step 3031: Perform down-sampling processing on the scene point cloud and the model point cloud respectively to extract the key points of the scene point cloud and the key points of the model point cloud.

[0056] ​Step 3031: According to the scene key point cloud and the model key point cloud, the scene key point feature descriptor and the model key point cloud feature descriptor are calculated based on the direction histogram feature, the point histogram feature, the angle histogram feature, the normal alignment radial feature, the moment of inertia feature or the eccentricity feature.

[0057] According to the scene key point cloud and the model key point cloud, the scene key point feature descriptor and the model key point cloud feature descriptor are calculated based on the direction histogram feature, the point histogram feature, the angle histogram feature, the normal alignment radial feature, the moment of inertia feature or the eccentricity feature.

[0058] In practical applications, for the scene point cloud and the identification module model point cloud, the normal vector of each point is calculated respectively, and uniform down-sampling is performed to calculate the key points. The direction histogram feature descriptor (scene key point feature descriptor and model key point cloud feature descriptor) is calculated based on the normal vector and the key points. The key points can be obtained based on the normal alignment radial feature or the scale invariant feature transformation.

[0059] Step 3031: According to the scene key point feature descriptor and the model key point cloud feature descriptor, the corresponding points of the identification module model in the scene are calculated, and the corresponding point pair set between the scene descriptor point cloud and the model descriptor point cloud is determined; the scene descriptor point cloud includes the scene key point feature descriptor; the model descriptor point cloud includes the model key point cloud feature descriptor.

[0060] In practical applications, for each point in the model key point cloud feature descriptor obtained in the previous step, an effective nearest neighbor search is performed in the Euclidean space using a Kd tree structure to find the corresponding points in the scene descriptor point cloud corresponding to the model descriptor point cloud.

[0061] Step 3031: The corresponding point pair set is subjected to corresponding point clustering processing, and the corresponding point clusters in the scene point cloud that match the model point cloud are identified, obtaining the identification module of the magnetoencephalography device, and determining the position coordinates of the identification module.

[0062] In practical applications, the Hough Voting or GC algorithm is used to cluster the corresponding point pair set obtained in the previous step, and all corresponding point clusters in the scene point cloud that match the identification module model point cloud are identified. The corresponding point cluster is the identification module on the magnetoencephalography device, and the identification module coordinates are obtained.

[0063] Step 304: determining the sensor array probe position coordinates according to the position coordinates of the identification module and the relative position relationship between the identification module and the sensor array. In practical applications, the magnetic sensor coordinates can be calculated according to the relative position relationship between the identification module and the sensor array, and then the sensor array probe position coordinates can be obtained.

[0064] Step 305: obtaining a three-dimensional MRI image of the subject's head, and separating the three-dimensional MRI image of the subject's head into a three-dimensional image of the scalp and a cerebral cortex anatomical structure. In practical applications, a T1 structural image of the subject's head is obtained using a nuclear magnetic resonance device, reconstructed into a three-dimensional image (three-dimensional MRI image of the subject's head), and separated into a three-dimensional image of the scalp and a cerebral cortex anatomical structure.

[0065] Step 306: extracting an MRI facial image according to the three-dimensional image of the scalp. In practical applications, a three-dimensional facial image (MRI facial image) is extracted from the three-dimensional image of the scalp.

[0066] Step 307: registering the MRI facial image with the subject's head image based on the sensor array probe position coordinates, so that the sensor array and the cerebral cortex anatomical structure are in a unified coordinate system.

[0067] Further, the step 307 specifically includes:

[0068] Step 3071: calculating the feature descriptors of the subject's head image and the MRI facial image respectively to obtain a head image feature descriptor and a facial image feature descriptor. The fast point histogram feature descriptors of the subject's head image obtained in step 301 and the MRI facial image obtained in step 306 are calculated respectively.

[0069] Step 3072: performing coarse registration on the subject's head image and the MRI facial image using a sample consistency registration algorithm according to the head image feature descriptor and the facial image feature descriptor to obtain a first coordinate transformation matrix. In practical applications, the coarse registration is performed using a sample consistency algorithm (SAC-IA) according to the descriptors in step 3071 to obtain a coordinate transformation matrix M1 (first coordinate transformation matrix).

[0070] Step 3073: performing normal distribution transformation on the first coordinate transformation matrix to obtain a second coordinate transformation matrix. In practical applications, the coarse registration result obtained in step 3072 is refined using normal distribution transformation (NDT) to obtain a coordinate transformation matrix M2 (second coordinate transformation matrix).

[0071] The coordinate transformation matrices M1 and M2 have the following expressions:

[0072]

[0073] wherein R 3*3 is a rotation matrix from the source point cloud to the target point cloud, T 3*1 is a translation matrix from the source point cloud to the target point cloud.

[0074] Step 3074: applying the first coordinate transformation matrix and the second coordinate transformation matrix to the sensor array probe position coordinates, completing the registration of the MRI face image and the subject head image, aligning the sensor array and the cerebral cortex anatomical structure in the same coordinate system.

[0075] Applying the coordinate transformation matrices M1 and M2 obtained in steps 3072 and 3073 to the magnetic sensor coordinates obtained in step 304, the registration of the magnetoencephalography device and the MRI image is completed, that is, the sensor array and the cerebral cortex anatomical structure are aligned in the same coordinate system, thereby improving the accuracy of subsequent signal tracing of magnetoencephalogram.

[0076] The present application relates to a magnetoencephalography device and a method for registering the magnetoencephalography device with an MRI image, which comprises a marker module arranged on the magnetoencephalography device, and realizes the registration of the head three-dimensional image of the subject wearing the magnetoencephalography device with the MRI image. The method simultaneously obtains the head image of the subject wearing the magnetoencephalography device and the digital point cloud model (marker module model) of the marker module in the magnetoencephalography device through a scanner, positions and identifies the marker module on the magnetoencephalography device based on the corresponding point clustering of key point identification, thereby obtaining the position information of the sensor array probe on the magnetoencephalography device according to the known relative position relationship between the marker module and the sensor array; and registering the MRI image of the subject through the sampling consistency algorithm (SAC-IA) and the normal distribution transformation (NDT).

[0077] Compared with the prior art, the present application has the following advantages:

[0078] The present application focuses on solving the scanning registration problem of wearable magnetoencephalogram and MRI image, and proposes an optical scanning and sensor identification method for the wearable magnetoencephalography device, as well as an accurate registration method with the MRI image.

[0079] Based on the known relative position of the marker module and the sensor on the magnetoencephalography device, a method for positioning the sensor coordinates by identifying the marker using a key point-based corresponding point clustering algorithm is proposed.

[0080] When performing three-dimensional optical scanning on the head of the subject wearing the magnetoencephalography device using the method for positioning the sensor according to the marker module, only the complete marker module shape and the subject's facial information need to be scanned, and all sensor probes do not need to be scanned, thereby greatly saving the scanning time and improving the overall efficiency.

[0081] The method is flexible, and the identification module is suitable for various wearable brain magnetic measurement devices and can solve the defect that the traditional magnetoencephalogram cannot be measured in a lying position.

[0082] The position of the identification module on the brain magnetic measurement device is adjustable, effectively avoiding the problem of the occlusion of the camera by the sensor cable during optical scanning.

[0083] The purpose of accurate matching is achieved through sampling consistency registration (SAC-IA) and normal distribution transformation (NDT), the shortcoming that the traditional nearest point iteration method (ICP) registration has a relatively high requirement for coarse registration and is easy to fall into local optimization is avoided, and the requirement for coarse registration is reduced.

[0084] The application discloses a wearable brain magnetic measurement device and MRI registration method, which has the characteristics of automation, high precision and strong adaptability, and is suitable for any actual use scene.

[0085] In the specification, each embodiment is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same or similar parts of each embodiment can be referred to each other.

[0086] In the specification, the principle and implementation mode of the application are described by applying specific examples, and the above embodiment is only used to help understand the method and core idea of the application; meanwhile, for the general technical personnel in the field, the specific implementation mode and application range of the application will be changed according to the idea of the application. In conclusion, the content of the specification should not be understood as the limitation of the application.

Claims

1. A method for registering a magnetoencephalogram (MEG) measuring device with an MRI scanner, characterized in that, include: Acquire head images of a subject wearing a magnetoencephalography (MEG) measuring device; the MEG measuring device includes a MEG helmet body, an identification module, and a sensor array; both the sensor array and the identification module are disposed on the upper surface of the MEG helmet body; the MEG helmet body is worn on the subject's head; the sensor array is used to detect the extracranial magnetic field generated by the current in brain neurons; the identification module is used to locate the sensor array; the identification module is a cross-shaped cylinder, a T-shaped cuboid, or a cone; The subject's head image was converted into a point cloud to obtain a scene point cloud. The point cloud of the scene and the point cloud of the constructed identification module model are subjected to corresponding point clustering processing to identify the identification module of the magnetoencephalogram (MEG) measurement device and determine the position coordinates of the identification module. Perform corresponding point clustering processing on the scene point cloud and the model point cloud of the constructed identifier module model to identify the identifier module of the magnetoencephalogram (MEG) measurement device and determine the position coordinates of the identifier module, specifically including: The scene point cloud and the model point cloud are downsampled respectively to extract key points of the scene point cloud and key points of the model point cloud. Based on the key points of the scene point cloud and the key points of the model point cloud, calculate the feature descriptors of the scene key points and the feature descriptors of the model point cloud key points; Based on the scene key point feature descriptor and the model point cloud key point feature descriptor, the corresponding points of the identification module model in the scene are calculated, and the set of corresponding point pairs between the scene descriptor point cloud and the model descriptor point cloud is determined; the scene descriptor point cloud includes scene key point feature descriptors; the model descriptor point cloud includes model point cloud key point feature descriptors. Perform corresponding point clustering processing on the set of corresponding point pairs, identify the corresponding point clusters in the scene point cloud that match the model point cloud, obtain the identification module of the magnetoencephalogram (MEG) measurement device, and determine the position coordinates of the identification module; The position coordinates of the sensor array probe are determined based on the position coordinates of the marking module and the relative positional relationship between the marking module and the sensor array. Acquire three-dimensional MRI images of the subject's head and separate the three-dimensional MRI images of the subject's head into three-dimensional images of the scalp and anatomical structures of the cerebral cortex; MRI facial images were extracted from the scalp 3D images; Based on the position coordinates of the sensor array probe, the MRI facial image is registered with the subject's head image, so that the sensor array and the anatomical structure of the cerebral cortex are in a unified coordinate system. Based on the position coordinates of the sensor array probes, the MRI facial image is registered with the subject's head image, so that the sensor array and the anatomical structures of the cerebral cortex are in a unified coordinate system. Specifically, this includes: The feature descriptors of the subject's head image and the MRI facial image are calculated respectively to obtain the head image feature descriptor and the facial image feature descriptor; Based on the head image feature descriptor and the facial image feature descriptor, a sampling consistency registration algorithm is used to coarsely register the subject's head image and the MRI facial image to obtain a first coordinate transformation matrix; The first coordinate transformation matrix is ​​subjected to a normal distribution transformation to obtain the second coordinate transformation matrix; The first coordinate transformation matrix and the second coordinate transformation matrix are applied to the position coordinates of the sensor array probe to complete the registration of the MRI facial image and the subject's head image, so that the sensor array and the anatomical structure of the cerebral cortex are aligned in the same coordinate system.

2. The method for registering the magnetoencephalogram (MEG) measurement device with an MRI scanner according to claim 1, characterized in that, The corresponding point clustering processing method is either the Hough voting algorithm or the geometric consistency clustering algorithm.

3. The method for registering the magnetoencephalogram (MEG) measurement device with an MRI scanner according to claim 1, characterized in that, Based on the key points of the scene point cloud and the key points of the model point cloud, feature descriptors for the scene key points and the model point cloud key points are calculated using features such as orientation histogram, point histogram, angle histogram, normal alignment radial feature, moment of inertia feature, or eccentricity feature.

Images