Individual Structural Target Localization Method for Transcranial Magnetic Stimulation Based on Diffusion-Weighted Imaging

Abstract

The present invention discloses a method for individual structural target localization of transcranial magnetic stimulation based on diffusion-weighted imaging. The present invention adopts a precise localization method for individual structures of transcranial magnetic stimulation based on diffusion-weighted imaging, which can not only solve the limitation that conventional TMS cannot have a specific effect on deep nuclei, but also transmit electrical stimulation from the superficial cortex of the brain to deep nuclei through white matter structural connections in a non-invasive manner, avoiding the adverse risks existing in traditional deep brain electrical stimulation. At the same time, when providing an individualized stimulation target plan for patients, this method uses a white matter structural connection index that is more stable than the functional connection strength calculated by resting-state functional magnetic resonance, providing a precise intervention guidance plan that is more stable and reliable in time and has neuroanatomical significance while solving the problem of strong individual heterogeneity in patients with mental disorders.

Description

Technical Field

[0001] The present invention relates to the field of biomedical image pattern recognition, and particularly to an individual structural target localization method for transcranial magnetic stimulation based on diffusion weighted imaging. Background Art

[0002] The prior art discloses Diffusion Weighted Imaging (DWI) technology, which is an imaging technology that indirectly reflects tissue microstructure by detecting the restricted direction and degree of water molecule diffusion in different physiological tissues. When applied to brain imaging, DWI technology can perform fiber tract tracing based on the characteristics of water molecule diffusion along fiber tracts in white matter. The morphological features of brain fiber tracts traced using DWI technology can to a certain extent reflect the actual running direction and distance of neuron groups in the brain, showing the white matter structure connections between different brain regions and the neural circuits formed by them.

[0003] Transcranial Magnetic Stimulation (TMS) is a non-invasive neuromodulation technique that uses an alternating current in a stimulating coil to generate an alternating magnetic field outside the brain to induce an induced current in the brain, thereby changing the electrical activity of neurons, and then triggering a series of neuroelectrophysiological effects and having a certain impact on an individual's cognition and behavior. Currently, TMS has been approved by the US Food and Drug Administration as a treatment for refractory depression, obsessive-compulsive disorder and other mental disorders, and a large number of clinical studies have shown that TMS can also play a certain therapeutic effect in the intervention of other mental disorders.

[0004] Although TMS has been proven to be an effective neuromodulation means, currently the direct stimulation of TMS can only reach a depth of 2-3 cm below the cortex, and there is a lack of an effective specific stimulation scheme for deep brain nuclei. In contrast, many studies have found that the pathological manifestations of most mental disorders do not only appear in some superficial cortices, but many involve abnormal changes in the connections and neural activities between deep brain nuclei.

[0005] However, existing studies have shown that transcranial magnetic stimulation acting on the superficial cortex of the brain can transmit the stimulation to deep brain nuclei through the white matter structure connection pathway, thereby causing changes in the neural activities of subcutaneous nuclei. Some invasive clinical studies based on deep brain electrical stimulation have also shown that when the stimulation site is selected on the white matter connection pathway between brain regions, the impact on deep brain regions is the most significant. Therefore, we need to adopt individualized stimulation targets to transmit the stimulation of TMS on the superficial cortex to the target nuclei in the deep brain through specific white matter structure connections, so as to achieve accurate and effective intervention and treatment effects.

[0006] Currently, the commonly used transcranial magnetic stimulation (TMS) in clinical practice generally adopts a unified standard target, lacking an individualized precise stimulation plan. For some mental disorders with strong individual heterogeneity, it cannot produce obvious intervention effects on all patients. At present, the stimulation targets guided by resting-state functional magnetic resonance internationally are limited by the time-varying nature of functional magnetic resonance images and the abstract nature of the concept of functional connectivity, and a stable, reliable target plan with anatomical and neurophysiological significance cannot be obtained.

[0007] Therefore, based on the existing research situation, the present invention proposes a precise positioning method for individual structural targets of transcranial magnetic stimulation based on diffusion-weighted imaging. Summary of the Invention

[0008] The object of the present invention is to provide a precise positioning method for individual structural targets of transcranial magnetic stimulation based on diffusion-weighted imaging. This method can not only solve the limitation that conventional TMS cannot have a specific impact on deep nuclei, but also transmit electrical stimulation from the superficial cerebral cortex to deep cerebral nuclei via white matter structural connections non-invasively, avoiding the adverse risks existing in traditional deep brain stimulation. At the same time, when providing an individualized stimulation target plan for patients, this method uses a white matter structural connection index that is more stable than the functional connection strength calculated by resting-state functional magnetic resonance, providing a precise intervention guidance plan that is more stable and reliable in time and has neuroanatomical significance while solving the problem of strong individual heterogeneity in patients with mental disorders.

[0009] To achieve the above object, the present invention is implemented according to the following technical solution:

[0010] The present invention includes the following steps:

[0011] S1: Collect the diffusion-weighted image and high-resolution T1-weighted structural image of the patient, preprocess the patient's diffusion-weighted image, and register the preprocessed diffusion-weighted image to the T1 image space so that the corresponding positions of the registered diffusion image space and the T1 space are the same positions of the subject's brain.

[0012] S2: Preprocess the patient's T1 structural image and perform T1 structural segmentation to obtain the individualized target nuclei and cortical surface stimulation brain area templates during transcranial magnetic stimulation of the subject.

[0013] S3: For the preprocessed diffusion-weighted image, estimate the probability density function of the fiber orientation at each voxel, and perform probabilistic fiber tractography in a constrained spherical deconvolution manner to reconstruct the fiber distribution and orientation at each voxel, and trace the white matter fiber bundle trajectory from the stimulation target nucleus to the cortical surface stimulation area based on the individual brain area template segmented in step S2.

[0014] S4: Display the results of the traced white matter fiber bundles, observe and locate the fiber bundle endpoints closest to the cortical surface from multiple angles, and extract the spatial coordinates of the endpoints; construct a spherical template with the fiber bundle endpoint closest to the cortical surface stimulation area as the center, and at the same time invert the brain-stripped image to obtain an individual extracranial space template;

[0015] S5: Use an edge detection algorithm to extract the surface of the cortical stimulation area, construct a spherical template with the fiber bundle endpoint closest to the cortical surface as the center of the sphere, iterate the radius of the sphere, obtain the radius when the sphere is tangent to the extracranial space, then calculate the spatial coordinates of the intersection area between the sphere and the outer surface of the cortical stimulation area, write the spatial coordinates of the intersection area into the preprocessed T1 structural image of the patient, generate a template with individual TMS structural target markers, and import the template with structural target markers into the precise navigation TMS instrument.

[0016] The preprocessing in step S1 includes noise reduction using principal component analysis, followed by removal of Gibbs artifacts, head motion correction, bias field correction, and calculation of the affine transformation matrix from the diffusion-weighted image space to the T1-weighted structural image space.

[0017] The preprocessing and T1 structural segmentation in step S1 use Freesurfer software, where the preprocessing includes motion correction, non-uniform intensity normalization, Talairach transformation calculation, intensity normalization, brain stripping, linear volume registration, CA intensity normalization, CA non-linear volume registration, removal of the neck, skull-stripped registration, CA labeling and statistics, secondary intensity normalization, white matter segmentation, correction of white matter using ASeg (subcortical segmentation), filling of shears, surface subdivision, original surface smoothing, inflation, automatic topology repair, generation of the final surface, secondary smoothing, secondary inflation, sphere mapping, sphere registration, ipsilateral and contralateral surface registration, mapping of mean curvature to the subject, cortical parcellation and statistics, creation of a cortical ribbon template, and mapping of cortical parcellation to ASeg.

[0018] In step S4, the results of the traced white matter fiber bundles are displayed using MRview software, and the ortho view function is used for multi-angle observation to locate and obtain the coordinates of the fiber bundle endpoints closest to the cortical surface; in MATLAB software, a spherical template is constructed with the fiber bundle endpoint closest to the cortical surface stimulation area as the center, and the brain-stripped image is processed by Freesurfer software.

[0019] In step S5, the edge detection algorithm is performed in MATLAB software.

[0020] The beneficial effects of the present invention are:

[0021] The present invention is a method for precisely localizing individual structural targets of transcranial magnetic stimulation (TMS) based on diffusion-weighted imaging. Compared with the prior art, the present invention addresses the problems that conventional TMS cannot specifically stimulate deep brain nuclei and the large individual variability among patients with mental disorders, and proposes a method for precisely localizing individual structural targets of TMS based on diffusion-weighted imaging. This method helps to precisely intervene according to the structural connection characteristics of patients, and provides an individualized intervention plan for the clinical treatment of mental disorders. This method uses diffusion-weighted images to perform fiber tract tracing from the targeted nucleus to the stimulated cortex, and then calculates the regional coordinates of the area closest to the fiber tract based on the traced fiber tract closest to the stimulated cortex for precise stimulation, guiding the stimulation from the superficial cortex to the target nucleus via the fiber tract specifically, and providing a more precise and effective individualized treatment navigation plan for TMS.

[0022] Therefore, based on the existing research status, the present invention adopts a method for precisely localizing individual structural targets of TMS based on diffusion-weighted imaging, which can not only solve the limitation that conventional TMS cannot specifically affect deep brain nuclei, but also transmit electrical stimulation from the superficial cortex of the brain to the deep brain nuclei via white matter structural connections non-invasively, avoiding the adverse risks of traditional deep brain electrical stimulation. At the same time, when providing an individualized stimulation target plan for patients, this method uses a white matter structural connection index that is more stable than the functional connection strength calculated by resting-state functional magnetic resonance imaging, solving the problem of strong individual heterogeneity among patients with mental disorders while providing a more stable, reliable and neurally anatomically significant precise intervention guidance plan in terms of time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic flow diagram of the method for precisely localizing individual structural targets of TMS based on diffusion-weighted imaging in the present invention.

[0024] Figure 2 They are the high-resolution T1-weighted images and diffusion-weighted images of the individual subject to be pre-collected in the present invention. Figure 2 Among them, (a) is the high-resolution T1-weighted image, and (b) is the diffusion-weighted image;

[0025] Figure 3 It is the result obtained after precisely segmenting the individual T1-weighted image in the present invention.

[0026] Figure 4 It is the fiber tract connection trajectory traced from the target nucleus to the cortical stimulation area in the present invention (here, the target nucleus is selected as the amygdala, and the cortical stimulation area is the frontal lobe).

[0027] Figure 5 It is the spatial localization template of the fiber tract endpoint closest to the surface of the stimulated cortex extracted in the present invention.

[0028] Figure 6 It is the extracerebral space template obtained by brain stripping and image inversion of the T1-weighted image in the present invention.

[0029] Figure 7 It is the frontal lobe surface template obtained by using the edge detection algorithm for the segmented frontal lobe template in the present invention.

[0030] Figure 8 It is the schematic diagram of the intersection of the sphere obtained by iterating the radius with the fiber bundle endpoints as the center of the sphere and the frontal lobe surface in the present invention.

[0031] Figure 9 It is the coordinate of the precise structural stimulation target area obtained in the present invention, which is the area on the surface of the target cortex stimulation area closest to the endpoints of the traced fiber bundle. Detailed implementation manners

[0032] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. The schematic embodiments and descriptions of the present invention are used to explain the present invention, but do not limit the present invention.

[0033] As Figure 1 shown: The present invention includes the following steps:

[0034] Step 1: Collect the diffusion-weighted image and high-resolution T1-weighted structural image of the patient, as Figure 2 shown;

[0035] Step 2: Denoise the diffusion-weighted image of the patient using principal component analysis, and then perform preprocessing such as removing Gibbs artifacts, head motion correction, and bias field correction. By calculating the affine transformation matrix from the diffusion-weighted image space to the T1-weighted structural image space, register the preprocessed diffusion-weighted image of the individual to the T1 image space, so that the corresponding positions of the registered diffusion image space and the T1 space at the same coordinates correspond to the same position of the subject's brain;

[0036] Step 3: Use the Freesurfer software to perform preprocessing and T1 structural segmentation processes on the patient's T1 structural image, including motion correction, non-uniform intensity normalization, Talairach transformation calculation, intensity normalization, brain stripping, linear volume registration, CA intensity normalization, CA non-linear volume registration, removing the neck, registering with the skull, CA labeling and statistics, secondary intensity normalization, white matter segmentation, correcting white matter with ASeg, filling shear, surface subdivision, original surface smoothing, inflation, automatic topology repair, generating the final surface, secondary smoothing, secondary inflation, sphere mapping, sphere registration, ipsilateral and contralateral surface registration, mapping the mean curvature to the subject, cortical parcellation and statistics, creating a cortical ribbon template, mapping the cortical parcellation to ASeg, etc. The preprocessing and brain region segmentation results are as Figure 3 shown, and obtain the individualized target nucleus and cortical surface stimulation brain region template for the subject during transcranial magnetic stimulation;

[0037] Step 4: For the preprocessed diffusion-weighted images, by estimating the probability density function of the fiber orientation at each voxel, probabilistic fiber tractography is performed in a constrained spherical deconvolution manner to reconstruct the fiber distribution and orientation at each voxel. Based on the individual brain region template segmented in Step 3, the white matter fiber tract trajectory from the stimulated target nucleus to the cortical surface stimulation area is traced, as Figure 4 shown;

[0038] Step 5: Use the MRview software to display the results of the traced white matter fiber tracts. Use the ortho view to observe and locate the fiber tract endpoints closest to the stimulated cortical surface from multiple angles, and extract the spatial coordinates of the endpoints. The results are as Figure 5 shown;

[0039] Step 6: In the MATLAB software, construct a sphere template with the fiber tract endpoint closest to the cortical surface stimulation area as the center. At the same time, invert the brain-stripped image processed by Freesurfer to obtain the individual extracerebral space template as Figure 6 shown;

[0040] Step 7: In the MATLAB software, use an edge detection algorithm to extract the surface of the cortical stimulation area as Figure 7 shown;

[0041] Step 8: Construct a sphere template with the fiber tract endpoint closest to the cortical surface as the center of the sphere. Iteratively loop through the sphere radius to obtain the radius when the sphere is tangent to the extracerebral space, and then calculate the spatial coordinates of the intersection region between the sphere and the outer surface of the cortical stimulation area as Figure 8 shown;

[0042] Step 9: Write the spatial coordinates of the intersection region into the preprocessed T1 structural image of the patient to generate a template with individual TMS structural target markers as Figure 9 shown;

[0043] Step 10: Import the template with structural target markers into the precise navigation TMS instrument.

[0044] In a preferred embodiment of the present invention, in Step 2, the affine transformation matrix for registering the individual diffusion-weighted image to the high-resolution T1-weighted image is calculated, and the diffusion image is registered into the T1 structural image space.

[0045] In a preferred embodiment of the present invention, in steps five to eight, a sphere is constructed by iteratively looping the radius with the endpoint of the fiber bundle closest to the stimulated cortex as the center of the sphere and inflated until it is tangent to the extracranial space. Then, the spatial coordinates of the intersection region between the sphere and the stimulated cortex are calculated. The coordinates of the intersection region are written as the structural target into the high-resolution T1 structural image to generate an individual TMS precise structural target template. The target coordinates obtained by this process are the positions where the fiber connection from the stimulated cortex to the target nucleus is closest to the superficial cortex. This indicates that when this region is stimulated, the fiber bundle can receive the stimulation located in the superficial cortex to the greatest extent and conduct the induced current to the deep target nucleus.

[0046] The technical solution of the present invention is not limited to the limitations of the above specific embodiments. Any technical deformation made according to the technical solution of the present invention falls within the protection scope of the present invention.

Claims

1. A method for individual structural target localization of transcranial magnetic stimulation based on diffusion weighted imaging, characterized in that, It includes the following steps: S1: Collect the diffusion-weighted imaging and high-resolution T1-weighted structural imaging of the patient, preprocess the diffusion-weighted image of the patient, and register the preprocessed diffusion-weighted image to the T1 imaging space so that the corresponding positions of the same coordinates in the registered diffusion image space and the T1 space are the same positions of the subject's brain; The preprocessing of the diffusion-weighted image of the patient includes noise reduction using principal component analysis, then removing Gibbs artifacts, head motion correction, bias field correction, and calculating the affine transformation matrix from the diffusion-weighted image space to the T1-weighted structural image space; S2: Preprocess the patient's T1 structural imaging and perform T1 structural segmentation to obtain the individualized target nuclei and cortical surface stimulation brain region templates during transcranial magnetic stimulation of the subject; The preprocessing and T1 structural segmentation of the patient's T1 structural imaging use Freesurfer software, where the preprocessing includes motion correction, non-uniform intensity normalization, Talairach transformation calculation, intensity normalization, brain stripping, linear volume registration, CA intensity normalization, CA non-linear volume registration, removing the neck, skull-stripped registration, CA labeling and statistics, secondary intensity normalization, white matter segmentation, correcting white matter with ASeg, filling cuts, surface tessellation, original surface smoothing, inflation, automatic topology repair, generating the final surface, secondary smoothing, secondary inflation, sphere mapping, sphere registration, ipsilateral and contralateral surface registration, mapping the mean curvature to the subject, cortical parcellation and statistics, creating a cortical ribbon template, and mapping the cortical parcellation to ASeg; S3: For the preprocessed diffusion-weighted image, by estimating the probability density function of the fiber orientation on each voxel, perform probabilistic fiber tractography in a constrained spherical deconvolution manner, reconstruct the fiber distribution and orientation on each voxel, and track the white matter fiber bundle trajectory from the stimulation target nucleus to the cortical surface stimulation area based on the individual brain region template segmented in step S2; S4: Display the results of the traced white matter fiber bundle, observe and locate the fiber bundle endpoint closest to the stimulated cortical surface from multiple angles, and extract the endpoint spatial coordinates; construct a sphere template with the fiber bundle endpoint closest to the cortical surface stimulation area as the center, and at the same time invert the brain-stripped image to obtain the individual extracranial space template; S5: Use an edge detection algorithm to extract the surface of the cortical stimulation area, construct a sphere template with the fiber bundle endpoint closest to the cortical surface as the center of the sphere, perform cyclic iteration on the sphere radius, obtain the radius when the sphere is tangent to the extracranial space, then calculate the spatial coordinates of the intersection area between the sphere and the outer surface of the cortical stimulation area, write the spatial coordinates of the intersection area into the patient's preprocessed T1 structural imaging, generate a template with individual TMS structural target markers, and import the template with structural target markers into a precise navigation TMS instrument.

2. The method for individual structural target localization of transcranial magnetic stimulation based on diffusion weighted imaging according to claim 1, wherein: In step S4, the white matter fiber bundle results obtained by display tracking are observed from multiple angles using the orthoview function of MRview software, and the coordinates of the fiber bundle endpoints closest to the cortical surface are located and obtained; in MATLAB software, a spherical template is constructed with the fiber bundle endpoint closest to the cortical surface stimulation area as the center, and the brain-stripped image is processed by Freesurfer software.

3. The transcranial magnetic stimulation individual structural target localization method based on diffusion weighted imaging according to claim 1, characterized in that: In step S5, the edge detection algorithm is performed in MATLAB software.

Images