Head positioning and path measurement method for non-navigated rTMS

Abstract

The present application belongs to the technical field of target positioning, and in particular relates to a scalp positioning and path measurement method for non-navigation rTMS, which constructs a continuous scalp shell layer that can be used for geometric calculation on an individual MRI; establishes a stable three-axis coordinate system based on a nasion point, an inion point, and left and right ear points; solves an internal center point under multiple constraint conditions as a geometric anchor point for the construction of a measurement plane; calculates the real scalp path length by means of "finding the intersection of a plane and the scalp surface to obtain a curve, and then finding the arc length of the curve", and outputs visual and numerical results; and thereby significantly improves the geometric consistency and repeatability of target positioning and measurement of non-navigation rTMS.

Description

Technical Field

[0001] This invention belongs to the field of target localization technology, and in particular to a scalp localization and path measurement method for non-navigation rTMS. Background Technology

[0002] Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neuromodulation technique widely used in the clinical treatment of mental illnesses (such as depression and anxiety) and neurological diseases (such as Parkinson's disease and post-stroke rehabilitation) as well as in cognitive neuroscience research. The efficacy of rTMS is highly dependent on the precise localization of the stimulation target. It usually requires mapping the target cortical region (such as the dorsolateral prefrontal cortex, motor cortex, etc.) onto the scalp surface to determine the placement of the coil and its geometric relationship with key anatomical landmarks on the scalp (such as the root of the nose, external occipital protuberance, and left and right tragus points).

[0003] With an optical navigation system, the aforementioned mapping and measurement can be performed by the navigation system with millimeter-level accuracy. However, optical navigation systems are expensive, complex to operate, space-consuming, and require specialized operators, which greatly limits their widespread application in routine clinical settings and primary healthcare institutions. Therefore, navigation-free positioning methods remain the mainstream approach in current rTMS clinical practice.

[0004] Currently, commonly used non-navigational localization methods include the 10–20 EEG system localization method, the empirical proportion method, or tape measure measurement, such as the non-navigational individual rTMS function-specific target localization method disclosed in the invention patent application number CN202411874788.9.

[0005] However, existing navigation-free methods typically suffer from the following shortcomings: 1) The scalp is a curved surface, and using linear Euclidean distance or simple proportions to replace the actual scalp arc length leads to systematic errors; 2) Significant individual head shape differences, such as variations in auricle, occipital region morphology, and scalp thickness, amplify positioning errors and result in insufficient repeatability; 3) The lack of stable internal reference points or a unified coordinate system makes it difficult to achieve consistent measurements by different operators and at different time points; 4) Existing MRI segmentation procedures often focus on brain tissue / skull, failing to adequately extract the "scalp shell layer suitable for scalp geometry measurement." Segmentation results often exhibit discontinuities, holes, and unstable thickness on the scalp's outer surface, making them unsuitable for subsequent surface intersection and arc length calculations. Therefore, there is an urgent need for a unified method that can automatically obtain a continuous scalp shell layer on MRI, construct a reproducible head coordinate system, determine a stable internal center point, and calculate the actual path length based on the scalp surface, thereby supporting target localization and standardized measurement in navigation-free rTMS. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a scalp localization and path measurement method for non-navigation rTMS, which solves the problems of insufficient repeatability, inaccurate scalp surface measurement, and lack of unified geometric reference in the existing non-navigation rTMS target localization.

[0007] To solve the above problems, the technical solution adopted by the present invention is: a scalp localization and path measurement method for non-navigation rTMS, comprising the following steps: S1. Obtain three-dimensional magnetic resonance images of the subject's head, preprocess the images, and construct a continuous triangular mesh surface of the scalp. S2. Construct an orthogonal coordinate system for an individual's head based on anatomical landmarks, including the root of the nose, external occipital protuberance, left ear point, and right ear point; S3. Solve for the internal center point inside the scalp shell; S4. Construct a measurement plane based on the internal center point and any two points on the surface of the scalp triangular mesh. Calculate the curve segment where the measurement plane intersects the surface of the scalp triangular mesh, and then calculate the arc length of the curve segment as the actual path length of the scalp.

[0008] Further, in step S1, preprocessing the image and constructing a continuous scalp triangular mesh surface specifically includes: Denoising, bias field correction, and intensity normalization are performed on structural magnetic resonance images; Gaussian smoothing is performed on the preprocessed image, and soft tissue candidate masks are obtained using adaptive thresholding segmentation; The candidate masks are sequentially subjected to morphological opening, closing, hole filling and connected component filtering to obtain continuous shell candidate masks. Multiple candidate masks are generated using different parameters, and the candidate masks are fused at the voxel level to obtain a stable shell mask; Isosurface extraction is performed on the stable shell mask to obtain a triangular mesh on the scalp surface.

[0009] Furthermore, voxel-level voting or average fusion is performed on the candidate masks: , Mscalp is a stable shell mask. Let I be the candidate mask, and r be the voting threshold, with a value ranging from 0 to 1. <r<1。

[0010] Further, step S2 includes: Acquire or mark four anatomical landmarks on the head: nasal root point (N), external occipital protuberance (I), left ear point (L), and right ear point (R); Define the unit vector in the forward and backward directions: a0 = (IN) / ||IN||; Project the left and right direction vectors b=RL onto a plane perpendicular to a0: , , We obtain a single vector b0 in the left and right directions; By orthogonally complementing and normalizing the space, we obtain the unit vector b0 in the left and right directions: Calculate the unit vector c0 in the up-down direction: c0 = a0 b0; This allows us to construct an orthogonal coordinate system (a0, b0, c0) for the individual head.

[0011] Furthermore, in step S3, solving for the internal center point specifically includes: Define the midpoint m of the left and right ears LR =(L+R) / 2, midpoint of the nasal occiput m NI =(N+I) / 2; Construct two constraint planes: The midpoints of the left and right ears and the single vectors in the left and right directions define the sagittal correlation plane: Πsag: (xm LR )^T b0 = 0, The coronal correlation plane is defined by the midpoint of the nasal occiput and the unit vector in the anterior-posterior direction: Πcor:(xm NI )^T a0=0; Define a candidate domain Ω within the scalp shell and solve for the objective function: The internal center point C is obtained. * , Where dist(c,l) is the distance from point C to line l, and dist(c,π) is the distance from point C to plane π. l λπ are preset weight hyperparameters.

[0012] Further, step S4 includes: For any two points Ps and Pt, construct a path through Ps, Pt, and C. * Measurement plane Πst: (xC * )^T nst=0, where nst=(Ps- C * ) (Pt-C) * ), is the normal vector of the plane Πst; Calculate the curve Γ where the measurement plane Πs intersects the surface of the scalp triangular mesh, and extract the curve segment γst connecting Ps and Pt from the curve Γ; Discretely sample the curve segment γst to obtain a discrete point sequence {P0, P1, ..., PN}, where P0 = Ps and PN = Pt. Calculate the arc length Lst: ; Output path 3D visualization, measurement plane intersection diagram and path length value.

[0013] The beneficial effects of this invention are: 1. Improve the continuity and geometric integrity of scalp shell extraction: Through multi-parameter fusion strategy and morphological repair, a scalp shell without pores and with stable thickness is obtained, providing a reliable data foundation for subsequent geometric calculations.

[0014] 2. Establish a robust individual head coordinate system: Construct an orthogonal coordinate system based on four key anatomical landmarks to eliminate the influence of head pose changes and provide a unified geometric reference framework for cross-individual comparisons and longitudinal tracking.

[0015] 3. Introduce internal center points as stable geometric anchor points: By solving for internal center points through multi-constraint optimization, the construction of the measurement plane is freed from excessive dependence on the local shape of the scalp, significantly improving the consistency of measurement results.

[0016] 4. Achieve precise scalp surface path measurement: The actual surface curve is obtained by intersecting the plane and scalp, and then the arc length is calculated, avoiding the systematic errors caused by traditional straight-line distance or tape measure measurement, resulting in higher accuracy. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram of scalp shell extraction; Figure 3 This is a schematic diagram of the head coordinate system construction; Figure 4 This is a schematic diagram for solving the internal center point problem; Figure 5 This is a schematic diagram illustrating the calculation of the scalp curve path length; Figure 6 This is a three-view overlay diagram of the scalp shell extraction results, with a local magnification shown to demonstrate the shell continuity and thickness stability. Figure 7 It is a three-dimensional visualization of the geometric relationship between the head coordinate system and the center point, showing the nasal root point N, external occipital protuberance I, left ear point L, right ear point R, coordinate axes a0, b0, c0 and constraint plane; Figure 8 This is a schematic diagram of the signed distance function (SDF) in the measurement plane and a schematic diagram of the path extraction results; Figure 9This is a 3D visualization and length output example of standard paths (Nasion, Inion, Left ear, Right ear): showing the spatial position of the intersection line on the scalp surface and the calculated path length value. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] The scalp localization and path measurement method for non-navigation rTMS of the present invention, such as Figure 1 As shown, it includes the following steps: S1, such as Figure 2 As shown, three-dimensional magnetic resonance images of the subject's head were acquired, the images were preprocessed, and a continuous triangular mesh surface of the scalp was constructed.

[0020] Specifically, three-dimensional structural magnetic resonance imaging (MRI) images of the subject's head are acquired, and the structural MRI images are then denoised, bias field corrected, and intensity normalized. Denoising, bias field correction, and intensity normalization are all conventional processing methods that can be performed using existing technologies.

[0021] Gaussian smoothing is performed on the preprocessed image, and adaptive thresholding segmentation is used to obtain the soft tissue candidate mask M0. Gaussian smoothing and adaptive thresholding segmentation can be performed using existing techniques.

[0022] The soft tissue candidate mask M0 is subjected to morphological opening, closing, and hole filling operations sequentially to remove fine connections and noise points, and to fill small holes and discontinuous areas on the scalp surface. Then, connected component filtering is used to retain the largest connected structure corresponding to the head shell, resulting in a continuous shell candidate mask M1. Existing techniques can be used for morphological opening, closing, hole filling, and connected component filtering.

[0023] Multiple candidate masks are generated using different parameters. Candidate masks are typically generated by combining the following three types of parameters: smoothing parameters (noise suppression, boundary stabilization): Gaussian smoothing scale parameter sigma_k; adaptive threshold parameter Delta; and morphological filtering radius r. By performing Gaussian smoothing, adaptive thresholding, and morphological opening and closing operations on the preprocessed image using various combinations of parameters, multiple sets of candidate masks are obtained.

[0024] Voxel-level fusion of multiple candidate masks yields a stable shell mask, Mscalp. Specifically, this is achieved by voxel-level averaging of the candidate masks. , The stable shell mask Mscalp is obtained, where I is the indicator function and r is the voting threshold, with a value ranging from 0 to 1. <r<1。

[0025] Isosurface extraction was performed on the stable shell mask Mscalp to obtain the triangular mesh S on the scalp surface: S = {(Vm, fn)}, where Vm ∈ R 3 , where is the coordinates of the vertex; fn = (a, b, c), is the nth triangular facet, where a, b, c are the vertex numbers used to indicate that the facet is composed of vertices Va, Vb, and Vc.

[0026] Figure 2 The superposition of shells in coronal / sagittal / axial positions is shown to prove that the shell boundaries are continuous and suitable for subsequent intersection calculations.

[0027] S2, such as Figure 3 As shown, an orthogonal coordinate system (a0, b0, c0) for an individual's head is constructed based on anatomical landmarks, including the root of the nose, external occipital protuberance, left ear point, and right ear point.

[0028] Specifically, four anatomical landmarks are manually acquired or calibrated in the three-dimensional image space: the root of the nose (N), the external occipital protuberance (I), the left ear (L), and the right ear (R).

[0029] The line connecting the nasal root point N and the external occipital protuberance I is used as the anterior-posterior direction line, and the line connecting the left ear point L and the right ear point R is used as the lateral direction line.

[0030] Define the unit vector in the forward and backward directions: a0 = (IN) / ||IN||; Project the left and right direction vectors b=RL onto a plane perpendicular to a0: , Calculate the single vector in the left and right directions ; Calculate the unit vector c0 in the up-down direction: c0 = a0 b0; This allows us to construct an orthogonal coordinate system (a0, b0, c0) for the individual head.

[0031] This coordinate system provides a unified reference for the stable definition of the measurement plane and cross-individual comparisons, such as... Figure 7 As shown in A and B.

[0032] S3, such as Figure 4 As shown, stable geometric anchor points are constructed inside the scalp shell, and the internal center point C is solved. * .

[0033] To obtain stable geometric anchor points required for the measurement plane construction, a candidate domain Ω is defined inside the scalp shell: First, a symbolic distance field Φ(x) is constructed based on the stable shell mask Mscalp, and the candidate domain Ω = {x|Mscalp(x) = 1, Φ(x) ≤ -d0} is defined, where d0 is the shrinkage threshold used to ensure that the candidate points are far away from the shell boundary.

[0034] Define the midpoint m of the left and right ears LR =(L+R) / 2, midpoint of the nasal occiput m NI =(N+I) / 2; Construct two constraint planes: The midpoints of the left and right ears and the single vectors in the left and right directions define the sagittal correlation plane: Πsag: (xm LR )^T b0 = 0, The coronal correlation plane is defined by the midpoint of the nasal occiput and the unit vector in the anterior-posterior direction: Πcor:(xm NI )^T a0=0; The candidate points are located within the candidate domain Ω. Solve the optimization objective function: The internal center point C is obtained. * , Where dist(c,l) is the distance from point C to line l, and dist(c,π) is the distance from point C to plane π. l λπ are preset weight hyperparameters, with equal weights by default. When a marker point or its derived geometric constraints (lines / planes) becomes unstable (e.g., large deviations in manual annotation or unreliable constraint geometry due to local shell noise), the corresponding weight is reduced. The optimization variable C is located in the three-dimensional Euclidean space R. 3 The coordinates are represented in the figure, but to ensure internality and stability, the feasible solution is further restricted to the candidate domain Ω(R) determined by the safety distance parameter R.

[0035] The optimized C * Used as a stable reference point for subsequent measurement planes.

[0036] The internal center point C* is a stable geometric anchor point inside the scalp shell. While satisfying the internal safety distance and midline symmetry constraints, it remains robust to individual head shape differences and shell noise. It can also achieve the optimal trade-off between symmetry, internality and landmark reliability through weight adjustment, thus providing a unified and reproducible benchmark for subsequent measurement plane and scalp path calculation.

[0037] The technical advantage of this optimization lies in shifting the "measurement planar structure" from directly relying on the local shape of the scalp to relying on a stable internal reference point, thereby improving the consistency of remeasurements (e.g., Figure 7(As shown in Part A of the middle section).

[0038] S4, such as Figure 5 As shown, a measurement plane is constructed based on the internal center point and any two points on the surface of the scalp triangular mesh. The curve segment intersecting the measurement plane with the surface S of the scalp triangular mesh is calculated, and then the arc length of the curve segment is calculated as the actual path length of the scalp.

[0039] Specifically, for any two points Ps and Pt, construct a path through Ps, Pt, and C. * Measurement plane Πst: (xC * )^T nst=0, where nst=(Ps- C * ) (Pt-C) * ), is the normal vector of the plane Πst.

[0040] Calculate the curve Γ where the measurement plane Πs intersects the scalp triangular mesh surface S, and extract the curve segment γst connecting Ps and Pt from the curve Γ; Discrete sampling is performed on the curve segment γst to obtain the discrete point sequence {P0, P1, ..., PN}, where P0 = Ps and PN = Pt.

[0041] The arc length of a curve is defined as: , In discrete implementation, the curve sampling points approximate: ; Output path 3D visualization, measurement plane intersection diagram and path length value. Figure 8 Provide the SDF zero level set and extraction path within the measurement plane; Figure 9 Examples of 3D visualization and length output for four standard paths are provided. The output actual scalp path lengths can be used for coil placement distance constraints in unguided rTMS, stimulation point standardization reports, etc.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for scalp localization and path measurement in non-navigation rTMS, characterized in that, Includes the following steps: S1. Obtain three-dimensional magnetic resonance images of the subject's head, preprocess the images, and construct a continuous triangular mesh surface of the scalp. S2. Construct an orthogonal coordinate system for an individual's head based on anatomical landmarks, including the root of the nose, external occipital protuberance, left ear point, and right ear point; S3. Solve for the internal center point inside the scalp shell; S4. Construct a measurement plane based on the internal center point and any two points on the surface of the scalp triangular mesh. Calculate the curve segment where the measurement plane intersects the surface of the scalp triangular mesh, and then calculate the arc length of the curve segment as the actual path length of the scalp.

2. The scalp localization and path measurement method for non-navigation rTMS as described in claim 1, characterized in that, In step S1, the image is preprocessed, and a continuous scalp triangular mesh surface is constructed, specifically including: Denoising, bias field correction, and intensity normalization are performed on structural magnetic resonance images; Gaussian smoothing is performed on the preprocessed image, and soft tissue candidate masks are obtained using adaptive thresholding segmentation; The candidate masks are sequentially subjected to morphological opening, closing, hole filling and connected component filtering to obtain continuous shell candidate masks. Multiple candidate masks are generated using different parameters, and a stable shell mask is obtained by voxel-level fusion of the candidate masks. Isosurface extraction is performed on the stable shell mask to obtain a triangular mesh on the scalp surface.

3. The scalp localization and path measurement method for non-navigation rTMS as described in claim 2, characterized in that, Voxel-level voting or average fusion of candidate masks: ， Mscalp is a stable shell mask. Let I be the candidate mask, and r be the voting threshold, with a value ranging from 0 to 1. <r<1。 4. The scalp localization and path measurement method for non-navigation rTMS as described in claim 1, characterized in that, Step S2 includes: Acquire or mark four anatomical landmarks on the head: nasal root point (N), external occipital protuberance (I), left ear point (L), and right ear point (R); Define the unit vector in the forward and backward directions: a0 = (IN) / ||IN||; Project the left and right direction vectors b=RL onto a plane perpendicular to a0: ， Calculate the single vector in the left and right directions ; Calculate the unit vector c0 in the up-down direction: c0 = a0 b0; This allows us to construct an orthogonal coordinate system (a0, b0, c0) for the individual head.

5. The scalp localization and path measurement method for non-navigation rTMS as described in claim 4, characterized in that, Step S3, solving for the internal center point specifically includes: Define the midpoint m of the left and right ears LR =(L+R) / 2, midpoint of the nasal occiput m NI =(N+I) / 2; Construct two constraint planes: The midpoints of the left and right ears and the single vectors in the left and right directions define the sagittal correlation plane: Πsag: (xm LR )^T b0 = 0, The coronal correlation plane is defined by the midpoint of the nasal occiput and the unit vector in the anterior-posterior direction: Πcor:(xm NI )^T a0=0; Define a candidate domain Ω within the scalp shell and solve for the objective function: The internal center point C is obtained. * , Where dist(c,l) is the distance from point C to line l, and dist(c,π) is the distance from point C to plane π. l λπ are preset weight hyperparameters.

6. The scalp localization and path measurement method for non-navigation rTMS as described in claim 1, characterized in that, Step S4 includes: For any two points Ps and Pt, construct a path through Ps, Pt, and C. * Measurement plane Πst: (xC * )^T nst=0, where nst=(Ps- C * ) (Pt-C) * ); Calculate the curve Γ where the measurement plane Πs intersects the surface of the scalp triangular mesh, and extract the curve segment γst connecting Ps and Pt from the curve Γ; Discretely sample the curve segment γst to obtain a discrete point sequence {P0, P1, ..., PN}, where P0 = Ps and PN = Pt. Calculate the arc length Lst: ； Output path 3D visualization, measurement plane intersection diagram and path length value.

Images