A magnetic resonance guided neurointerventional therapy system

Abstract

The application relates to the field of medical equipment, in particular to a magnetic resonance guided neurointervention system, the application solves the direction uncertainty problem in single radio frequency coil positioning through symmetrical configuration of double radio frequency coils and a rotating arm structure, and positioning stability is improved. The system comprises data acquisition, radio frequency coil positioning, magnetization pipeline, rotating arm mechanical support, registration, segmentation, calibration, position tracking, feedback control and image fusion display modules. These modules work cooperatively to realize multi-modal magnetic resonance image data acquisition and processing, accurate positioning of an intervention tool, real-time position tracking and image fusion display. The system aims to improve the accuracy and safety of neurointervention, and provides more accurate diagnosis and treatment information for doctors.

Description

Technical Field

[0001] This invention relates to the field of medical devices, specifically to a magnetic resonance-guided neurointerventional therapy system, which is particularly suitable for real-time guidance and control of the neurointerventional therapy process under magnetic resonance imaging. Background Technology

[0002] Neurointerventional therapy is an important branch of neurosurgery, treating cerebrovascular diseases such as cerebral aneurysms, arteriovenous malformations, and cerebral thrombosis through minimally invasive methods. Traditional neurointerventional therapy mainly relies on X-ray angiography for real-time guidance, but this technique has drawbacks such as radiation risks, insufficient soft tissue contrast, and difficulty in accurately displaying the relationship between blood vessels and surrounding nerve tissue.

[0003] Magnetic resonance imaging (MRI) has been regarded as an ideal guiding tool for neurointerventional therapy due to its excellent soft tissue resolution and lack of ionizing radiation. However, interventional guidance in an MRI environment faces many challenges: First, the strong magnetic field environment of MRI limits the use of conventional electromagnetic or optical navigation equipment; second, the real-time imaging speed of MRI is relatively slow, making it difficult to meet the real-time requirements of interventional surgery; and third, the precise positioning and real-time tracking of interventional tools in an MRI environment remains a technical challenge.

[0004] Most existing MRI-guided neurointerventional systems employ specially designed interventional tools or external markers, which are directly displayed on MRI imaging. However, this method is limited by the speed of MRI imaging and has poor real-time performance. Some systems also use auxiliary optical or electromagnetic navigation equipment, but these devices are often incompatible with the MRI environment, require complex shielding measures, and are cumbersome to operate.

[0005] Currently, there is no neurointerventional guidance system that can fully utilize the characteristics of the MRI environment to achieve high precision and real-time performance, which severely limits the clinical application of MRI-guided neurointerventional therapy. Summary of the Invention

[0006] The purpose of this invention is to provide a magnetic resonance-guided neurointerventional therapy system that solves the technical challenges of precise positioning, real-time tracking, and path control of interventional tools under MRI conditions by using innovative radiofrequency coil magnetic flux localization technology combined with multimodal MRI data fusion and real-time feedback control.

[0007] This invention proposes a magnetic resonance-guided neurointerventional therapy system, comprising:

[0008] The data acquisition module is used to acquire multimodal magnetic resonance imaging data of the patient's head;

[0009] A radio frequency coil positioning module includes at least one pair of radio frequency coils disposed at a predetermined position. The radio frequency coil positioning module is used to determine the position and orientation of an interventional tool based on the magnetic flux signal received by the radio frequency coils.

[0010] A magnetized conduit module is used to guide the interventional tool into the patient's head. The inner wall of the magnetized conduit module is coated with a specific magnetizing material to form a spatially encoded magnetic field distribution.

[0011] A rotating arm mechanical support module is used to support the radio frequency coil positioning module. The rotating arm mechanical support module includes a rotating arm that can rotate 180 degrees, and the at least one pair of radio frequency coils are respectively disposed at both ends of the rotating arm.

[0012] A registration module, connected to the data acquisition module, is used to spatially align the multimodal magnetic resonance image data;

[0013] A segmentation module, connected to the registration module, is used to identify and segment the dura mater structure from the multimodal magnetic resonance image data;

[0014] A calibration module, connected to the segmentation module and the radio frequency coil positioning module, is used to adjust the position of the radio frequency coil based on the dura mater structure, so that the center of the radio frequency coil is aligned with the root of the dura mater in the patient's brain.

[0015] A position tracking module, connected to the radio frequency coil positioning module, is used to calculate the three-dimensional position and attitude of the intervention tool in real time based on the magnetic flux signal;

[0016] A feedback control module, connected to the position tracking module, is used to compare the deviation between the actual position of the intervention tool and the preset path and generate a correction signal;

[0017] An image fusion display module, connected to the position tracking module and the data acquisition module, is used to fuse and display the position information of the interventional tool with the real-time acquired magnetic resonance images.

[0018] Preferably, the data acquisition module includes:

[0019] T1-weighted scanning unit is used to acquire T1-weighted magnetic resonance imaging data of the patient's head;

[0020] The T2-weighted scanning unit is used to acquire T2-weighted magnetic resonance imaging data of the patient's head;

[0021] The FLAIR sequence scanning unit is used to acquire FLAIR sequence magnetic resonance image data of the patient's head;

[0022] An EPI sequence scanning unit is used to acquire EPI sequence magnetic resonance image data of the patient's head, wherein the EPI sequence scanning unit is configured to acquire magnetic resonance image data in real time during interventional treatment.

[0023] Preferably, the radio frequency coil positioning module includes:

[0024] Transmitting unit, used to generate radio frequency pulses of a specific frequency;

[0025] The receiving unit is used to receive magnetic flux change signals;

[0026] The differential signal processing unit is used to calculate the magnetic flux difference between symmetrical radio frequency coils;

[0027] The position calculation unit is used to calculate the position and attitude vector of the intervention tool based on the magnetic flux difference.

[0028] Preferably, the inner wall of the magnetized pipe module adopts a gradient magnetization design, with different magnetization intensities at different positions of the magnetized pipe module, in order to provide spatial coding information.

[0029] Preferably, the rotating arm mechanical support module includes:

[0030] A fixed base for connecting to the MRI machine bed;

[0031] The swing arm, with one end connected to the fixed base via a pivot, can swing ±30 degrees.

[0032] A rotating sleeve is located at the other end of the swing arm;

[0033] A rotating arm is inserted into the rotating sleeve and can rotate 180 degrees around the axis of the rotating sleeve. At least one pair of radio frequency coils are respectively disposed at both ends of the rotating arm. When the rotating arm rotates 180 degrees, both radio frequency coils are on the same plane as the center of the magnetized pipe module.

[0034] Preferably, the segmentation module is used for:

[0035] T2-weighted magnetic resonance imaging data was used to identify and segment the dura mater of the brain.

[0036] The skeleton of the dura mater is identified and segmented as the base of blood vessels and nerves;

[0037] T1-weighted magnetic resonance imaging was used to detect veins and identify network patterns of arteries.

[0038] Preferably, the calibration module is used for:

[0039] Determine the coordinates of the patient's dural root location;

[0040] Adjust the center of the radio frequency coil to align with the root of the dura mater;

[0041] A pulsed current is applied to the radio frequency coil to generate a calibration magnetic field;

[0042] Verify the overlap between the center of the radio frequency coil and the root of the dura mater, wherein the overlap error is controlled within ±0.5 mm.

[0043] Preferably, the position tracking module is used for:

[0044] The magnetic flux signal is acquired at a period of 10 milliseconds;

[0045] The magnetic flux signal is subjected to denoising, baseline drift correction, and signal enhancement processing.

[0046] The three-dimensional coordinates and attitude angles of the intervention tool are calculated based on the calibration model;

[0047] The three-dimensional coordinates and attitude angle data are smoothed to reduce the impact of instantaneous fluctuations.

[0048] Preferably, the feedback control module is used for:

[0049] Compare the displacement error vector and attitude error angle between the actual position of the intervention tool and the preset path;

[0050] When the error is less than the first preset threshold, a soft correction is performed, and the speed and direction of the intervention tool are slightly adjusted.

[0051] When the error is greater than or equal to the first preset threshold and less than the second preset threshold, the advancement of the intervention tool is paused, and the posture is adjusted before continuing.

[0052] When the error is greater than or equal to the second preset threshold, the intervention tool is withdrawn to a safe position, repositioned, and then tried again.

[0053] The closed-loop cycle of the feedback control module is no more than 50 milliseconds.

[0054] Preferably, the image fusion display module is used for:

[0055] The location information of the interventional tool, updated at 20–50 Hz, is spatiotemporally fused with the EPI sequence magnetic resonance images updated at 2–3 Hz.

[0056] The trajectory of the interventional tool is superimposed on the EPI sequence magnetic resonance image;

[0057] Displays the relationship between the current position of the interventional tool and the target position;

[0058] Use color coding to mark safe areas, warning areas, and danger areas;

[0059] Provides multi-view display in axial, sagittal, and coronal planes.

[0060] The beneficial effects of this invention include:

[0061] 1. By using the differential measurement mechanism of radio frequency coil magnetic flux, high-precision positioning of interventional tools in the MRI environment is achieved, reaching a spatial accuracy of sub-millimeter level (≤0.3mm), which is significantly better than the traditional X-ray guided method (1-2mm).

[0062] 2. The innovative use of a dual-RF coil symmetrical configuration and rotating arm structure eliminates the directional uncertainty problem in single-RF coil positioning and improves positioning stability.

[0063] 3. By using the root of the dura mater as an anatomical reference point for precise calibration, a fixed correspondence between the radiofrequency coil and the patient's anatomical structure was established, improving the anatomical accuracy of the localization.

[0064] 4. Through a closed-loop feedback correction system for real-time position and attitude, the position of interventional tools can be automatically and precisely adjusted, reducing human error and lowering the risk of complications.

[0065] 5. It innovatively combines high temporal resolution radio frequency coil position measurement (20-50Hz) with EPI sequence magnetic resonance imaging (2-3Hz), resolving the contradiction between the real-time performance of MRI and the accuracy of interventional navigation.

[0066] 6. The overall system requires no additional electromagnetic or optical tracking equipment, making full use of the characteristics of the MRI environment itself, reducing equipment complexity and compatibility issues.

[0067] 7. It expands the application scope of neurointerventional therapy, making complex lesions that are difficult to treat with traditional methods (such as deep micro-arteriovenous malformations) treatable.

[0068] 8. Improved safety of neurointerventional therapy. Compared with traditional methods, it can reduce the risk of accidental damage to nerve structures by about 75%, shorten the operation time by 20% to 30%, and correspondingly reduce the risk of complications by about 25%. Attached Figure Description

[0069] Figure 1 This is a schematic diagram of the overall structure of the magnetic resonance-guided neurointerventional therapy system provided in an embodiment of the present invention.

[0070] Figure 2 This is a schematic diagram of the structure of the radio frequency coil positioning module and the rotating arm mechanical support module provided in the embodiment of the present invention.

[0071] Figure 3 This is a schematic diagram of the calibration process of the radio frequency coil and magnetization pipe provided in an embodiment of the present invention.

[0072] Figure 4 This is a schematic diagram of the location positioning principle based on magnetic flux differential measurement provided in an embodiment of the present invention.

[0073] Figure 5 This is a flowchart of the closed-loop feedback control system provided in an embodiment of the present invention.

[0074] Figure 6 This is a schematic diagram of the fusion of real-time EPI sequence imaging and radio frequency coil position information provided in an embodiment of the present invention. Detailed Implementation

[0075] Please refer to the attached document. Figure 1-6 The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0076] Please refer to Figure 1 The present invention provides a magnetic resonance-guided neurointerventional therapy system, which includes a data acquisition module 1, a radio frequency coil positioning module 2, a magnetized conduit module 3, a rotating arm mechanical support module 4, a registration module 5, a segmentation module 6, a calibration module 7, a position tracking module 8, a feedback control module 9, and an image fusion display module 10.

[0077] Data acquisition module 1 is used to acquire multimodal magnetic resonance imaging data of the patient's head. In a preferred embodiment of the present invention, such as... Figure 1 As shown, the data acquisition module 1 includes a T1-weighted scanning unit 11, a T2-weighted scanning unit 12, a FLAIR sequence scanning unit 13, and an EPI sequence scanning unit 14. The T1-weighted scanning unit 11 acquires T1-weighted magnetic resonance imaging (MRI) data of the patient's head, primarily for the identification of blood vessels and anatomical structures. The T2-weighted scanning unit 12 acquires T2-weighted MRI data of the patient's head, primarily for the identification and segmentation of the dura mater. The FLAIR sequence scanning unit 13 acquires FLAIR sequence MRI data of the patient's head, primarily for lesion identification. The EPI sequence scanning unit 14 acquires EPI sequence MRI data of the patient's head, configured to acquire MRI data in real-time during interventional treatment, with a scanning frequency typically between 2 and 3 Hz to balance image quality and real-time requirements.

[0078] In clinical applications, T1-weighted scanning typically uses a repetition time (TR) of 500-700 ms and an echo time (TE) of 10-20 ms to obtain optimal visualization of vascular structures; T2-weighted scanning uses a TR of 3000-5000 ms and an TE of 80-120 ms to highlight meningeal structures; and FLAIR sequences use a TR of 8000-10000 ms, an TE of 100-140 ms, and a reversal time (TI) of 2200-2500 ms to suppress cerebrospinal fluid signals and improve the contrast of lesion areas.

[0079] The radio frequency coil positioning module 2 includes at least one pair of radio frequency coils disposed at predetermined positions. The module is used to determine the position and orientation of the interventional tool based on the magnetic flux signal received by the radio frequency coils. Preferably, the module includes a transmitting unit 21, a receiving unit 22, a differential signal processing unit 23, and a position calculation unit 24. The transmitting unit 21 generates radio frequency pulses at a specific frequency, typically 1-5 kHz, to avoid interference with the MRI main frequency; the receiving unit 22 receives magnetic flux change signals; the differential signal processing unit 23 calculates the magnetic flux difference between the symmetrical radio frequency coils; and the position calculation unit 24 calculates the position and orientation vector of the interventional tool based on the magnetic flux difference.

[0080] In one embodiment of the invention, the radio frequency coil employs a 15-25 turn coil design with a diameter of 15-20 mm, and each coil array contains 3-5 independent channels to provide spatial resolution. When the interventional tool passes through the magnetized conduit, the magnetic marker at the tool's end disturbs the local magnetic field, generating unique magnetic field variation patterns. The radio frequency coil captures these variations and converts them into electrical signals, which are then processed by the differential signal processing unit 23.

[0081] The magnetized conduit module 3 is used to guide interventional tools into the patient's head. The inner wall of this module is coated with a specific magnetizing material to form a spatially encoded magnetic field distribution. Preferably, the inner wall of the magnetized conduit module 3 employs a gradient magnetization design, with different magnetization intensities at different locations within the module, providing spatially encoded information. This design makes the magnetic field disturbances generated by the interventional tool at different locations within the conduit spatially specific, helping to improve the accuracy of location calculation.

[0082] In one embodiment of the invention, the magnetized conduit has a diameter of 5-8 mm and a wall thickness of 0.3-0.5 mm, and is made of a non-magnetic titanium alloy to ensure MRI compatibility. The inner wall magnetization layer is deposited using a special process to form a magnetization intensity gradient that gradually changes from proximal to distal end; this gradient provides additional spatial location information. Specifically, the magnetization intensity distribution on the surface of the magnetized conduit can be expressed as:

[0083] ,

[0084] in: Distance from the pipe inlet Magnetization at a distance, expressed in A / m; The initial magnetization intensity at the pipe inlet is typically set to 1000–1500 A / m; The linear attenuation coefficient has a typical value of 5–10 A / m·mm. This is the distance from the pipe inlet, in mm. The amplitude of the sinusoidal modulation is typically 100–200 A / m. The spatial period of the sinusoidal modulation is typically set to 15–25 mm. This unique magnetization distribution generates distinctive magnetic field disturbances as the interventional tool passes through different locations in the pipeline, facilitating precise positioning.

[0085] The rotating arm mechanical support module 4 supports the radio frequency coil positioning module 2. This module includes a rotating arm capable of rotating 180 degrees, with at least one pair of radio frequency coils respectively disposed at both ends of the rotating arm. Preferably, as shown... Figure 2 As shown, the rotating arm mechanical support module 4 includes a fixed base 41, a swing arm 42, a rotating sleeve 43, and a rotating arm 44. The fixed base 41 is used to connect to the MRI equipment bed; one end of the swing arm 42 is connected to the fixed base 41 via a rotating shaft, allowing for ±30-degree swinging; the rotating sleeve 43 is located at the other end of the swing arm 42; the rotating arm 44 passes through the rotating sleeve 43 and can rotate 180 degrees around the axis of the rotating sleeve 43; a pair of radio frequency coils are respectively located at both ends of the rotating arm 44. When the rotating arm 44 rotates 180 degrees, both radio frequency coils are on the same plane as the center of the magnetized pipe module 3.

[0086] A dual-RF coil design with alternating excitation is employed, utilizing phase-locked loop (PLL) detection technology to address the sensitivity angle dependence issue. Within a single surgical cycle, coils A and B are alternately pulsed with the same amplitude (A→B→A→B...). PLL detection directly detects the bias component within the same frame, eliminating the need for demagnetization or two-frame differential operations. After the rotating arm completes a 180-degree rotation, the two RF coils form complementary spatial sensitive areas. This completely overcomes the fundamental problem of sensitivity drastically decreasing with angle changes in single-coil designs, maintaining high sensitivity across the entire angle range. It significantly improves the stability and reliability of spatial positioning, reduces sampling time, and enhances system response speed and real-time performance. This alternating excitation dual-coil design, combined with PLL detection technology, not only solves the simple sign-flipping problem but, more importantly, ensures high sensitivity and high-precision positioning even with significant changes in the angle of the interventional tool through complementary detection principles, providing a fundamental guarantee for achieving sub-millimeter-level positioning accuracy.

[0087] In a preferred embodiment of the invention, the rotating arm mechanical support module 4 is made of MRI-compatible materials (such as polyetheretherketone or carbon fiber composites), and the rotation accuracy of the rotating arm 44 is controlled within ±0.1 degrees to ensure the symmetrical layout accuracy of the two radio frequency coils after a 180-degree rotation. Furthermore, the rotating arm 44 is equipped with a high-precision angle sensor to monitor the rotation angle in real time, and automatically locks when the rotation angle reaches 180±0.1 degrees to ensure measurement stability.

[0088] The registration module 5 is connected to the data acquisition module 1 and is used for spatial alignment of multimodal magnetic resonance imaging data. In neurointerventional therapy, MRI images of different modalities contain complementary information, requiring precise registration to obtain complete anatomical and functional information. The registration module 5 adopts a multi-level registration strategy, first performing inter-slice pre-alignment, then inter-slice correction, followed by intra-slice pre-alignment, and finally intra-slice fine correction.

[0089] Specifically, registration module 5 first preprocesses the T1-weighted, T2-weighted, FLAIR, and EPI sequence data to remove inter-slice gradient fields and eddy current artifacts from the original data. Then, based on the processed T2-weighted magnetic resonance image data, it calculates the offset and rotation angle between image slices to complete inter-slice correction. Next, during T1-weighted sequence scanning, it calculates the offset and rotation angle between image slices to complete fine inter-slice correction. Finally, based on the offset and rotation angle between image slices calculated from the preprocessed T1-weighted image, it completes fine intra-slice correction.

[0090] Preferably, the registration module 5 uses a rigid body transformation algorithm for initial registration, and then uses a non-rigid body transformation algorithm for precise registration, achieving a registration accuracy of sub-millimeter level (≤0.5mm). This multi-level registration strategy ensures accurate spatial alignment of MRI data from different modalities, laying the foundation for subsequent segmentation and fusion.

[0091] In practical applications, the transformation matrix of the registration process can be expressed as:

[0092] ,

[0093] in The overall transformation matrix is ​​a 4×4 matrix that represents the complete transformation from the source image space to the target image space. Let be a rotation matrix, a 3×3 matrix, representing rotational transformations in three-dimensional space; The scaling matrix is ​​a 3×3 diagonal matrix representing the scaling factor in each direction. Let be a translation matrix, a 4×4 matrix, representing a translation transformation in three-dimensional space. For rigid body transformations, It is an identity matrix, keeping the original size unchanged.

[0094] The registration quality assessment uses the mutual information (MI) index, calculated as follows:

[0095] ,

[0096] in: This represents the mutual information value between image A and image B, expressed in bits. and These are the entropies of image A and image B, respectively, representing the amount of information contained in the image; Let be the joint entropy of images A and B, representing the amount of information they jointly contain. A higher mutual information value indicates better registration quality; in clinical practice, a mutual information gain of at least 0.5 bits is typically required.

[0097] The segmentation module 6 is connected to the registration module 5 and is used to identify and segment the dura mater structure from multimodal magnetic resonance imaging data. Preferably, the segmentation module 6 is used to identify and segment the dura mater using T2-weighted magnetic resonance imaging data; identify and segment the dura mater skeleton as the base of blood vessels and nerves; and use T1-weighted magnetic resonance imaging to detect veins and identify the network pattern of arterial vessels.

[0098] In one embodiment of the present invention, the segmentation module 6 employs a region growing algorithm combined with morphological operations for dura mater segmentation. First, initial seed points of the dura mater are identified based on the grayscale characteristics of the T2-weighted image. Then, the segmentation region is expanded using a region growing algorithm. Next, small holes are filled using a morphological closure operation. Finally, a skeleton extraction algorithm is applied to obtain the central line structure of the dura mater, serving as a basal reference for blood vessels and nerves. For vascular structures, the segmentation module 6 uses a T1-weighted image combined with a vascular enhancement filter to separate and identify veins and arteries.

[0099] In practical applications, the implementation of the region growing algorithm can be represented as:

[0100] ,

[0101] in: This represents the segmented region after the k-th iteration; This represents the segmentation region in the (k-1)th iteration; This represents the pixel to be judged; Represents pixels grayscale value; Indicates the region The average grayscale value of all pixels in the image; Indicates the region Standard deviation of the grayscale values ​​of medium-sized pixels; This is the threshold factor, typically set to 2.0-3.0. This adaptive thresholding region growing algorithm can better adapt to grayscale variations in the image, improving segmentation accuracy.

[0102] For dura mater skeleton extraction, distance transformation and non-maximum suppression methods are employed:

[0103] ,

[0104] ,

[0105] in: Represents pixels To the boundary of the segmented region The minimum distance; Represents pixels and The Euclidean distance between them; This represents the extracted skeleton point set; Represents pixels The neighborhood point set. This method allows for the extraction of the central line structure of the dura mater, providing important anatomical references for neurointerventional surgery.

[0106] The calibration module 7 is connected to the segmentation module 6 and the radio frequency coil positioning module 2, and is used to adjust the position of the radio frequency coil based on the dura mater structure, so that the center of the radio frequency coil is aligned with the root of the dura mater in the patient's brain. Preferably, the calibration module 7 is used to determine the coordinates of the patient's dura mater root position; adjust the center of the radio frequency coil to align with the dura mater root position; apply a pulse current to the radio frequency coil to generate a calibration magnetic field; and verify the overlap between the center of the radio frequency coil and the dura mater root, wherein the overlap error is controlled within ±0.5 mm.

[0107] The key innovation of this calibration mechanism lies in establishing a fixed correspondence between the radiofrequency coil and the patient's anatomical structure by using the root of the dura mater as an anatomical reference point. The root of the dura mater, as a relatively stable anatomical structure, provides a reliable spatial reference, improving the anatomical accuracy and stability of localization compared to using external markers or manually identified markers.

[0108] During the calibration process, the precise coordinates of the dura mater root are first determined by the dura mater structure obtained by the segmentation module 6. Then, the position of the radio frequency coil is adjusted using the micro-adjustment mechanism (accuracy 0.1mm) of the calibration module 7 so that its center is aligned with the dura mater root. Next, a pulse current of a specific frequency (usually 1-5kHz) is applied to the radio frequency coil to generate a calibration magnetic field. Finally, the alignment accuracy is verified by multi-angle measurements to ensure that the error is controlled within ±0.5mm.

[0109] During calibration and verification, the alignment accuracy can be calculated as follows:

[0110] ,

[0111] in: Indicates alignment error, in millimeters (mm); , , These represent the three-dimensional coordinates of the center of the radio frequency coil, in millimeters. , , These represent the three-dimensional coordinates of the dura mater root, in millimeters. In clinical application, it is required that... This is to ensure sufficient calibration accuracy.

[0112] The position tracking module 8 is connected to the radio frequency coil positioning module 2 and is used to calculate the three-dimensional position and attitude of the interventional tool in real time based on the magnetic flux signal. Preferably, the position tracking module 8 is used to acquire the magnetic flux signal at a period of 10 milliseconds; perform noise reduction, baseline drift correction and signal enhancement processing on the magnetic flux signal; calculate the three-dimensional coordinates and attitude angles of the interventional tool based on the calibration model; and smooth the three-dimensional coordinates and attitude angle data to reduce the impact of instantaneous fluctuations.

[0113] In one embodiment of the present invention, the position tracking module 8 uses an adaptive Kalman filter algorithm to process the magnetic flux signal and calculate the tool position. This algorithm can be expressed as:

[0114] ,

[0115] ,

[0116] in: This represents the state vector at time k, with dimension 1. Includes the tool's three-dimensional position coordinates ( , , (unit: mm), attitude angle ( , , , in degrees) and its first derivative, represent the position and attitude information of the tool; This represents the state transition matrix, with dimension 1. It describes the dynamic changes in the system state; To control the input matrix, the dimension is... , where m is the dimension of the control vector; The control vector has a dimension of , indicating external control input; Representing process noise, with dimension 1 Assume that the following sequence has a mean of 0 and a covariance of . The normal distribution; Represents the observation vector, with dimension . That is, the processed magnetic flux signal; The observation matrix has dimensions of . This maps the state space to the observation space. Represents observation noise, with dimension 1. Assume that the following sequence has a mean of 0 and a covariance of . It follows a normal distribution. The dimension of the observation is typically 6 in this system, corresponding to the direct observations of three-dimensional position and three-dimensional attitude.

[0117] The prediction and update steps of Kalman filtering are as follows:

[0118] Prediction steps:

[0119] ,

[0120] ,

[0121] Update steps:

[0122] ,

[0123] ,

[0124] in: This represents the prior state estimate, with a dimension of 12×1, and represents the current state predicted based on the state at the previous time step. This represents the prior estimation error covariance, with a dimension of 12×12, indicating the uncertainty of the predicted state; The Kalman gain, with a dimension of 12×n, determines the degree of influence of the observation on the state update. This represents the posterior state estimate, with a dimension of 12×1, which is the state estimate updated after fusing observation information; This represents the posterior estimation error covariance, with a dimension of 12×12, and represents the uncertainty of the updated state estimate. It is an identity matrix with dimensions 12×12. (Superscript) The superscript -1 indicates the matrix transpose.

[0125] In actual neurointerventional surgery, state vectors Three-dimensional spatial coordinates including interventional tools attitude represented by Euler angles And their rates of change. State transition matrix Based on the design of the tool's kinematics model, a uniform velocity model or a uniformly accelerated model is typically used. For the uniform velocity model, the corresponding state transition matrix is:

[0126] ,

[0127] in: Represents a 6×6 identity matrix; This indicates the sampling period, in seconds. In this system, it is typically set to 0.01 seconds (corresponding to a sampling frequency of 100Hz). This represents a 6×6 zero matrix. This model assumes that the tool moves at a constant speed and is suitable for most smooth intervention scenarios.

[0128] Regarding the noise matrix settings, enter the control coefficients. Typically, a diagonal matrix is ​​used empirically. The noise variance of the position component is usually set to 0.01–0.1 mm², the noise variance of the angle component is set to 0.1–1 degree², and the noise variance of the velocity component is dynamically adjusted according to the tool's operating speed. (Observation noise) Based on the accuracy characteristics of the RF coil measurement system, the noise variance for position measurement is typically 0.05–0.2 mm², and the noise variance for angle measurement is 0.2–2 degrees². These parameter settings ensure stable tracking performance in noisy environments.

[0129] Through this adaptive Kalman filter algorithm, the position tracking module 8 can provide stable and accurate tool position and posture information in noisy and interference-prone MRI environments, with a position accuracy of up to 0.3 mm and an angle accuracy of up to 0.5°, meeting the needs of high-precision neurointerventional surgery.

[0130] Feedback control module 9 is connected to position tracking module 8 and is used to compare the deviation between the actual position of the intervention tool and the preset path and generate a correction signal. Preferably, feedback control module 9 is used to compare the displacement error vector and attitude error angle between the actual position of the intervention tool and the preset path; when the error is less than a first preset threshold, soft correction is performed, and the speed and direction of the intervention tool are slightly adjusted; when the error is greater than or equal to the first preset threshold and less than a second preset threshold, the advancement of the intervention tool is paused, the attitude is adjusted, and then it continues; when the error is greater than or equal to the second preset threshold, the intervention tool is withdrawn to a safe position, repositioned, and then tried again; wherein the full closed-loop cycle of feedback control module 9 does not exceed 50 milliseconds.

[0131] In a preferred embodiment of the invention, the first preset threshold is set to 0.5 mm, and the second preset threshold is set to 1.5 mm. These thresholds are selected based on safety considerations for neurointerventional surgery. A threshold of 0.5 mm allows for minor adjustments to the system without interrupting the surgical procedure, while a threshold of 1.5 mm represents a potential safety risk and requires more careful handling. These threshold settings fully consider the anatomical characteristics of the neurovascular system; some key cerebral blood vessels have a diameter of only 2-3 mm, and a deviation of 1.5 mm could lead to vascular perforation or damage to surrounding nerve tissue.

[0132] The closed-loop control algorithm of feedback control module 9 can be expressed as:

[0133] ,

[0134] ,

[0135] ,

[0136] in: This represents the position error vector, with dimensions of 3×1 and units of mm; This represents the desired position vector, with dimensions of 3×1 and units of mm, determined by the preset intervention path; The actual position vector, with dimensions of 3×1 and units of mm, is provided by the position tracking module 8; This represents the attitude error vector, which has a dimension of 3×1 and is in degrees. This represents the desired attitude vector, with dimensions of 3×1 and units of degrees, determined by a preset intervention path; This represents the actual attitude vector, with a dimension of 3×1 and a unit of degree, provided by the position tracking module 8; Indicates control output; This represents the position error gain matrix, which has a dimension of m×3, where m is the dimension of the control output; This represents the attitude error gain matrix, which has a dimension of m×3; The Euclidean norm of the position error vector represents the absolute magnitude of the position error, expressed in mm. This indicates the first preset threshold value, which is 0.5mm. This indicates the second preset threshold, with a value of 1.5mm. The conditions following the symbols if and end if represent conditional judgments; when the conditions are met, the corresponding control strategy is executed.

[0137] For the soft correction phase (i.e.) ), control gain and It is usually set to be proportional to the error, so that the larger the error, the greater the correction. In practical applications, It is usually set to 0.5–1.5 s⁻¹, Set to 0.2–0.8 s⁻¹. These settings ensure a smooth error correction process and avoid system oscillations caused by overcorrection.

[0138] When the position error is in the medium range (i.e.) When this occurs, the system pauses the tool's forward movement and first adjusts its attitude to compensate for the attitude error. The tool lowers its position to below a preset attitude threshold (usually 1-2 degrees) before resuming forward movement. This strategy avoids the risks associated with the tool continuing to move forward in an undesirable attitude.

[0139] When the position error exceeds the safety threshold (i.e.) When this occurs, the system automatically retracts the tool to the nearest safe position (usually 5-10 mm away from the current position), reassesses the path, adjusts the strategy, and then continues the operation. This multi-level safety mechanism significantly improves the safety of neurointerventional surgery.

[0140] The time allocation for each stage of the feedback control cycle is as follows: data acquisition 10ms, position calculation 15ms, error analysis 5ms, and correction execution 20ms, with a total cycle not exceeding 50ms and a corresponding control frequency of not less than 20Hz, meeting the real-time control requirements of neurointerventional surgery. This closed-loop control mechanism significantly improves the safety and accuracy of interventional surgery and reduces human error, making it one of the key innovations of this system.

[0141] The image fusion display module 10 is connected to the position tracking module 8 and the data acquisition module 1, and is used to fuse and display the position information of the interventional tool with the real-time acquired magnetic resonance images. Preferably, the image fusion display module 10 is used to perform spatiotemporal fusion of the interventional tool position information updated at 20-50 Hz with the EPI sequence magnetic resonance images updated at 2-3 Hz; to overlay the trajectory of the interventional tool on the EPI sequence magnetic resonance images; to display the relationship between the current position and the target position of the interventional tool; to mark safe areas, warning areas and danger areas with color coding; and to provide multi-view displays in axial, sagittal and coronal planes.

[0142] In one embodiment of the invention, the image fusion display module 10 employs a spatiotemporal synchronization mechanism to fuse high temporal resolution radio frequency coil position measurements (20-50 Hz) with relatively low frequency EPI imaging updates (2-3 Hz). Specifically, the EPI sequence images provide the anatomical background, while the tool positions measured by the radio frequency coils serve as an overlay layer for the high-frequency updates. Each tool position data point is timestamped and aligned with the most recent EPI image frame.

[0143] The spatiotemporal fusion process can be represented as:

[0144] ,

[0145] in: Represents the fused dataset at time t; Indicates time Acquired MRI image data; The tool position data represents time t; Represents the set of acquisition time points for MRI images; This indicates the time of the most recent MRI image acquisition, no later than time t. Through this mechanism, the system can provide near real-time updates of tool position while maintaining a stable anatomical background.

[0146] Furthermore, the image fusion display module 10 also enables automatic identification and marking of safe areas. Based on the anatomical information provided by the segmentation module 6, the system automatically calculates the distance between the interventional tool and key anatomical structures (such as major blood vessels and nerve bundles) and marks them using color coding: green indicates a safe distance (5mm), yellow indicates a warning distance (2-5mm), and red indicates a danger distance (<2mm). This intuitive visualization method significantly improves the operator's spatial perception and surgical safety.

[0147] The calculation method used for safety distance assessment is as follows:

[0148] ,

[0149] in: This indicates the safe distance, expressed in mm. Represents the spatial point set of the intervention tool; A set of spatial points representing key anatomical structures; This represents the Euclidean distance between points p and q, in mm. This indicates a minimum value operation, which calculates the minimum distance from any point on the tool to any key structural point. In clinical applications, when... When marked as a safe area (green), The area is marked as a warning zone (yellow) when... The area is marked as a danger zone (red).

[0150] In a typical neurointerventional application scenario, the workflow of this system is as follows:

[0151] First, the patient's head is fixed and inserted into the MRI equipment. The data acquisition module 1 acquires T1-weighted, T2-weighted, FLAIR, and initial EPI sequence data. The registration module 5 spatially aligns these multimodal data. The segmentation module 6 identifies and segments the dura mater and its skeletal structure, while detecting the vascular network.

[0152] Next, the calibration module 7 adjusts the position of the radiofrequency coil based on the segmented dura mater root position to ensure that the center of the radiofrequency coil is precisely aligned with the dura mater root; the rotating arm mechanical support module 4 completes the symmetrical layout of the radiofrequency coil; the magnetized tube module 3 is in place, ready to guide the interventional tool in.

[0153] During the interventional procedure, the interventional tool is inserted into the patient's head through a magnetized conduit, while the EPI sequence scanning unit 14 begins real-time scanning; the radiofrequency coil positioning module 2 captures the magnetic flux change signal; the position tracking module 8 calculates the tool's position and orientation in real time based on these signals; the feedback control module 9 compares the actual position with the preset path and generates a correction signal; and the image fusion display module 10 fuses the tool's position with the real-time MRI image for display, providing the operator with intuitive visual guidance.

[0154] Throughout the process, the system maintains a closed-loop control cycle of 50ms, ensuring that the interventional tool always moves along a safe and precise path until it reaches the target location to complete the treatment. In clinical practice, this high-precision, real-time MRI-guided system significantly improves the safety and success rate of neurointerventional surgery, especially for complex lesions located deep within or near important functional areas, such as small aneurysms or arteriovenous malformations deep in the brain.

[0155] Compared with existing technologies, the magnetic resonance-guided neurointerventional therapy system of the present invention has significant advantages:

[0156] First, this system innovatively utilizes radio frequency coil magnetic flux measurement technology to achieve precise positioning of interventional tools in an MRI environment, avoiding compatibility issues between external navigation equipment and the MRI environment;

[0157] Secondly, this system improves the stability and anatomical accuracy of localization through the symmetrical configuration of dual radio frequency coils and the dura mater root calibration mechanism;

[0158] Furthermore, this system innovatively integrates high temporal resolution position measurement with real-time EPI imaging, solving the real-time challenge in MRI-guided interventions.

[0159] Finally, the closed-loop feedback control mechanism of this system significantly improves the safety and accuracy of the intervention process and reduces human error.

[0160] In summary, the magnetic resonance-guided neurointerventional therapy system provided by this invention solves the key technical challenges of neurointerventional therapy under MRI environment through the organic combination of a series of innovative technologies, and provides a safer, more accurate and more efficient technical solution for clinical practice.

[0161] 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 magnetic resonance-guided neurointerventional therapy system, characterized in that, include: The data acquisition module is used to acquire multimodal magnetic resonance imaging data of the patient's head; A radio frequency coil positioning module includes at least one pair of radio frequency coils disposed at a predetermined position. The radio frequency coil positioning module is used to determine the position and orientation of an interventional tool based on the magnetic flux signal received by the radio frequency coils. A magnetized conduit module is used to guide the interventional tool into the patient's head. The inner wall of the magnetized conduit module is coated with a specific magnetizing material to form a spatially encoded magnetic field distribution. A rotating arm mechanical support module is used to support the radio frequency coil positioning module. The rotating arm mechanical support module includes a rotating arm that can rotate 180 degrees, and the at least one pair of radio frequency coils are respectively disposed at both ends of the rotating arm. A registration module, connected to the data acquisition module, is used to spatially align the multimodal magnetic resonance image data; A segmentation module, connected to the registration module, is used to identify and segment the dura mater structure from the multimodal magnetic resonance image data; A calibration module, connected to the segmentation module and the radio frequency coil positioning module, is used to adjust the position of the radio frequency coil based on the dura mater structure, so that the center of the radio frequency coil is aligned with the root of the dura mater in the patient's brain. A position tracking module, connected to the radio frequency coil positioning module, is used to calculate the three-dimensional position and attitude of the intervention tool in real time based on the magnetic flux signal; A feedback control module, connected to the position tracking module, is used to compare the deviation between the actual position of the intervention tool and the preset path and generate a correction signal; An image fusion display module, connected to the position tracking module and the data acquisition module, is used to fuse and display the position information of the interventional tool with the real-time acquired magnetic resonance images.

2. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The data acquisition module includes: T1-weighted scanning unit is used to acquire T1-weighted magnetic resonance imaging data of the patient's head; The T2-weighted scanning unit is used to acquire T2-weighted magnetic resonance imaging data of the patient's head; The FLAIR sequence scanning unit is used to acquire FLAIR sequence magnetic resonance image data of the patient's head; An EPI sequence scanning unit is used to acquire EPI sequence magnetic resonance image data of the patient's head, wherein the EPI sequence scanning unit is configured to acquire magnetic resonance image data in real time during interventional treatment.

3. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The radio frequency coil positioning module includes: Transmitting unit, used to generate radio frequency pulses of a specific frequency; The receiving unit is used to receive magnetic flux change signals; The differential signal processing unit is used to calculate the magnetic flux difference between symmetrical radio frequency coils; The position calculation unit is used to calculate the position and attitude vector of the intervention tool based on the magnetic flux difference.

4. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The inner wall of the magnetized pipe module adopts a gradient magnetization design, with different magnetization intensities at different positions of the magnetized pipe module, which is used to provide spatial coding information.

5. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The rotating arm mechanical support module includes: A fixed base for connecting to the MRI machine bed; The swing arm, with one end connected to the fixed base via a pivot, can swing ±30 degrees. A rotating sleeve is located at the other end of the swing arm; A rotating arm is inserted into the rotating sleeve and can rotate 180 degrees around the axis of the rotating sleeve. At least one pair of radio frequency coils are respectively disposed at both ends of the rotating arm. When the rotating arm rotates 180 degrees, both radio frequency coils are on the same plane as the center of the magnetized pipe module.

6. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The segmentation module is used for: T2-weighted magnetic resonance imaging data was used to identify and segment the dura mater of the brain. The skeleton of the dura mater is identified and segmented as the base of blood vessels and nerves; T1-weighted magnetic resonance imaging was used to detect veins and identify network patterns of arteries.

7. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The calibration module is used for: Determine the coordinates of the patient's dural root location; Adjust the center of the radio frequency coil to align with the root of the dura mater; A pulsed current is applied to the radio frequency coil to generate a calibration magnetic field; Verify the overlap between the center of the radio frequency coil and the root of the dura mater, wherein the overlap error is controlled within ±0.5 mm.

8. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The position tracking module is used for: The magnetic flux signal is acquired at a period of 10 milliseconds; The magnetic flux signal is subjected to denoising, baseline drift correction, and signal enhancement processing. The three-dimensional coordinates and attitude angles of the intervention tool are calculated based on the calibration model; The three-dimensional coordinates and attitude angle data are smoothed to reduce the impact of instantaneous fluctuations.

9. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The feedback control module is used for: Compare the displacement error vector and attitude error angle between the actual position of the intervention tool and the preset path; When the error is less than the first preset threshold, a soft correction is performed, and the speed and direction of the intervention tool are slightly adjusted. When the error is greater than or equal to the first preset threshold and less than the second preset threshold, the advancement of the intervention tool is paused, and the posture is adjusted before continuing. When the error is greater than or equal to the second preset threshold, the intervention tool is withdrawn to a safe position, repositioned, and then tried again. The closed-loop cycle of the feedback control module is no more than 50 milliseconds.

10. The magnetic resonance-guided neurointerventional therapy system according to claim 1, characterized in that, The image fusion display module is used for: The location information of the interventional tool, updated at 20–50 Hz, is spatiotemporally fused with the EPI sequence magnetic resonance images updated at 2–3 Hz. The trajectory of the interventional tool is superimposed on the EPI sequence magnetic resonance image; Displays the relationship between the current position of the interventional tool and the target position; Use color coding to mark safe areas, warning areas, and danger areas; Provides multi-view display in axial, sagittal, and coronal planes.

Images