An adaptive magnetic resonance imaging system and method for weight-bearing knee joint

Abstract

This application provides an adaptive detection system and method for weight-bearing MRI of the knee joint, belonging to the field of MRI detection technology. The system includes: a patient support platform disposed in the lower part of the patient receiving section of the magnet chamber; a sensing module fixedly installed on the inner wall of the magnet chamber; a first drive mechanism and a second drive mechanism both disposed inside the magnet chamber, and capable of driving the anterior half-coil and the posterior half-coil to move in three-dimensional space, respectively; a status feedback module for detecting the real-time position and engagement status of the anterior half-coil and the posterior half-coil; and a main controller connected to the sensing module, the anterior half-coil module, the posterior half-coil module, and the status feedback module. This application achieves automatic and precise positioning and engagement of the knee joint weight-bearing MRI coil, improving imaging quality and diagnostic accuracy, ensuring comparability between different scans, shortening scan preparation time, and optimizing the patient experience.

Description

Technical Field

[0001] This application belongs to the field of magnetic resonance imaging technology, specifically relating to an adaptive magnetic resonance imaging system and method for weight-bearing knee joint. Background Technology

[0002] Accurate diagnosis of knee joint diseases (such as meniscus tears, cartilage wear, and ligament injuries) relies heavily on magnetic resonance imaging (MRI). Traditional knee MRI examinations are typically performed with the patient in a supine position. However, the supine position cannot simulate the actual weight-bearing state of the body when standing and walking, leading to a disconnect between imaging and clinical symptoms. Specifically, when the knee joint is under weight-bearing, occult tears in the posterior trochanter of the meniscus may open due to pressure changes and become visible on imaging. Key diagnostic information such as cartilage compression deformation and the condition of the tibial plateau can only be captured under weight-bearing conditions. This information is crucial for developing surgical plans such as osteotomy and unicompartmental knee replacement.

[0003] To meet this need, weight-bearing knee MRI systems have emerged, allowing patients to undergo scanning while simulating a standing or semi-squatting posture. However, existing weight-bearing MRI technology has significant limitations: the quality of MRI images is highly dependent on the precise alignment of the radiofrequency coil with the scanning site. Since the coil, as a signal receiving component, is directly related to the distance between itself and tissues such as cartilage and the meniscus, the signal-to-noise ratio is determined. Currently, coil positioning relies entirely on the radiologist's visual observation and manual adjustment. This process is cumbersome and inefficient, requiring the technician to repeatedly enter and exit the scanning room for adjustments. This not only significantly reduces equipment cycle time but, more importantly, makes it difficult to guarantee the accuracy of manual positioning. This results in inaccurate repetition of image layers between different scans or between different patients, severely hindering precise imaging comparisons of disease progression or postoperative recovery.

[0004] Therefore, there is an urgent need in this field for a technical solution that can automatically, accurately, and adaptively complete the MRI coil positioning of the knee joint in weight-bearing positions, in order to free up manpower, ensure the repeatability and comparability of image quality, and ultimately improve the accuracy and efficiency of diagnosis. Summary of the Invention

[0005] In a first aspect, embodiments of this application provide a knee joint weight-bearing position magnetic resonance adaptive detection system, including a magnet chamber, a patient support platform, a sensing module, an anterior half-coil module, a posterior half-coil module, a status feedback module, and a main controller;

[0006] The magnet chamber includes two magnet sections, front and rear, forming the patient receiving area; the patient support platform is located at the lower part of the patient receiving area of ​​the magnet chamber.

[0007] The sensing module is fixedly installed on the inner wall of the magnet chamber to collect spatial information of the patient's knee joint;

[0008] The front half-coil module includes a first drive mechanism and a front half-coil;

[0009] The first drive mechanism is located inside the magnet compartment and can drive the front half coil to move forward, backward, left, right and up in three-dimensional space.

[0010] The rear half-coil module includes a second drive mechanism and a rear half-coil;

[0011] The second drive mechanism is located inside the magnet compartment and can drive the rear half coil to move forward, backward, left, right and up in three-dimensional space.

[0012] The status feedback module is used to detect the real-time position and engagement status of the front and rear half coils;

[0013] The main controller is connected to the sensing module, the front half-coil module, the rear half-coil module, and the status feedback module, and is configured as follows:

[0014] The data collected by the sensing module is processed to calculate the optimal imaging center point coordinates of the knee joint;

[0015] Based on the optimal imaging center point coordinates, the first drive mechanism and the second drive mechanism are controlled to move together, so that the front half coil and the rear half coil move to the target position and engage.

[0016] Based on the signal from the status feedback module, the magnetic resonance scan is started after confirming that the front half coil and the rear half coil are properly engaged.

[0017] Furthermore, both the first drive mechanism and the second drive mechanism are three-axis motion platforms composed of X-axis and Y-axis linear modules and Z-axis ball screw module;

[0018] An automatic locking mechanism is provided between the rear half-coil and the front half-coil. The automatic locking mechanism adopts an electromagnetic lock or a mechanical locking mechanism.

[0019] Furthermore, the perception module includes a 3D vision camera unit and a terahertz imaging unit;

[0020] The 3D vision camera unit is fixedly installed on the side wall of the magnet cabin. The field of view of the 3D vision camera unit covers the entire area of ​​the patient receiving part and is used to acquire three-dimensional point cloud data of the patient's lower limbs.

[0021] The terahertz imaging unit includes a transmitter and a receiver, which are fixedly mounted on the side wall of the magnet chamber. The beam of the terahertz imaging unit is directed towards the intended placement area of ​​the knee joint in the patient housing, for transmitting terahertz waves to the intended placement area of ​​the knee joint and receiving reflected signals.

[0022] Furthermore, the status feedback module includes a first position encoder, a second position encoder, a macro sensor, and a pressure sensor;

[0023] The first position encoder is located at the first drive mechanism and is used to provide feedback on the real-time position of the front half coil.

[0024] The second position encoder is located at the second drive mechanism and is used to provide feedback on the real-time position of the rear half coil.

[0025] The macro sensor and pressure sensor are integrated on the mating surface of the front half-coil and / or the rear half-coil;

[0026] The pressure sensor is used to monitor the fastening force. When the fastening force is within the preset threshold range, the main controller determines that the fastening is in place.

[0027] Secondly, embodiments of this application also provide an adaptive magnetic resonance imaging method for weight-bearing knee joints, comprising the following steps:

[0028] S1. Spatial information of the patient's knee joint is collected by a sensing module fixed to the inner wall of the magnet chamber, and the data is fused to calculate the optimal imaging center point coordinates of the knee joint under load.

[0029] S2. Based on the optimal imaging center point coordinates, control the first drive mechanism and the second drive mechanism to move together, drive the front half coil and the rear half coil to the target position and complete the automatic engagement;

[0030] S3. Based on the signal verification of the state feedback module, the engagement status of the front half coil and the rear half coil is verified. After confirming that the engagement is correct, the magnetic resonance scanning system is controlled to execute the scanning sequence.

[0031] Furthermore, the specific steps of step S1 are as follows:

[0032] S11. Scan the patient's lower limbs using a 3D vision camera unit to generate three-dimensional point cloud data, and construct an external contour model of the leg using computer vision algorithms. Combined with the anatomical position of the knee joint, output preliminary knee joint space coordinates. ;

[0033] S12. Terahertz waves are emitted towards the knee joint region through a terahertz imaging unit, and the reflected signals are received. The intensity of the reflected signals is analyzed. The distribution of the internal structures of the knee joint is used to output the optimal imaging point coordinates. ;

[0034] S13. Initial knee joint space coordinates Optimal imaging point coordinates of the internal structures of the knee joint The data is input into the data fusion algorithm to calculate the final optimal imaging center point coordinates. .

[0035] Furthermore, in step S11, preliminary knee joint space coordinates are output based on the anatomical position of the knee joint. The calculation formula is as follows:

[0036]

[0037] in, The coordinates of the i-th point in the three-dimensional point cloud data of the expected area of ​​the knee joint are represented, and N is the total number of points in the three-dimensional point cloud of the expected area of ​​the knee joint.

[0038] In step S12, the optimal imaging point coordinates for the internal structures of the knee joint are determined by finding the maximum value of the terahertz reflected signal intensity. The calculation formula is as follows:

[0039]

[0040] in, This represents the intensity of the terahertz reflected signal received at spatial coordinates (x, y, z);

[0041] In step S13, the data fusion algorithm uses a weighted average method, and its calculation formula is as follows:

[0042]

[0043] in, and The weighting coefficients are dynamically adjusted based on the real-time signal-to-noise ratio of visual data and terahertz data, and satisfy the following conditions: .

[0044] Furthermore, the specific process of constructing the leg's external contour model in step S11 is as follows:

[0045] S111. Denoising and filtering preprocessing are performed on the 3D point cloud data, and the lower limb point cloud clusters are extracted using the region growing segmentation algorithm;

[0046] S112. Based on the random sampling consensus algorithm, the extracted lower limb point cloud is geometrically fitted to construct the external contour model of the leg;

[0047] After receiving the reflected terahertz wave signal in step S12, filtering and noise reduction processing is performed, using a bandpass filter to retain the effective signal in the terahertz frequency band; the transfer function of the bandpass filter is:

[0048] ,when ;

[0049] ,when or ;

[0050] in, This is the lower limit frequency of the terahertz band. This is the upper limit frequency of the terahertz band.

[0051] Furthermore, the specific process of constructing the leg's external contour model in step S111 is as follows:

[0052] The 3D point cloud data was preprocessed by denoising and filtering, and the lower limb point cloud clusters were extracted using a region growing segmentation algorithm.

[0053] Denoising of 3D point cloud data employs a statistical outlier removal filter, specifically including:

[0054] For each point in the 3D point cloud data Calculate the average distance to the k nearest neighbors. ;

[0055] Calculate the mean distance μ and standard deviation σ of the entire point cloud;

[0056] Remove those that meet the conditions The point, where m is an adjustable multiple threshold;

[0057] Step S111 uses a region growing segmentation algorithm to extract the lower limb point cloud clusters, specifically including:

[0058] Calculate the normal vector of each point in the 3D point cloud data. and curvature ;

[0059] Select the point with the smallest curvature as the seed point and begin growth;

[0060] For the current seed point Check neighboring points If neighboring points With the current seed point If the angle between the normal vectors is less than the threshold θ and the curvature difference is less than the curvature difference threshold c, then the neighboring point is merged into the current region and used as a new seed point.

[0061] Repeat the above process until no new points can be merged, thus completing the region growth.

[0062] The specific steps of step S112 are as follows:

[0063] The leg contour is approximated by fitting a cylindrical model using a random sampling consensus algorithm. The parameters of the cylindrical model are the central axis direction vector, a point on the central axis, and the radius.

[0064] Randomly select several points to calculate the initial cylindrical model, calculate the distance from all points in the 3D point cloud data to the initial cylindrical model, and consider points with a distance less than the threshold t as interior points;

[0065] After multiple iterations, the cylindrical model with the most interior points is selected as the fitting result.

[0066] Based on the fitted cylindrical model, the leg axis is determined, and the coordinates of the center point of the knee joint space are estimated by combining the anatomical position of the knee joint. .

[0067] Furthermore, the specific steps of step S2 are as follows:

[0068] S21. Set the optimal imaging center point coordinates The command is converted into a displacement command. First, the first drive mechanism is controlled to move the front half coil to the target position in three-dimensional space. Then, the second drive mechanism is controlled to move the rear half coil in three-dimensional space and a motion path is planned to avoid contact with the patient.

[0069] S22. After both the current half-coil and the rear half-coil have moved to the target position, the automatic fastening mechanism is controlled to perform the fastening action, and the fastening status is monitored by the micro sensor and the pressure sensor;

[0070] When the distance value detected by the macro sensor reaches the preset closing distance threshold, and the force value detected by the pressure sensor is within the preset force threshold range, it is determined that the fastening is in place.

[0071] As can be seen from the above technical solutions, this application has the following advantages:

[0072] The knee joint weight-bearing magnetic resonance adaptive detection system and method provided in this application, by integrating a 3D vision camera and terahertz imaging technology, can accurately perceive the three-dimensional spatial position of the knee joint, ensuring optimized and consistent imaging quality for each scan, and improving the quality and comparability of image data; it shortens the tedious process that originally required technicians to repeatedly enter and exit the scanning room and make manual adjustments to minutes or even seconds, improving equipment turnover rate; it reduces patient preparation and waiting time, and the automated process avoids discomfort that may be caused by improper manual operation; the path planning based on sensor data can completely avoid accidental collisions between mechanical parts and patients, and the force of the latching process is also precisely controlled by the system; it can intelligently adapt to patients of different heights, weights, body types, and different physiological structures of the left and right legs, ensuring the accuracy and consistency of each scan. Attached Figure Description

[0073] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 This is a schematic diagram of the structure of the knee joint weight-bearing position magnetic resonance adaptive detection system of the present invention.

[0075] Figure 2 This is a schematic diagram of the front and rear coils and two drive mechanisms of the present invention.

[0076] Figure 3 This is a schematic diagram of the control of the knee joint weight-bearing position magnetic resonance adaptive detection system of the present invention.

[0077] Figure 4 This is a schematic flowchart of the knee joint weight-bearing position magnetic resonance adaptive detection method of the present invention.

[0078] Among them, 1-patient support platform; 2-main controller; 3-magnet chamber; 4-front half coil; 5-rear half coil; 6-sensing module; 7-status feedback module; 8-first drive mechanism; 9-second drive mechanism. Detailed Implementation

[0079] The various embodiments of this disclosure will be described more fully in the following detailed description of the knee joint weight-bearing magnetic resonance adaptive detection system. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0080] For example, accurate diagnosis of knee joint diseases, such as meniscus tears, cartilage wear, or ligament injuries, relies heavily on magnetic resonance imaging (MRI). Traditional knee MRI scans are typically performed with the patient in a supine position. However, this position cannot simulate the actual weight-bearing state of the body when standing or walking, leading to a disconnect between imaging results and clinical symptoms. Specifically, under weight-bearing conditions, occult tears may occur in the posterior horn of the meniscus, opening due to pressure changes and thus becoming visible on imaging. Furthermore, crucial diagnostic information such as cartilage compression deformation and tibial plateau alignment can only be captured under weight-bearing conditions, and this information is essential for developing surgical plans such as osteotomy or unicompartmental knee replacement.

[0081] To meet this need, weight-bearing knee MRI systems were developed, allowing patients to undergo scanning in a simulated standing or semi-squatting position. However, existing weight-bearing MRI technology has a significant bottleneck: image quality is highly dependent on the precise alignment of the radiofrequency coil with the scanning site. As a signal receiving component, the distance between the coil and tissues such as cartilage and meniscus directly determines the signal-to-noise ratio. Currently, coil positioning relies entirely on the radiologist's visual observation and manual adjustment, a cumbersome and inefficient process. The technician needs to repeatedly enter and exit the scanning room for adjustments, which not only significantly reduces the equipment's efficiency but also makes it difficult to guarantee the accuracy of manual positioning. This results in inaccurate repetition of image layers between different scans or between different patients, severely hindering precise imaging comparisons of disease progression or postoperative recovery.

[0082] Therefore, there is an urgent need in this field for a technical solution that can automatically, accurately, and adaptively perform MRI coil localization of the knee joint in weight-bearing positions. This solution aims to free up manpower, ensure the repeatability and comparability of image quality, and ultimately improve the accuracy and efficiency of diagnosis.

[0083] To address the aforementioned issues, this embodiment provides an adaptive detection system for weight-bearing MRI of the knee joint. Through automated coil positioning and engagement, it improves the imaging quality and repeatability of weight-bearing MRI of the knee joint, while significantly shortening scan preparation time and optimizing the patient experience.

[0084] 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] Please see Figure 1 , Figure 2 and Figure 3 The diagram shown is a schematic of a knee joint weight-bearing position magnetic resonance adaptive detection system in a specific embodiment. The system includes a magnet chamber 3, a patient support platform 1, a sensing module 6, an anterior half-coil module, a posterior half-coil module, a status feedback module 7, and a main controller 2.

[0086] The magnet chamber 3 includes two magnet sections, one at the front and one at the back, forming the patient's accommodating area;

[0087] It should be noted that the magnet chamber 3 is the supporting structure of the entire system. The magnet chamber 3 provides physical protection for each component and ensures the stability of the system operation; it also provides the necessary magnetic field environment for MRI imaging, ensuring the quality and accuracy of the imaging.

[0088] The patient support platform 1 is located in the lower part of the patient receiving section of the magnet chamber 3;

[0089] The patient support table 1 provides a stable support platform to ensure that the patient maintains the correct posture and position during the scanning process;

[0090] The sensing module 6 is fixedly installed on the inner wall of the magnet chamber 3 to collect spatial information of the patient's knee joint;

[0091] It should be noted that the perception module 6 uses a 3D vision camera and terahertz imaging technology to accurately collect spatial information of the patient's knee joint, providing data support for coil positioning; by fusing visual data and terahertz data, the accuracy and reliability of positioning are improved.

[0092] The front half-coil module includes a first drive mechanism 8 and a front half-coil 4;

[0093] The first drive mechanism 8 is located inside the magnet compartment 3 and can drive the front half coil 4 to move forward, backward, left, right and up in three-dimensional space.

[0094] It should be noted that the anterior coil module is driven by the first drive mechanism 8, which enables the anterior coil 4 to move flexibly in three-dimensional space to adapt to the knee joint position of different patients; the anterior coil 4, as a signal receiving component, works in conjunction with the posterior coil 5 to ensure the quality of MRI imaging.

[0095] The rear half-coil module includes a second drive mechanism 9 and a rear half-coil 5;

[0096] The second drive mechanism 9 is located inside the magnet compartment 3 and can drive the rear half coil 5 to move forward, backward, left, right and up in three-dimensional space.

[0097] It should be noted that the rear half-coil module, through the second drive mechanism 9, enables the rear half-coil 5 to move in the front-back, left-right, and up-down directions, achieving precise spatial positioning; together with the front half-coil 4, it forms a complete signal receiving system, ensuring the integrity and accuracy of imaging.

[0098] The status feedback module 7 is used to detect the real-time position and engagement status of the front half coil 4 and the rear half coil 5.

[0099] It should be noted that the status feedback module 7 monitors the position and engagement status of the front and rear half coils in real time through the position encoder and sensors, ensuring the accuracy and safety of the operation; and provides feedback signals to the main controller 2 to ensure the precise engagement of the coils and the stable operation of the system.

[0100] The main controller 2 is connected to the sensing module 6, the front half-coil module, the rear half-coil module, and the status feedback module 7 via signals, and is configured as follows:

[0101] The data collected by the sensing module 6 is processed to calculate the optimal imaging center point coordinates of the knee joint;

[0102] The first drive mechanism 8 and the second drive mechanism 9 are controlled to move in coordination according to the coordinates of the optimal imaging center point, so that the front half coil 4 and the rear half coil 5 are moved to the target position and engaged.

[0103] Based on the signal from the status feedback module 7, after confirming that the front half coil 4 and the rear half coil 5 are properly engaged, the magnetic resonance scan is started.

[0104] It should be noted that the main controller processes the data collected by the sensing module 6, calculates the optimal imaging center point, and controls the drive mechanism to move in coordination to ensure the precise alignment of the coil; it also coordinates the work of each module to ensure the efficient and stable operation of the entire system.

[0105] Taking a middle-aged male patient who is 175cm tall and weighs 70kg and has suspected meniscus injury due to knee pain as an example, when using this system for detection, the patient stands on the patient support platform 1 with his lower limbs hanging naturally vertically. The sensing module 6 in the magnet chamber is immediately activated, and the 3D vision camera unit quickly scans the patient's lower limbs, generating three-dimensional point cloud data including the calf, knee joint and lower thigh. At the same time, the terahertz imaging unit emits a terahertz wave with a frequency of 0.3THz to the knee joint area and receives the reflected signal. The main controller 2 fuses the two types of data and calculates the optimal imaging center point coordinates of the patient's knee joint as (X: 500mm, Y: 320mm, Z: 180mm) (with the lower left corner of the magnet chamber as the origin of the coordinates). Subsequently, the first drive mechanism 8 drives the front half coil 4 to move in the XY plane, reaching the plane position (X: 500mm, Y: 320mm) within 3 seconds; the second drive mechanism 9 drives the rear half coil 5 to move along the X, Y, and Z axes in a coordinated manner, with the planned motion path avoiding the patient's legs, accurately reaching the target position within 5 seconds; the automatic locking mechanism (using an electromagnetic lock) is energized and engaged, the micro-sensor of the status feedback module 7 detects that the coil spacing is only 0.5mm, the pressure sensor shows that the locking force is 8N (preset threshold range 5-10N), the main controller 2 determines that the locking is in place, and then starts the magnetic resonance scan. The entire preparation process takes only 12 seconds, far less than the 5-8 minutes of traditional manual adjustment;

[0106] In this embodiment, the main controller 2 coordinates the sensing module 6, the rear half-coil module, the front half-coil module, and the status feedback module 7 to achieve automatic coil positioning and engagement, eliminating human error and ensuring repeatability and consistency of each positioning. By automatically completing coil positioning and engagement, the scan preparation time is significantly shortened, and the efficiency of the equipment is improved. It can automatically adjust the coil position according to the patient's different body shape and physiological structure to achieve personalized and accurate positioning.

[0107] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process in this embodiment, another knee joint weight-bearing magnetic resonance adaptive detection system is provided. The system includes a magnet chamber 3, a patient support platform 1, a sensing module 6, an anterior half-coil module, a posterior half-coil module, a status feedback module 7, and a main controller 2.

[0108] The magnet chamber 3 includes two magnet sections, one at the front and one at the back, forming the patient's accommodating area;

[0109] The patient support platform 1 is located in the lower part of the patient receiving section of the magnet chamber 3;

[0110] The sensing module 6 is fixedly installed on the inner wall of the magnet chamber 3 to collect spatial information of the patient's knee joint;

[0111] The front half-coil module includes a first drive mechanism 8 and a front half-coil 4;

[0112] The first drive mechanism 8 is located inside the magnet compartment 3 and can drive the front half coil 4 to move forward, backward, left, right and up in three-dimensional space.

[0113] The rear half-coil module includes a second drive mechanism 9 and a rear half-coil 5;

[0114] The second drive mechanism 9 is located inside the magnet compartment 3 and can drive the rear half coil 5 to move forward, backward, left, right and up in three-dimensional space.

[0115] The status feedback module 7 is used to detect the real-time position and engagement status of the front half coil 4 and the rear half coil 5.

[0116] The main controller 2 is connected to the sensing module 6, the front half-coil module, the rear half-coil module, and the status feedback module 7 via signals, and is configured as follows:

[0117] The data collected by the sensing module 6 is processed to calculate the optimal imaging center point coordinates of the knee joint;

[0118] The first drive mechanism 8 and the second drive mechanism 9 are controlled to move in coordination according to the coordinates of the optimal imaging center point, so that the front half coil 4 and the rear half coil 5 are moved to the target position and engaged.

[0119] Based on the signal from the status feedback module 7, after confirming that the front half coil 4 and the rear half coil 5 are properly engaged, the magnetic resonance scan is started.

[0120] The first drive mechanism 8 and the second drive mechanism 9 are both three-axis motion platforms composed of X-axis and Y-axis linear modules and Z-axis ball screw module;

[0121] The X-axis and Y-axis linear modules perform horizontal (i.e., forward / backward and left / right) movement according to a preset horizontal accuracy (e.g., ±0.05mm) and motion path planning, such as guide rail model or motor specifications; the travel range of the X-axis and Y-axis linear modules can be set to 500-1000mm, the repeatability can be set to ±0.05mm, and the speed can be set to 0.1-1m / s; for example, using HIWIN or THK linear module series;

[0122] Z-axis ball screw module: Performs vertical (i.e., up and down) movement according to preset high precision (e.g., achieved through ball screw); the vertical stroke of the Z-axis ball screw module can be set to 200-500mm, the load capacity can be set to 5-20kg, and the accuracy can be set to ±0.02mm; for example, using NSK or PMI ball screws;

[0123] The modules work together to form a three-axis platform, which is coordinated by the main controller 2 to avoid contact with the patient.

[0124] An automatic locking mechanism is provided between the rear half-coil 5 and the front half-coil 4. The automatic locking mechanism adopts an electromagnetic lock or a mechanical locking mechanism.

[0125] The perception module 6 includes a 3D vision camera unit and a terahertz imaging unit;

[0126] The 3D vision camera unit is fixedly installed on the side wall of the magnet compartment 3. The vertical screen of the 3D vision camera unit covers the entire area of ​​the patient receiving part and is used to acquire three-dimensional point cloud data of the patient's lower limbs.

[0127] The terahertz imaging unit includes a transmitter and a receiver, which are fixedly mounted on the side wall of the magnet chamber; the beam of the terahertz imaging unit is directed towards the intended placement area of ​​the knee joint in the patient receiving part, for transmitting terahertz waves to the intended placement area of ​​the knee joint and receiving reflected signals;

[0128] The status feedback module 7 includes a first position encoder, a second position encoder, a macro sensor, and a pressure sensor;

[0129] The first position encoder is located at the first drive mechanism 8 and is used to provide feedback on the real-time position of the front half coil 4.

[0130] The second position encoder is located at the second drive mechanism 9 and is used to provide feedback on the real-time position of the rear half coil 5.

[0131] The macro sensor and pressure sensor are integrated on the mating surface of the rear half-coil 5 and / or the front half-coil 4;

[0132] The pressure sensor is used to monitor the fastening force. When the fastening force is within the preset threshold range, the main controller 2 determines that the fastening is in place.

[0133] like Figure 4 As shown, the following are embodiments of the knee joint weight-bearing magnetic resonance adaptive detection method provided in this disclosure. This method and the knee joint weight-bearing magnetic resonance adaptive detection system of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the knee joint weight-bearing magnetic resonance adaptive detection method, please refer to the embodiments of the knee joint weight-bearing magnetic resonance adaptive detection system described above.

[0134] The method includes the following steps:

[0135] S1. Spatial information of the patient's knee joint is collected by a sensing module fixed to the inner wall of the magnet chamber, and the data is fused to calculate the optimal imaging center point coordinates of the knee joint under load.

[0136] It should be noted that by collecting spatial information of the patient's knee joint through the sensing module and performing fusion processing, the optimal imaging center point coordinates are calculated, ensuring the accuracy of coil positioning and providing accurate data support for subsequent coil movement and engagement, thereby improving the automation level of MRI detection.

[0137] S2. Based on the optimal imaging center point coordinates, control the first drive mechanism and the second drive mechanism to move together, drive the front half coil and the rear half coil to the target position and complete the automatic engagement;

[0138] It should be noted that, based on the calculated optimal imaging center point coordinates, the drive mechanism is controlled to move in tandem, so that the front and rear half coils automatically move to the target position and complete the engagement, eliminating the error of manual positioning; the automated operation significantly shortens the scanning preparation time and improves the efficiency of the equipment.

[0139] S3. Based on the signal verification of the state feedback module, the engagement status of the front half coil and the rear half coil is verified. After confirming that the engagement is correct, the magnetic resonance scanning system is controlled to execute the scanning sequence.

[0140] It should be noted that the signal verification coil's engagement status based on the status feedback module ensures that the engagement is correct before starting the scan, thus guaranteeing the quality and reliability of the imaging; the verification process ensures the accuracy and safety of the scan.

[0141] This embodiment ensures the standardization and repeatability of the detection process by collecting spatial information and executing coil fastening and scanning; it automatically completes the entire process from positioning to scanning, reducing human intervention and improving detection efficiency and accuracy.

[0142] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, another adaptive detection method for weight-bearing magnetic resonance imaging of the knee joint is provided, which includes the following steps:

[0143] S1. Spatial information of the patient's knee joint is acquired through a sensing module fixed to the inner wall of the magnet chamber, and the data is fused to calculate the optimal imaging center point coordinates of the knee joint under load. The specific steps of step S1 are as follows:

[0144] S11. Scan the patient's lower limbs using a 3D vision camera unit to generate three-dimensional point cloud data, and construct an external contour model of the leg using computer vision algorithms. Combined with the anatomical position of the knee joint, output preliminary knee joint space coordinates. ;

[0145] In step S11, preliminary knee joint space coordinates are output based on the anatomical location of the knee joint. The calculation formula is as follows:

[0146]

[0147] in, The coordinates of the i-th point in the three-dimensional point cloud data of the expected area of ​​the knee joint are represented, and N is the total number of points in the three-dimensional point cloud of the expected area of ​​the knee joint.

[0148] The specific process of constructing the leg's external contour model in step S11 is as follows:

[0149] S111. Denoising and filtering preprocessing are performed on the 3D point cloud data, and the lower limb point cloud clusters are extracted using the region growing segmentation algorithm;

[0150] The specific process of constructing the leg's external contour model in step S111 is as follows:

[0151] The 3D point cloud data was preprocessed by denoising and filtering, and the lower limb point cloud clusters were extracted using a region growing segmentation algorithm.

[0152] Denoising of 3D point cloud data employs a statistical outlier removal filter, specifically including:

[0153] For each point in the 3D point cloud data Calculate the average distance to the k nearest neighbors. ;

[0154] Calculate the mean distance μ and standard deviation σ of the entire point cloud;

[0155] Remove those that meet the conditions The point, where m is an adjustable multiple threshold;

[0156] Step S111 uses a region growing segmentation algorithm to extract the lower limb point cloud clusters, specifically including:

[0157] Calculate the normal vector of each point in the 3D point cloud data. and curvature ;

[0158] Select the point with the smallest curvature as the seed point and begin growth;

[0159] For the current seed point Check neighboring points If neighboring points With the current seed point If the angle between the normal vectors is less than the threshold θ and the curvature difference is less than the curvature difference threshold c, then the neighboring point is merged into the current region and used as a new seed point.

[0160] Repeat the above process until no new points can be merged, thus completing the region growth.

[0161] S112. Based on the random sampling consensus algorithm, the extracted lower limb point cloud is geometrically fitted to construct the external contour model of the leg;

[0162] The specific steps of step S112 are as follows:

[0163] The leg contour is approximated by fitting a cylindrical model using a random sampling consensus algorithm. The parameters of the cylindrical model are the central axis direction vector, a point on the central axis, and the radius.

[0164] Randomly select several points to calculate the initial cylindrical model, calculate the distance from all points in the 3D point cloud data to the initial cylindrical model, and consider points with a distance less than the threshold t as interior points;

[0165] After multiple iterations, the cylindrical model with the most interior points is selected as the fitting result.

[0166] Based on the fitted cylindrical model, the leg axis is determined, and the coordinates of the center point of the knee joint space are estimated by combining the anatomical position of the knee joint. ;

[0167] S12. Terahertz waves are emitted towards the knee joint region through a terahertz imaging unit, and the reflected signals are received. The intensity of the reflected signals is analyzed. The distribution of the internal structures of the knee joint is used to output the optimal imaging point coordinates. ;

[0168] In step S12, the optimal imaging point coordinates for the internal structures of the knee joint are determined by finding the maximum value of the terahertz reflected signal intensity. The calculation formula is as follows:

[0169]

[0170] in, This represents the intensity of the terahertz reflected signal received at spatial coordinates (x, y, z);

[0171] After receiving the reflected terahertz wave signal in step S12, filtering and noise reduction processing is performed, using a bandpass filter to retain the effective signal in the terahertz frequency band; the transfer function of the bandpass filter is:

[0172] ,when ;

[0173] ,when or ;

[0174] in, This is the lower limit frequency of the terahertz band. This is the upper limit frequency of the terahertz band;

[0175] S13. Initial knee joint space coordinates Optimal imaging point coordinates of the internal structures of the knee joint The data is input into the data fusion algorithm to calculate the final optimal imaging center point coordinates. ;

[0176] In step S13, the data fusion algorithm uses a weighted average method, and its calculation formula is as follows:

[0177]

[0178] in, and The weighting coefficients are dynamically adjusted based on the real-time signal-to-noise ratio of visual data and terahertz data, and satisfy the following conditions: ;

[0179] S2. Based on the optimal imaging center point coordinates, control the first drive mechanism and the second drive mechanism to move together, drive the front half coil and the rear half coil to the target position and complete the automatic engagement;

[0180] The specific steps of step S2 are as follows:

[0181] S21. Set the optimal imaging center point coordinates The command is converted into a displacement command. First, the first drive mechanism is controlled to move the front half coil to the target position in three-dimensional space. Then, the second drive mechanism is controlled to move the rear half coil in three-dimensional space and a motion path is planned to avoid contact with the patient.

[0182] S22. After both the current half-coil and the rear half-coil have moved to the target position, the automatic fastening mechanism is controlled to perform the fastening action, and the fastening status is monitored by the micro sensor and the pressure sensor;

[0183] When the distance value detected by the macro sensor reaches the preset closing distance threshold, and the force value detected by the pressure sensor is within the preset force threshold range, it is determined that the fastening is in place;

[0184] S3. Based on the signal verification of the state feedback module, the engagement status of the front half coil and the rear half coil is verified. After confirming that the engagement is correct, the magnetic resonance scanning system is controlled to execute the scanning sequence.

[0185] For example, taking a teenage female patient who is 150cm tall and weighs 45kg (undergoing knee joint examination due to sports injury) as an example, the specific execution process of step S1 is as follows:

[0186] The S11 3D vision camera unit (e.g., a Basler blaze 1016-100 3D vision camera) scans the patient's lower limbs at a frame rate of 30fps, generating 3D point cloud data containing 1.2 million points. First, a statistical outlier removal filter is used for noise reduction, with k=20 (taking the 20 nearest neighbors of each point). The average distance of the entire point cloud is calculated to be μ=0.8mm, and the standard deviation σ=0.3mm. Outliers satisfying the specified parameters are then removed. Outliers were identified, and 15,600 noise points were removed. Then, a region growing segmentation algorithm was used to extract the lower limb point cloud clusters. The normal vector and curvature of each point were calculated, and the point with the smallest curvature (0.02) was selected as the seed point. A threshold for the angle between the normal vectors was set. With a curvature difference threshold of c=0.05, a complete lower limb point cloud cluster was obtained through growth. Finally, a cylindrical model was fitted using the random sampling consensus algorithm. After 500 iterations, a model with 1.02 million interior points was selected as the fitting result, and the leg axis direction vector was determined to be (0.02, 1, 0.01). Combined with the anatomical position of the knee joint (the angle between the thigh and lower leg axes is approximately 175°), the preliminary knee joint space coordinates were calculated. = (X: 420mm, Y: 280mm, Z: 160mm);

[0187] S12. Terahertz imaging unit (transmission frequency: 0.2-0.5THz) emits terahertz waves towards the knee joint region. The receiver collects the reflected signal and uses a bandpass filter (lower limit frequency) to filter the signal. =0.2THz, upper limit frequency =0.5THz) retains the effective signal while filtering out low-frequency interference (such as environmental electromagnetic noise) and high-frequency clutter. Analysis of the reflected signal intensity distribution reveals that the signal intensity reaches its maximum at coordinates (X: 421mm, Y: 282mm, Z: 161mm). =256 (grayscale value), determine the optimal imaging point coordinates for the internal structures of the knee joint. = (421, 282, 161).

[0188] S13. Due to the signal-to-noise ratio (SNR) of 3D visual data (SNR1) being 45dB and the signal-to-noise ratio (SNR) of terahertz data (SNR2) being 38dB, dynamically adjust the weighting coefficients. =0.55, =0.45 (satisfied) The optimal imaging center point coordinates are calculated using the weighted average method. =0.55×(420, 280, 160)+0.45×(421, 282, 161)=(420.45mm, 280.9mm, 160.45mm), accurate to one decimal place, providing precise coordinates for subsequent coil positioning;

[0189] The specific execution process of step S2 is as follows:

[0190] S21. The main controller 2 will assign the optimal imaging center point coordinates. = (420.45, 280.9, 160.45) is converted into a displacement command; after receiving the command, the first drive mechanism 8 (XY platform, accuracy: ±0.05mm) drives the front half coil 4 to move in three-dimensional space. The distance moved in the X-axis direction is 420.45-380=40.45mm (initial position X=380mm), and the distance moved in the Y-axis direction is 280.9-250=30.9mm (initial position Y=250mm). It reaches the target position (420.45, 280.9, 160.45) within 2.5 seconds. 0.45, 280.9); The second drive mechanism 9 (X-axis and Y-axis linear module accuracy ±0.05mm, Z-axis ball screw module accuracy ±0.02mm) drives the rear half coil 5 to move, moving 38.6mm in the X-axis direction, 29.8mm in the Y-axis direction, and descending from the initial height of 250mm to 160.45mm in the Z-axis direction. The motion path is planned to first move along the X and Y axes to 10mm directly above the knee joint, and then descend along the Z-axis, avoiding the patient's leg throughout the entire process. The movement takes 4 seconds.

[0191] S22. After the front and rear half coils reach the target position, the automatic fastening mechanism (mechanical locking mechanism) is activated, the locking tongue extends and inserts into the lock; the micro-sensor (accuracy: ±0.01mm) detects that the coil spacing is 0.3mm, reaching the preset closing distance threshold (≤0.5mm); the pressure sensor (range: 0-50N, accuracy: ±0.1N) detects that the fastening force is 6.2N, which is within the preset force threshold range (4-8N), and the main controller 2 determines that the fastening is in place and sends a fastening success signal;

[0192] The specific execution process of step S3 is as follows: The first position encoder of the status feedback module 7 feeds back the real-time position of the front half coil 4 as (420.45mm, 280.9mm), which deviates from the target position by only 0.02mm; the second position encoder feeds back the real-time position of the rear half coil 5 as (420.45mm, 280.9mm, 160.45mm), which deviates from the target position by 0.03mm, and the position accuracy meets the requirements. Meanwhile, the micro-sensor continuously monitors the coil spacing, stabilizing it between 0.3-0.4 mm within 10 seconds, while the pressure sensor displays a stable engagement force of 6.2-6.3 N without significant fluctuations. The main controller 2 integrates all feedback signals to confirm that the engagement state is stable and correct, and sends a start command to the magnetic resonance scanning system (e.g., using a MAGSPIN Armous Formulla III magnetic resonance scanning system). The system immediately executes the preset knee joint-specific scanning sequence (including T1-weighted images, T2-weighted images, and PDWI sequences). During the scanning process, the coil and the knee joint maintain precise contact throughout. In the final generated image, structures such as the meniscus and cartilage are clearly visible, and the signal-to-noise ratio reaches 32 dB, meeting the requirements for clinical diagnosis.

[0193] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0194] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A knee joint weight-bearing position magnetic resonance adaptive detection system, characterized in that, It includes a magnet chamber (3), a patient support table (1), a sensing module (6), a front half coil module, a rear half coil module, a status feedback module (7), and a main controller (2); The magnet compartment (3) includes two magnet sections, one at the front and one at the back, forming the patient housing. The patient support table (1) is located at the lower part of the patient receiving section of the magnet chamber (3); The sensing module (6) is fixedly installed on the inner wall of the magnet chamber (3) to collect spatial information of the patient's knee joint; The front half-coil module includes a first drive mechanism (8) and a front half-coil (4); The first drive mechanism (8) is located inside the magnet compartment (3) and can drive the front half coil (4) to move forward, backward, left, right and up in three-dimensional space; The rear half-coil module includes a second drive mechanism (9) and a rear half-coil (5). The second drive mechanism (9) is located inside the magnet compartment (3) and can drive the rear half coil (5) to move forward, backward, left, right and up in three-dimensional space; The status feedback module (7) is used to detect the real-time position and engagement status of the front half coil (4) and the rear half coil (5); The main controller (2) is connected to the sensing module (6), the front half-coil module, the rear half-coil module, and the status feedback module (7) respectively, and is configured as follows: The data collected by the sensing module (6) is processed to calculate the optimal imaging center point coordinates of the knee joint; The first drive mechanism (8) and the second drive mechanism (9) are controlled to move in coordination according to the optimal imaging center point coordinates, so that the front half coil (4) and the rear half coil (5) move to the target position and engage. Based on the signal from the status feedback module (7), after confirming that the front half coil (4) and the rear half coil (5) are properly engaged, the magnetic resonance scan is started.

2. The knee joint weight-bearing position magnetic resonance adaptive detection system according to claim 1, characterized in that, The first drive mechanism (8) and the second drive mechanism (9) are both three-axis motion platforms composed of X-axis and Y-axis linear modules and Z-axis ball screw module; An automatic locking mechanism is provided between the rear half coil (5) and the front half coil (4), and the automatic locking mechanism adopts an electromagnetic lock or a mechanical locking mechanism.

3. The knee joint weight-bearing position magnetic resonance adaptive detection system according to claim 2, characterized in that, The perception module (6) includes a 3D vision camera unit and a terahertz imaging unit; The 3D vision camera unit is fixedly installed on the side wall of the magnet cabin (3). The field of view of the 3D vision camera unit covers the entire area of ​​the patient receiving part and is used to acquire three-dimensional point cloud data of the patient's lower limbs. The terahertz imaging unit includes a transmitter and a receiver, which are fixedly mounted on the side wall of the magnet chamber. The beam of the terahertz imaging unit is directed towards the intended placement area of ​​the knee joint in the patient housing, for transmitting terahertz waves to the intended placement area of ​​the knee joint and receiving reflected signals.

4. The knee joint weight-bearing position magnetic resonance adaptive detection system according to claim 1, characterized in that, The status feedback module (7) includes a first position encoder, a second position encoder, a macro sensor, and a pressure sensor; The first position encoder is located at the first drive mechanism (8) and is used to provide feedback on the real-time position of the front half coil (4); The second position encoder is located at the second drive mechanism (9) and is used to provide feedback on the real-time position of the rear half coil (5); The macro sensor and pressure sensor are integrated on the mating surface of the front half coil (4) and / or the rear half coil (5); The pressure sensor is used to monitor the fastening force. When the fastening force is within the preset threshold range, the main controller (2) determines that the fastening is in place.

5. An adaptive magnetic resonance imaging method for weight-bearing knee joint, characterized in that, Includes the following steps: S1. Spatial information of the patient's knee joint is collected by a sensing module fixed to the inner wall of the magnet chamber, and the data is fused to calculate the optimal imaging center point coordinates of the knee joint under load. S2. Based on the optimal imaging center point coordinates, control the first drive mechanism and the second drive mechanism to move together, drive the front half coil and the rear half coil to the target position and complete the automatic engagement; S3. Based on the signal verification of the state feedback module, the engagement status of the front half coil and the rear half coil is verified. After confirming that the engagement is correct, the magnetic resonance scanning system is controlled to execute the scanning sequence.

6. The knee joint weight-bearing position magnetic resonance adaptive detection method according to claim 5, characterized in that, The specific steps of step S1 are as follows: S11. Scan the patient's lower limbs using a 3D vision camera unit to generate three-dimensional point cloud data, and construct an external contour model of the leg using computer vision algorithms. Combined with the anatomical position of the knee joint, output preliminary knee joint space coordinates. ; S12. Terahertz waves are emitted towards the knee joint region through a terahertz imaging unit, and the reflected signals are received. The intensity of the reflected signals is analyzed. The distribution of the internal structures of the knee joint is used to output the optimal imaging point coordinates. ; S13. Initial knee joint space coordinates Optimal imaging point coordinates of the internal structures of the knee joint The data is input into the data fusion algorithm to calculate the final optimal imaging center point coordinates. .

7. The knee joint weight-bearing position magnetic resonance adaptive detection method according to claim 6, characterized in that, In step S11, preliminary knee joint space coordinates are output based on the anatomical location of the knee joint. The calculation formula is as follows: in, The coordinates of the i-th point in the three-dimensional point cloud data of the expected area of ​​the knee joint are represented, and N is the total number of points in the three-dimensional point cloud of the expected area of ​​the knee joint. In step S12, the optimal imaging point coordinates for the internal structures of the knee joint are determined by finding the maximum value of the terahertz reflected signal intensity. The calculation formula is as follows: in, This represents the intensity of the terahertz reflected signal received at spatial coordinates (x, y, z); In step S13, the data fusion algorithm uses a weighted average method, and its calculation formula is as follows: in, and The weighting coefficients are dynamically adjusted based on the real-time signal-to-noise ratio of visual data and terahertz data, and satisfy the following conditions: .

8. The knee joint weight-bearing position magnetic resonance adaptive detection method according to claim 6, characterized in that, The specific process of constructing the leg's external contour model in step S11 is as follows: S111. Denoising and filtering preprocessing are performed on the 3D point cloud data, and the lower limb point cloud clusters are extracted using the region growing segmentation algorithm; S112. Based on the random sampling consensus algorithm, the extracted lower limb point cloud is geometrically fitted to construct the external contour model of the leg; After receiving the reflected terahertz wave signal in step S12, filtering and noise reduction processing is performed, using a bandpass filter to retain the effective signal in the terahertz frequency band; the transfer function of the bandpass filter is: ,when ; ,when or ; in, This is the lower limit frequency of the terahertz band. This is the upper limit frequency of the terahertz band.

9. The knee joint weight-bearing position magnetic resonance adaptive detection method according to claim 8, characterized in that, The specific process of constructing the leg's external contour model in step S111 is as follows: The 3D point cloud data was preprocessed by denoising and filtering, and the lower limb point cloud clusters were extracted using a region growing segmentation algorithm. Denoising of 3D point cloud data employs a statistical outlier removal filter, specifically including: For each point in the 3D point cloud data Calculate the average distance to the k nearest neighbors. ; Calculate the mean distance μ and standard deviation σ of the entire point cloud; Remove those that meet the conditions The point, where m is an adjustable multiple threshold; Step S111 uses a region growing segmentation algorithm to extract the lower limb point cloud clusters, specifically including: Calculate the normal vector of each point in the 3D point cloud data. and curvature ; Select the point with the smallest curvature as the seed point and begin growth; For the current seed point Check neighboring points If neighboring points With the current seed point If the angle between the normal vectors is less than the threshold θ and the curvature difference is less than the curvature difference threshold c, then the neighboring point is merged into the current region and used as a new seed point. Repeat the above process until no new points can be merged, thus completing the region growth. The specific steps of step S112 are as follows: The leg contour is approximated by fitting a cylindrical model using a random sampling consensus algorithm. The parameters of the cylindrical model are the central axis direction vector, a point on the central axis, and the radius. Randomly select several points to calculate the initial cylindrical model, calculate the distance from all points in the 3D point cloud data to the initial cylindrical model, and consider points with a distance less than the threshold t as interior points; After multiple iterations, the cylindrical model with the most interior points is selected as the fitting result. Based on the fitted cylindrical model, the leg axis is determined, and the coordinates of the center point of the knee joint space are estimated by combining the anatomical position of the knee joint. .

10. The knee joint weight-bearing position magnetic resonance adaptive detection method according to claim 5, characterized in that, The specific steps of step S2 are as follows: S21. Set the optimal imaging center point coordinates The command is converted into a displacement command. First, the first drive mechanism is controlled to move the front half coil to the target position in three-dimensional space. Then, the second drive mechanism is controlled to move the rear half coil in three-dimensional space and a motion path is planned to avoid contact with the patient. S22. After both the current half-coil and the rear half-coil have moved to the target position, the automatic fastening mechanism is controlled to perform the fastening action, and the fastening status is monitored by the micro sensor and the pressure sensor; When the distance value detected by the macro sensor reaches the preset closing distance threshold, and the force value detected by the pressure sensor is within the preset force threshold range, it is determined that the fastening is in place.

Images