Esophageal electromagnetic navigation system

Abstract

The application discloses an esophagus electromagnetic navigation system, a planning module is used for constructing a three-dimensional model of a patient's esophagus based on computer tomography image data, determining a target point position, and calibrating esophagus feature points; a registration module is used for acquiring spatial coordinates of the feature points, matching the spatial coordinates with coordinates of the feature points in the three-dimensional model, and constructing an electromagnetic navigation coordinate system; and a navigation module is used for guiding a navigation probe to move into the patient's esophagus, and taking the esophagus structure as a registration reference, thus fundamentally solving the deformation error problem inherent in traditional methods which depend on soft tissue matching, and realizing accurate and rapid positioning and navigation of esophagus lesions.

Description

Technical Field

[0001] This invention belongs to the field of electromagnetic navigation technology, specifically relating to an esophageal electromagnetic navigation system. Background Technology

[0002] Early diagnosis and treatment of esophageal diseases such as esophageal cancer heavily rely on esophagoscopy. However, traditional esophagoscopy primarily depends on the doctor's tactile and visual experience. During the examination, the doctor observes the esophageal wall through endoscopic video, performing biopsies or marking suspicious lesions, or using esophagoscopy with ultrasound to diagnose and treat extraesophageal lesions (or lymph nodes). This method has significant limitations:

[0003] 1. Inaccurate localization: The esophagus is a soft, peristaltic muscular tube, and its spatial position within the body cavity can shift and deform with respiration, heartbeat, and changes in patient position. Deep or extraesophageal lesions that doctors find on imaging (such as CT scans) may be difficult to locate precisely under endoscopic visualization, leading to errors in biopsy sampling or missed lesions.

[0004] 2. Lack of three-dimensional spatial correlation: Endoscopy provides a two-dimensional surface view of the inside of the lumen, which cannot be effectively correlated and fused with the preoperative three-dimensional CT images. It is difficult for doctors to accurately correspond the specific location under the endoscope with the anatomical structures (such as lymph nodes and blood vessels) outside the esophagus shown in the CT images.

[0005] 3. Reliance on experience: The accuracy of the examination is highly correlated with the personal experience of the doctor performing the procedure, and there may be significant differences in the procedures between different doctors.

[0006] Existing electromagnetic navigation technology has been applied to bronchoscopy and intestinal examinations, but its application in the esophagus faces unique challenges. Traditional soft tissue registration methods suffer from significant physiological peristalsis and respiratory displacement in the esophagus, resulting in large registration errors and navigation accuracy that cannot meet clinical requirements. Therefore, there is an urgent need for a navigation system that can overcome esophageal deformation and achieve high-precision positioning. Summary of the Invention

[0007] The technical problem that the present invention aims to solve is the problem of inaccurate positioning, lack of three-dimensional guidance, and large navigation error caused by soft tissue deformation in traditional esophagoscopy.

[0008] To solve the above-mentioned technical problems, the present invention provides an esophageal electromagnetic navigation system, comprising: The planning module is used to construct a three-dimensional model of the patient's esophagus based on computed tomography image data of the esophageal region, determine the location of the target point in the three-dimensional model, and use the esophageal structure as the registration reference. Based on the location of the target point, the module marks the feature points for registration in the esophageal structure. The feature points have relatively fixed anatomical locations. The registration module is used during surgery to obtain the actual spatial coordinates of feature points through a navigation probe with positioning sensors, and match them with the coordinates of the corresponding feature points in the 3D model to construct an electromagnetic navigation coordinate system that associates the 3D model with the real patient space. The navigation module is used to track the position of the navigation probe with positioning sensors in real time based on the constructed electromagnetic navigation coordinate system, and guide it to move to the target point in the patient's esophagus. The positioning module of the internal electromagnetic field generation module is used to blow air into the patient's esophagus based on preset target parameters during the operation of the registration module and the navigation module. By controlling the air pressure in the esophagus, the esophageal shape is stabilized, its peristalsis is inhibited, and it is kept in a relatively expanded and fixed state, thereby significantly improving the accuracy of registration and navigation. Both the electromagnetic field generating module and the positioning sensor have coils inside. The electromagnetic field generating module generates a stable electromagnetic field covering the surgical area where the patient is located. The positioning sensor senses the magnetic field generated by the electromagnetic field generating module to generate an electromagnetic field coordinate system, which enables the navigation probe to obtain the actual spatial coordinates of the feature points. Guided by an esophagoscope, the tip of the navigation probe gently touches the feature points on the inner wall of the patient's esophagus that correspond to the planning module. Each time a point is touched, its spatial coordinates in the electromagnetic field coordinate system are recorded. After recording the spatial coordinates of all feature points, the registration module starts the registration algorithm to match the physical spatial coordinate set of the collected feature points with the corresponding point coordinate set calibrated on the 3D model. By minimizing the distance error between the two sets of point sets, the algorithm accurately calculates a transformation matrix. This matrix defines the transformation relationship from the CT image coordinate space to the actual physical space of the surgery, realizing the electromagnetic navigation coordinate system that links the 3D model with the actual patient space.

[0009] Preferably, the feature points include the thoracic inlet, the level of the zona venosa, the level of the right lower pulmonary vein, and the pylorus.

[0010] Preferably, a medical image segmentation framework based on the combination of Transformer and U-Net is used to process the computed tomography image data by leveraging the powerful global context modeling capabilities of Transformer to compensate for the skip connection structure of traditional U-Net and restore local details. The three-dimensional model of the patient's esophagus is then automatically or semi-automatically reconstructed using medical image processing software.

[0011] Preferably, the navigation probe is inserted into the patient's esophagus along with the esophagoscope, and the three-dimensional spatial coordinates and attitude information of the navigation probe are acquired in real time by the positioning sensor.

[0012] This invention provides an esophageal electromagnetic navigation system. A planning module constructs a three-dimensional model of the patient's esophagus based on computed tomography (CT) image data, determines the location of target points, and calibrates esophageal feature points. A registration module acquires the spatial coordinates of the feature points and matches them with the coordinates of the feature points in the three-dimensional model to construct an electromagnetic navigation coordinate system. A navigation module guides the navigation probe to move into the patient's esophagus, using the esophageal structure as a registration reference. This fundamentally solves the deformation error problem inherent in traditional methods that rely on soft tissue matching, enabling precise and rapid localization and navigation of esophageal lesions. Attached Figure Description

[0013] Figure 1 A schematic diagram of the overall module composition of the esophageal electromagnetic navigation system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of feature point calibration of the planning module in the esophageal electromagnetic navigation system provided in an embodiment of the present invention; Figure 3 A schematic diagram illustrating the principle of spatial coordinate matching in the registration module of the esophageal electromagnetic navigation system provided in this embodiment of the invention; Figure 4 This is a schematic diagram of the esophageal electromagnetic navigation device provided in an embodiment of the present invention; Figure label: 100 - Target point; 200 - Thoracic inlet; 300 - Azygos vein level; 400 - Right lower pulmonary vein level; 600 - Pylorus; 500 - Esophageal electromagnetic navigation device; 510 - System control unit; 520 - Positioning sensor interface unit; 530 - Magnetic field generator; 540 - Positioning guide wire; 550 - Medical imaging component; 560 - Body position detector. Detailed Implementation

[0014] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0015] To address the issues mentioned in the background art, such as inaccurate positioning, spatial displacement and deformation due to breathing, heartbeat, and patient position changes leading to biopsy sampling errors or missed lesions; lack of three-dimensional spatial correlation, making effective spatial correlation and fusion with preoperative three-dimensional CT images impossible; and reliance on experience, with accuracy highly correlated with the operator's individual experience, potentially resulting in significant operational differences between different doctors, the esophageal electromagnetic navigation system provided in this invention uses the esophageal structure as a registration benchmark, achieving high registration accuracy, strong reliability, and a blind-zone-free positioning range. Figure 1 As shown, the system includes: Planning module: Used to construct a 3D model of the patient's esophagus based on computed tomography (CT) image data of the esophageal region, determine the location of target points (such as lesions) in the 3D model, and calibrate feature points for registration. For example... Figure 2 As shown, the feature points include the thoracic inlet at 200°, the azygos vein level at 300°, the right inferior pulmonary vein level at 400°, and the pylorus at 600°. These feature points are located on the esophageal structures and have relatively fixed anatomical positions, serving as stable registration benchmarks.

[0016] The planning module specifically includes: Image acquisition unit: Used to acquire computed tomography (CT) image data of the patient's esophageal region.

[0017] 3D Reconstruction Unit: Used to construct a 3D model of the patient's esophagus based on computed tomography (CT) image data, and to mark several feature points in the esophageal structure based on the target point location.

[0018] It should be noted that when calibrating several feature points, the target point position is first determined in the three-dimensional model of the patient's esophagus. Then, appropriate feature points or point sets are selected in the esophageal structure based on the target point position. The esophageal structure is used as the registration reference, which fundamentally solves the deformation error problem inherent in traditional methods that rely on soft tissue matching.

[0019] In this embodiment, the 3D reconstruction unit automatically or semi-automatically reconstructs a 3D model of the patient's esophagus using medical image processing software. Based on a medical image segmentation framework combining Transformer and U-Net, the powerful global context modeling capability of Transformer is used to compensate for the skip connection structure of traditional U-Net and restore local details, thereby achieving a more accurate segmentation effect. The model clearly shows the esophageal structure.

[0020] like Figure 3As shown, the registration module is used during surgery to obtain the actual spatial coordinates of the aforementioned feature points in the patient's body through a navigation probe equipped with a positioning sensor, and match them with the coordinates of the corresponding feature points in the 3D model, thereby constructing an electromagnetic navigation coordinate system that associates the virtual 3D model with the real patient space.

[0021] like Figure 4 As shown, the navigation module is used to track the position of the navigation probe in real time based on the constructed electromagnetic navigation coordinate system and guide it to the target point in the patient's esophagus.

[0022] The navigation module specifically includes: Navigation unit: includes a navigation probe, which is equipped with a positioning sensor for real-time acquisition of its position information. The navigation probe can be inserted into the patient's esophagus along with the esophagoscope (positioning guide), and the positioning sensor can acquire the three-dimensional spatial coordinates and attitude information of the navigation probe in real time.

[0023] Control unit: Used to receive position information from positioning sensors, determine the real-time spatial coordinates of the navigation probe in the three-dimensional model, calculate its relative position and path to the target point, and guide the navigation probe to the target point in the esophagus of the patient through the electromagnetic navigation coordinate system.

[0024] The positioning sensor contains three orthogonal coils, which are used to sense the magnetic field generated by the electromagnetic field generation module.

[0025] The navigation unit is coupled to the control unit.

[0026] Positioning module: including the ventilator unit, used for positioning during registration and navigation by displaying the three-dimensional navigation arrows and target location.

[0027] In this embodiment, the positioning module further includes an electromagnetic field generation module: it is coupled to the control unit and is used to generate a stable electromagnetic field covering the surgical area where the patient is located, providing a spatial reference frame for the positioning sensor.

[0028] Inside the electromagnetic field generating module are three orthogonal coils, which are sequentially energized with alternating currents of different frequencies, thereby generating alternating electromagnetic fields in three directions in space.

[0029] Display module: Used to synchronously display computed tomography (CT) images of the patient's esophageal region, 3D model images, feature points and target points, and the real-time virtual position of the navigation probe in the 3D model, providing doctors with an intuitive visual navigation interface.

[0030] During registration in the registration module, the patient lies in the surgical position, and the electromagnetic field generating module is placed above or below the patient's chest area, covering the area with its generated electromagnetic field. Then, guided by an esophagoscope, the tip of the navigation probe gently touches the actual feature points on the inner wall of the patient's esophagus that correspond to the planning module. Each time a point is touched, its spatial coordinates in the electromagnetic field coordinate system are recorded.

[0031] like Figure 3 As shown, after recording the spatial coordinates of all real feature points, the registration module initiates a registration algorithm (such as the Iterative Closest Point (ICP) algorithm). This algorithm matches the set of physical spatial coordinates of multiple feature points acquired intraoperatively with the set of corresponding point coordinates calibrated on the 3D model preoperatively. By minimizing the distance error between the two sets of points, the algorithm accurately calculates a transformation matrix. This matrix defines the transformation relationship from the CT image coordinate space to the actual physical space of the surgery. After this step, the virtual CT 3D model is precisely aligned with the patient's actual anatomical structure, establishing a stable, global navigation coordinate system.

[0032] The basic principle of ICP (Iterative Closest Point) registration is: Through continuous iteration, one point cloud (source point cloud) is gradually brought closer to another point cloud (target point cloud) until they overlap as much as possible. Specific steps: Find corresponding points: For each point in the source point cloud, find the nearest point in the target point cloud as a matching pair. Calculate the optimal transformation: Based on these matching point pairs, calculate a rigid body transformation (including rotation and translation) that minimizes the overall distance between the source and target point clouds after the transformation. Apply the transformation: Update the position of the source point cloud using this transformation. Repeat the above process until the error between the point clouds no longer decreases significantly (i.e., convergence). The core idea is: "First guess the correspondence, then optimize the position; then guess again, then optimize again…," repeating this process until high-precision alignment of the two point clouds is achieved.

[0033] ICP objective function:

[0034] It is a rotation matrix that describes the rotation direction of the source point cloud relative to the target point cloud; It is a translation vector that describes the positional offset of the source point cloud in space; It is the i-th point in the source point cloud, and the point set position signal acquired electromagnetically; It is the closest point in the target point cloud corresponding to Pi; It is the weight for matching the i-th point, which improves robustness and avoids noise points affecting the overall registration result.

[0035] The innovation of the registration module lies in its registration process, which uses pre-calibrated esophageal feature points on the CT image 3D model (such as 200 at the thoracic inlet, 300 at the azygos vein level, 400 at the right lower pulmonary vein level, and 600 at the pylorus) to register with the spatial coordinates of corresponding feature points actually touched and collected by the navigation probe with positioning sensors used during surgery on the patient's body surface or inside the esophagus. This registration method utilizes the stability of the esophageal structure, reduces the impact of significant changes in body position during surgery and CT scans, and avoids the deformation effects of bronchial or skin surface registration points, thereby establishing a high-precision, globally unified coordinate system.

[0036] Since the electromagnetic field strength is inversely proportional to the cube of the distance, the voltage signal induced by the sensor coil (i.e., the raw values ​​of the 9-channel ADC) contains the sensor's position (x, y, z) and attitude (azimuth, elevation, roll) relative to the electromagnetic field generating module. The specific process is as follows: Step S1: Zero bias removal. The purpose is to eliminate the inherent background noise of the sensor and the DC bias caused by static magnetic fields (such as the geomagnetic field) in the environment on the ADC readings. In the process, the zero bias data is first obtained, that is, before the tracking starts, the sensor is placed in an environment that is "non-magnetic" or known to be free of FG magnetic field interference, and the readings of the 9-channel ADC are recorded at this time. This set of readings is the zero bias value (Bias or Offset), denoted as ADC_offset[9]. After that, real-time subtraction is performed, that is, in the subsequent normal tracking process, the zero bias value is subtracted from the 9-channel raw ADC value ADC_raw[9] acquired in each frame. ADC_corrected[i] = ADC_raw[i] - ADC_offset[i] (for i = 0 to 8).

[0037] Step S2: Calibration Matrix. Its purpose is to convert the zero-biased 9-channel ADC voltage values ​​into a "cleaner" vector representing the physical magnetic field strength. This matrix compensates for the sensitivity differences of each sensor coil, minor non-orthogonality, and the characteristics of the FG (electromagnetic field generation module coil). During processing, the calibration matrix is ​​first obtained. This matrix C[9x9] contains unique parameters for each sensor, generated by NDI during factory calibration and stored in the sensor's ROM. These 81 values ​​need to be read from the sensor ROM via the API or communication protocol provided by NDI. Then, matrix multiplication is performed, that is, for each frame of zero-biased 9-channel signals ADC_corrected[9] (a 9x1 column vector), the 9x9 calibration matrix C is multiplied on the left. B[9x1] = C[9x9] ADC_corrected[9x1] Step S3: Dipole Model Optimization. The purpose is to use the magnetic dipole physical model to calculate the sensor position (x, y, z) from the 3x3 magnetic field matrix B_mat obtained in step S2.

[0038] The mathematical model (dipole field formula) is as follows: For a transmitting coil with magnetic moment M, the magnetic field B generated at a distance r is: B = (μ0 / 4π) [(3) (M · r) r) / |r| 5 - M / |r|³ ], where r = [x, y,z] is the position vector of the sensor relative to the transmitting coil.

[0039] The optimization process is as follows: First, for each of the three transmitting coils (Mx, My, Mz), there exists a measured magnetic field vector (i.e., a column in B_mat). The unknown is the position vector r = [x, y, z]. Next, a nonlinear least-squares cost function is established. This function calculates the difference (residual) between the magnetic field predicted using the dipole formula at the currently guessed position r and the magnetic field B_mat "actually measured" in step S2. Cost(r) = Σ ||B_measured_i - B_dipole_model_i(r) ||² (summed over the three transmitting coils). Finally, a nonlinear optimization algorithm (e.g., the Levenberg-Marquardt algorithm) is used to find the position r that minimizes the cost function Cost(r). This process is iterative and requires an initial guess (typically the position from the previous frame, or a rough estimate, can be used as a starting point). After the optimization algorithm converges, the optimal three-dimensional coordinates [x, y, z] of the positioning sensor are obtained.

[0040] Step S4: SVD Attitude Determination. The purpose is to calculate the attitude (rotation matrix) of the positioning sensor from the 3x3 magnetic field matrix B_mat obtained in Step S2, based on the known position [x, y, z]. The core idea is that the orientation of the FG coil in its own coordinate system is known (e.g., the magnetic moment direction of the X coil is [1,0,0]). According to the dipole model, if the positioning sensor and FG have the same attitude (i.e., no rotation), then the measured B_mat will be a specific matrix. Any attitude change is equivalent to rotating this "ideal" matrix.

[0041] During processing, the theoretical field is first calculated, that is, using the position [x, y, z] obtained in the third step, and substituting it into the dipole formula, to calculate the position when the positioning sensor and the FG coordinate system are perfectly aligned (i.e., the rotation matrix R is the identity matrix). should The measured ideal magnetic field matrix B_ideal. This is a 3x3 matrix. B_ideal = f(x, y, z, Mx, My, Mz) (Mx, My, Mz are the known magnetic moment vectors of the FG coil). A relationship is then established: the actually measured B_mat can be considered as the result of the ideal field B_ideal being rotated by R by the positioning sensor. The relationship between them can be expressed as: B_measured = R B_ideal (this is a simplified model, but it reveals the core relationship: the rotation matrix R connects the theoretical field and the observed field). Finally, SVD decomposition is performed. To find the optimal rotation matrix R from B_measured and B_ideal, the problem can be transformed into: Find R to minimize || B_measured - R B_ideal ||_F (where F denotes the Frobenius norm). The solution to this standard problem can be obtained using SVD (Singular Value Decomposition): for matrix H = B_measured B_ideal Performing SVD decomposition: [U, S, V] = SVD(H), the optimal rotation matrix R is: R = U V Ensure a right-handed coordinate system: Check if det(R) is +1. If it is -1, the algorithm needs to be modified (e.g., invert the last column of V). Output the rotation matrix R[3x3]. This matrix describes the rotation from the FG coordinate system to the sensor coordinate system.

[0042] From Rotation Matrix to Euler Angles: If you need more easily understood azimuth, elevation, and roll angles, you can decompose the rotation matrix R into Euler angles rotating around ZYX (or another sequence). Note that Euler angles suffer from gimbal lock, but in most applications, rotation matrices are a more commonly used and stable form of attitude representation.

[0043] By fully implementing the above four steps, the nine raw magnetic signals from the three coils of the electromagnetic field generation module and the positioning sensor are converted into 6-DOF pose coordinates.

[0044] The esophageal electromagnetic navigation system is used in the following esophageal electromagnetic navigation device 500, see [link to device 500]. Figure 4The device includes a system control unit 510, a positioning sensor interface unit 520, a magnetic field generator 530, a positioning guide wire 540, a medical imaging component 550, and a body position detector 560. The positioning guide wire 540 and the body position detector 560 are connected to the positioning sensor interface unit 520. The system control unit 510 is connected to the medical imaging component 550. The positioning sensor interface unit 520 and the magnetic field generator 530 are both coupled to the system control unit 510. The esophageal electromagnetic navigation device achieves sub-millimeter-level spatial positioning through the magnetic field generator and sensors, and its control unit is the core module of the system. The system control unit employs the esophageal electromagnetic navigation system described above, including: a multi-channel magnetic field generator drive circuit, each channel containing a programmable H-bridge and an overcurrent protection module; an environmental magnetic field monitoring subsystem configured to acquire background magnetic field noise in real time; a dynamic frequency adjuster that switches the operating frequency in the range of 5-20kHz according to the type of connected sensor; and an embedded calibration module that stores magnetic field distortion correction parameters for the surgical area.

[0045] Example 1: This example uses a 65-year-old male patient weighing 60 kg. During a routine physical examination, a high-resolution computed tomography (CT) scan revealed an isolated target point in his esophagus. Imaging characteristics showed that the vertical distance from the center of this location to the nearest visceral esophagus was approximately 12 mm, making it difficult to visualize visually under conventional esophagoscopy. The specific procedure for determining the target point using the aforementioned esophageal electromagnetic navigation system is as follows: S1: Scale module.

[0046] S1-1: The image acquisition unit of the planning module acquires enhanced CT scan data of the patient's chest and upper abdomen. The CT data is imported into the 3D reconstruction unit, where image segmentation algorithms (such as region growing and level sets) automatically or semi-automatically reconstruct a detailed 3D model of the esophagus, surrounding blood vessels (such as the azygos vein and right lower pulmonary vein), and stomach (to the pylorus). The physician identifies and marks the target lesion (target point) on the 3D model. Subsequently, the system automatically or the physician manually marks four key feature points on the esophageal structure of the 3D model: point A (esophagus at the thoracic inlet), point B (where the azygos vein arch compresses the esophagus), point C (esophagus at the level of the right lower pulmonary vein), and point D (the esophagogastric junction, i.e., the pylorus). These points have fixed anatomical locations and are not easily moved significantly with respiration and peristalsis.

[0047] S1-3: Target point planning: The system automatically calculates the navigation path from the esophageal opening to the target point based on the identified target point location.

[0048] S2: Registration module.

[0049] S2-1: Electromagnetic Navigation System Setup: The patient is brought into the operating room and placed in a supine position to fully expose the right esophagus. After successful anesthesia, endotracheal intubation is performed, and single esophageal ventilation is initiated (left esophageal ventilation, right esophageal collapse). The electromagnetic field generating module is placed below the patient's right chest, ensuring that the generated electromagnetic field stably and uniformly covers the entire surgical area. The ventilator unit of the positioning module is activated, and air is blown into the patient's esophagus according to preset parameters (e.g., peak airway pressure 20 cmH2O, PEEP 10 cmH2O, maintenance time 10 seconds), temporarily bringing the esophagus into a stable and dilated state.

[0050] S2-2: Esophageal Feature Point Acquisition: Establishing a standard esophagoscopy operating channel. A handheld navigation probe with an integrated positioning sensor at its tip is inserted into the esophagus through the main operating port. Under real-time monitoring by the esophagoscopy camera, the navigation probe is manipulated, guided by the display module, to sequentially position the tip of the probe at four feature points A, B, C, and D within the patient's body. Each successful contact with a point is confirmed by pressing the foot pedal, and the real-time spatial coordinates of that point in the electromagnetic field coordinate system are immediately recorded.

[0051] S2-3: Feature Point Registration: After acquiring the coordinates of the four points, the registration module automatically starts the Rigid CPD (Coherent Point Drift) registration algorithm. This algorithm iteratively matches the three physical space coordinate point sets acquired intraoperatively with the three CT model coordinate point sets planned preoperatively. By minimizing the root mean square error between the two sets of point sets through transformation (rotation and translation only), the algorithm accurately calculates a unique transformation matrix within seconds. After registration, the virtual CT 3D model is accurately aligned with the patient's actual anatomical structure in the surgical space, and a stable, global navigation coordinate system is established.

[0052] S3: Navigation module.

[0053] S3-1: Real-time Tracking and Guidance: On the display module, the navigation software interface integrates a semi-transparent 3D model of the patient with real-time esophagoscopy images. A virtual probe model is displayed synchronously in the 3D model in real time, with its position and orientation completely consistent with the physical probe. The system automatically highlights the target—esophageal positioning point P1—and displays a navigation interface similar to a "crosshair" for guidance.

[0054] S3-2: The physician continues to operate the esophagoscope and navigation probe to move towards the target point. The control unit continuously receives data from the positioning sensors, calculates and updates the probe's position in the 3D model in real time. The system guides the physician to adjust the probe's direction using graphics (such as distance bars and directional arrows) and / or sound until the virtual probe coincides with the target point in the 3D model. At this point, the tip of the real probe has accurately reached the lesion location, allowing for precise biopsy or treatment.

[0055] Therefore, the above technical solution uses the esophageal structure as the registration reference, which has high registration accuracy and high reliability. It does not depend on the tracheal path and can accurately locate any position in the esophagus, especially the position close to the esophagus, the position that is blocked by bones or the position that is not accessible by airway. At the same time, the entire positioning process does not require puncture of the esophageal tissue, which greatly improves safety.

[0056] The present invention relates to Electromagnetic Navigation Esophageal (ENE), a cutting-edge interventional technique for deep lesions in the mediastinum and periesophageal region (such as lymph nodes, early esophageal cancer, and neurogenic tumors). It combines preoperative high-resolution imaging (such as CT / MRI) with real-time electromagnetic positioning during surgery, providing doctors with a "see-through eye" and "GPS" that surpasses the traditional endoscopic field of vision. It has profound significance in empowering doctors' operational capabilities and improving navigation accuracy.

[0057] In addition to the above embodiments, the present invention may have other implementation methods; all technical solutions formed by equivalent substitution or equivalent transformation fall within the protection scope claimed by the present invention.

[0058] The beneficial effects of the embodiments of the present invention are as follows: 1. The ENE system displays the position of the catheter / needle tip with electromagnetic sensors in real time on a 3D view and precisely overlays it onto the preoperatively reconstructed personalized 3D anatomical model of the patient. Doctors can intuitively see the spatial relationship between the instrument and the target lesion, realizing a leap from "presumption" to "direct vision" and greatly reducing the blindness of the operation.

[0059] 2. Achieved true three-dimensional real-time navigation: Through electromagnetic navigation technology, the preoperative CT three-dimensional model is perfectly integrated with the real-time spatial position during the operation, enabling doctors to "see through" the esophageal wall and clearly see the precise position of the probe in three-dimensional space and its relative relationship with the target lesion, guiding the operation directly to the target.

[0060] 3. Improved accuracy and efficiency of examination: The system can accurately guide biopsy or treatment instruments to small or deep lesions, greatly reducing repeated operations and missed lesions caused by inaccurate positioning, improving the detection rate of early esophageal cancer, and shortening the operation time.

[0061] 4. By providing clear and intuitive navigation paths and real-time location feedback, ENE significantly lowers the technical barrier to complex mediastinal punctures. Even physicians with relatively little experience can complete the procedure more safely and accurately under the system's guidance, which helps to promote standardized precision medicine procedures in different medical centers.

Claims

1. An esophageal electromagnetic navigation system, characterized in that, The system includes: The planning module is used to construct a three-dimensional model of the patient's esophagus based on computed tomography image data of the esophageal region, determine the location of the target point in the three-dimensional model, and use the esophageal structure as the registration reference. Based on the location of the target point, the module marks the feature points for registration in the esophageal structure. The feature points have relatively fixed anatomical locations. The registration module is used during surgery to obtain the actual spatial coordinates of feature points through a navigation probe with positioning sensors, and match them with the coordinates of the corresponding feature points in the 3D model to construct an electromagnetic navigation coordinate system that associates the 3D model with the real patient space. The navigation module is used to track the position of the navigation probe with positioning sensors in real time based on the constructed electromagnetic navigation coordinate system, and guide it to move to the target point in the patient's esophagus. The positioning module of the internal electromagnetic field generation module is used to blow air into the patient's esophagus based on preset target parameters during the operation of the registration module and the navigation module. By controlling the air pressure in the esophagus, the esophageal shape is stabilized, its peristalsis is inhibited, and it is kept in a relatively expanded and fixed state, thereby significantly improving the accuracy of registration and navigation. Both the electromagnetic field generating module and the positioning sensor have coils inside. The electromagnetic field generating module generates a stable electromagnetic field covering the surgical area where the patient is located. The positioning sensor senses the magnetic field generated by the electromagnetic field generating module to generate an electromagnetic field coordinate system, which enables the navigation probe to obtain the actual spatial coordinates of the feature points. Guided by an esophagoscope, the tip of the navigation probe gently touches the feature points on the inner wall of the patient's esophagus that correspond to the planning module. Each time a point is touched, its spatial coordinates in the electromagnetic field coordinate system are recorded. After recording the spatial coordinates of all feature points, the registration module starts the registration algorithm to match the physical spatial coordinate set of the collected feature points with the corresponding point coordinate set calibrated on the 3D model. By minimizing the distance error between the two sets of point sets, the algorithm accurately calculates a transformation matrix. This matrix defines the transformation relationship from the CT image coordinate space to the actual physical space of the surgery, realizing the electromagnetic navigation coordinate system that links the 3D model with the actual patient space.

2. The esophageal electromagnetic navigation system as described in claim 1, characterized in that, The feature points include the thoracic inlet, the level of the zygomatic vein, the level of the right lower pulmonary vein, and the pylorus.

3. The esophageal electromagnetic navigation system as described in claim 1, characterized in that, A medical image segmentation framework based on the combination of Transformer and U-Net utilizes the powerful global context modeling capabilities of Transformer to compensate for the skip connection structure of traditional U-Net and restore local details. The computed tomography image data is processed, and the three-dimensional model of the patient's esophagus is automatically or semi-automatically reconstructed using medical image processing software.

4. The esophageal electromagnetic navigation system as described in claim 1, characterized in that, The navigation probe is inserted into the patient's esophagus along with the esophagoscope, and the positioning sensor acquires the three-dimensional spatial coordinates and attitude information of the navigation probe in real time.

Images