Multi-modal image fusion chordoma invasion boundary real-time navigation system

Abstract

The invention relates to the field of medical instruments, and discloses a multi-modal image fusion chordoma invasion boundary real-time navigation system which comprises an integrated optical flow acoustic detection probe, an optical positioning tracking device, a pneumatic negative pressure control device, a frequency sweep optical coherence tomography host, an augmented reality display terminal and a main control computer terminal. The main control computer terminal obtains and fuses the tissue rheological characteristics analyzed by the pneumatic negative pressure control device and the tissue hardness characteristics analyzed by the sweep frequency optical coherence tomography host to generate a tissue identification result; and in combination with the probe position provided by the optical positioning tracking device, an augmented reality navigation image is generated and displayed on an augmented reality display terminal in real time. According to the invention, through fusion analysis of multi-modal physical characteristics, the accuracy of boundary identification is improved; through an augmented reality technology, intuitive visual guidance in an operation is realized; and the resection boundary can be dynamically recommended, so that the operation safety is enhanced.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, specifically to a real-time navigation system for chordoma invasion boundaries based on multimodal image fusion. Background Technology

[0002] Chordoma is a locally invasive tumor originating from embryonic notochord remnants. Its growth often invades and encircles the neural and vascular structures of the skull base and spinal region, making surgical resection challenging. The extent of tumor resection is a decisive factor affecting postoperative recurrence rate and prognosis. However, due to the lack of a clear macroscopic boundary between the tumor tissue and its microscopic invasion zone, accurately identifying its boundaries during surgery remains a technical challenge in the field of surgery.

[0003] Existing intraoperative tissue identification techniques, such as intraoperative ultrasound, optical coherence tomography, or elastography, typically rely on a single physical parameter, such as acoustic impedance, optical scattering properties, or tissue stiffness. For tumors like chordomas, which exhibit high tissue heterogeneity, their physical characteristics are similar to those of surrounding edematous or fibrotic tissues in different regions.

[0004] Therefore, methods that rely on a single parameter are prone to producing ambiguous identification results in the transition zone between tumors and normal tissues, resulting in insufficient accuracy in determining the boundaries of microscopic invasion.

[0005] In surgical guidance, traditional surgical navigation systems display preoperative images and the positions of surgical instruments on a separate external monitor. This information separation requires the surgeon to frequently shift their gaze between the surgical field and the external monitor, and to perform coordinate matching between the surgical space and the image space. This not only disrupts the continuity of the surgical procedure but also increases the surgeon's cognitive load, especially when fine boundary discrimination is required, as the real-time and positional nature of the guidance information is insufficient.

[0006] Furthermore, when determining the final resection extent, the setting of the safe margin distance relies primarily on the surgeon's clinical experience and interpretation of static preoperative imaging. This decision-making process is statically set, lacking a quantitative mechanism for dynamic adjustment based on real-time intraoperative tissue property data and the immediate relative positions of surgical instruments with high-risk anatomical structures (such as the internal carotid artery and brainstem). This makes it difficult to quantify the balance between ensuring radical tumor resection and avoiding damage to key functional structures, increasing the complexity and potential risks of surgical decision-making. Summary of the Invention

[0007] The purpose of this invention is to provide a real-time navigation system for chordoma invasion boundaries based on multimodal image fusion, which aims to solve the problem of insufficient intraoperative real-time, in-situ, and quantitative identification of the microscopic invasion boundaries of chordoma in the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A real-time navigation system for chordoma invasion boundaries based on multimodal image fusion includes: an integrated optical flow acoustic detection probe, an optical positioning and tracking device, a pneumatic negative pressure control device, a swept-frequency optical coherence tomography (OCT) host, an augmented reality display terminal, and a main control computer terminal. The main control computer terminal establishes data communication connections with each device and the host.

[0009] The main control computer terminal functions to receive and process data from various components. Based on data acquired from the pneumatic negative pressure control device, it analyzes the rheological characteristics of the tissue; based on data acquired from the scanned frequency optical coherence tomography (OCT) host, it analyzes the tissue stiffness characteristics. Subsequently, the main control computer terminal fuses the rheological and stiffness characteristics to generate a tissue identification result characterizing the tumor invasion boundary. Finally, based on this tissue identification result and the real-time probe position provided by the optical positioning and tracking device, it generates an augmented reality navigation image and drives the augmented reality display terminal to display it.

[0010] As a specific implementation method of this technical solution: The integrated optical flow acoustic detection probe has a specific structure. It includes a mid-section rigid extension sleeve assembly and a distal optical flow coupled detection head. The sleeve assembly adopts a double-layer coaxial metal tube structure, including an outer stainless steel sleeve and an inner protective sleeve. The gap between the two sleeves forms an annular negative pressure channel for communication with a pneumatic negative pressure control device. The detection head adopts a non-flush end face design, with the end face of the outer stainless steel sleeve extending axially beyond the end face of the inner protective sleeve. This structure allows a gas-liquid coupling cavity for stable detection to be formed between the optical exit surface of the inner protective sleeve and the contact tissue.

[0011] To obtain the rheological characteristics of the tissue, the pneumatic negative pressure control device incorporates a high-precision differential pressure sensor and a thermal mass flow meter. A rheological parameter calculation module within the main control computer terminal calculates the dynamic rheological impedance index and the fluid behavior index based on the instantaneous pressure difference data measured by the high-precision differential pressure sensor and the fluid volumetric flow rate data measured by the thermal mass flow meter, in order to determine the tissue type.

[0012] To obtain the tissue stiffness characteristics, the integrated optical flow acoustic probe also includes a piezoelectric actuation element for exciting micro-shear waves. The swept-frequency optical coherence tomography host detects the micrometer-level displacement of the tissue caused by shear wave propagation by analyzing the phase term changes in the interference signal. An elastic imaging inversion module within the main control computer terminal calculates the Young's modulus of the tissue based on the propagation velocity of the shear wave within the tissue detected by optical methods.

[0013] In the surgical navigation process, the main control computer terminal first performs image space registration. Its internal image rigid registration algorithm unit uses an automatic registration algorithm based on mutual information to map the coordinate space of the magnetic resonance imaging data to the coordinate space of the computed tomography (CT) scan data, thereby constructing a preoperative virtual 3D model. Subsequently, the spatial registration logic unit uses a least-squares point cloud matching algorithm to solve for the rotation matrix and translation vector, thus establishing the geometric mapping relationship between the surgical physical space and the image virtual space.

[0014] To achieve intelligent switching of navigation modes, the main control computer terminal marks the tumor outline on the constructed preoperative virtual 3D model and extends it to generate a microscopic detection activation zone. During the operation, when the optical positioning and tracking device tracks the tip of the integrated photoacoustic probe into this activation zone, the main control computer terminal automatically triggers a signal, switching the entire system's operating mode from macroscopic geometric navigation to microscopic physical property detection mode.

[0015] In the microscopic physical property detection mode, the main control computer terminal performs feature fusion through a multimodal feature fusion unit. This unit first constructs a multidimensional feature vector from calculated parameters such as the effective Young's modulus, fluid behavior index, and consistency coefficient. Next, a classification decision engine calculates the statistical distance between the current feature vector and a pre-stored chordoma benchmark feature library, thereby deriving a normalized chordoma similarity index. Finally, based on this index, a hierarchical judgment logic is executed, outputting the specific tissue classification label for the current detection point.

[0016] To visualize the detection results, the main control computer terminal generates augmented reality navigation images through a visualization rendering logic unit. Within this logic unit, an organization attribute texture mapping engine uses a non-linear color transfer function to map the aforementioned chordoma similarity index onto corresponding mesh vertices on the surface of the preoperative three-dimensional anatomical model, assigning each mesh vertex a color value that varies between a marker color representing normal soft tissue and a warning color representing chordoma tissue.

[0017] To achieve the overlay of virtual information with real-world scenes, the visualization rendering logic unit also includes an augmented reality viewport compositing engine. This engine employs an adaptive transparency control algorithm, dynamically calculating a compositing opacity coefficient based on the chordoma similarity index corresponding to each pixel. Subsequently, an alpha blending operation is performed, compositing the virtual color layer generated by the tissue attribute texture mapping engine with the real video layer captured by equipment such as surgical microscopes.

[0018] Furthermore, this technical solution also provides an auxiliary decision-making function. The main control computer terminal also includes a dynamic surgical boundary calculation engine. This engine calculates and recommends a resection margin distance in real time based on the chordoma similarity index, the effective Young's modulus of the tissue, and the nearest distance of the current probe point to high-risk anatomical structures such as the brainstem and internal carotid artery. This calculation process includes a distance-based safety penalty function, the value of which decays as the probe point approaches the high-risk anatomical structure; the calculation result is used to narrow the recommended resection range. This invention provides a real-time navigation system for chordoma invasion boundaries based on multimodal image fusion. It offers the following advantages: 1. This invention utilizes an integrated optical flow acoustic detection probe, combined with a pneumatic negative pressure control device and a swept-frequency optical coherence tomography host, to simultaneously acquire tissue hardness characteristics (Young's modulus) and rheological characteristics (fluid behavior index, etc.). The multimodal feature fusion unit of the main control computer terminal fuses and analyzes these two different physical dimensions of features. Compared with detection methods that rely on only a single physical parameter, this invention can more effectively distinguish chordoma tissue with complex heterogeneity from surrounding normal tissue, reducing misjudgments caused by overlapping tissue characteristics.

[0019] 2. This invention utilizes an optical positioning and tracking device to obtain the precise spatial position of the probe, and renders the real-time calculated chordoma similarity index onto a 3D model in the form of a continuous color spectrum through a visualization rendering logic unit. The augmented reality viewport synthesis engine overlays and fuses this virtual color layer with the real surgical field of view under the microscope, enabling doctors to directly see the microscopic boundary distribution of the tumor in the surgical area, transforming abstract physical data into intuitive visual information, thereby accelerating the judgment speed and improving the smoothness of the surgical procedure.

[0020] 3. This invention identifies boundaries through a dynamic surgical boundary calculation engine. Based on tissue attributes and the distance between the probe point and high-risk anatomical structures, it determines the resection margin distance. A safety penalty function ensures that when the probe point is close to critical structures such as the brainstem or internal carotid artery, the system automatically suggests a more conservative resection range. This provides doctors with quantitative decision support between pursuing radical resection and protecting important functional areas, thereby effectively reducing the risk of surgical complications while ensuring the resection range. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart of the negative pressure rheological feedback subsystem of the present invention; Figure 3 This is a schematic diagram illustrating the spatial registration and real-time tracking coordinate transformation relationship of the present invention. Detailed Implementation

[0022] The technical solutions in 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.

[0023] See attached document Figure 1 The present invention provides a real-time navigation system for chordoma invasion boundaries based on multimodal image fusion. The system includes: a main control computer terminal, a multimodal image storage unit, an optical positioning and tracking device, an integrated optical flow acoustic detection probe, a swept-frequency optical coherence tomography host, a pneumatic negative pressure control device, and an augmented reality display terminal.

[0024] As the core of the system, the main control computer terminal establishes data communication connections with each component and is responsible for performing data acquisition, multimodal data fusion, rheological characteristic analysis, elastic modulus inversion, and visualization rendering.

[0025] The macroscopic navigation function is achieved through the collaboration of algorithms within the main control computer terminal and external optical positioning and tracking devices, aiming to guide the integrated optical flow acoustic detection probe to the preset target area.

[0026] First, the main control computer terminal retrieves the patient's computed tomography (CT) and magnetic resonance imaging (MRI) data from the multimodal image storage unit. To achieve multimodal image fusion, the image rigid registration algorithm unit inside the main control computer terminal uses an automatic registration algorithm based on mutual information to map the coordinate space of the MRI data to the coordinate space of the CT data. This algorithm iteratively optimizes the spatial transformation matrix to maximize the mutual information value between the two sets of images; its calculation model is as follows: ; in, The mutual information value between CT image A and MRI image B; and The Shannon entropy of the edges of the two images are respectively; Let be the joint entropy of the two images.

[0027] After registration, the system constructs a preoperative virtual 3D model containing complete bone and soft tissue information. The main control computer terminal marks the tumor outline on the model and generates a microscopic detection activation area by extending it outward by a preset distance (e.g., 5 mm).

[0028] At the start of the surgery, a mapping relationship needs to be established between the physical surgical space and the virtual imaging space. The optical positioning and tracking device includes an infrared binocular camera and a reference frame fixed to the patient's head. The main control computer terminal acquires the physical coordinates of anatomical landmarks on the patient's surface and the reflective tracer sphere on the reference frame through the infrared binocular camera, and matches them with the corresponding virtual coordinates in the model. The optimal rotation matrix is ​​solved using a least-squares point cloud matching algorithm. Translation vector Its objective function is as follows: ; In the above formula, The objective function representing the registration error; Indicates the number of anatomical landmarks selected; Indicates the first The position vector of each marker point in the patient coordinate system; Indicates the first The position vector of each marker point in the preoperative virtual 3D model coordinate system; Indicates the problem to be solved Rotation matrix; Indicates the problem to be solved Translation vector. The main control computer terminal minimizes the above objective function. Determine the optimal rotation matrix Translation vector This allows for precise alignment between the surgical space and the imaging space.

[0029] During the procedure, the optical positioning and tracking device tracks the position of the integrated optical flow acoustic detection probe in real time, and the main control computer terminal maps it onto the three-dimensional model according to the transformation matrix. When the probe tip enters the microscopic detection activation area, the main control computer terminal automatically triggers a signal, and the system switches from macroscopic geometric navigation to microscopic physical detection mode.

[0030] The microscopic detection function is achieved through the coordinated operation of a pneumatic negative pressure control device and a swept-frequency optical coherence tomography host, which is used to analyze the multimodal physical properties of tissues in situ.

[0031] Pneumatic negative pressure control and rheological measurement: The pneumatic negative pressure control device is connected to an integrated photoacoustic detection probe via a flexible pneumatic conduit to establish and maintain a precise fluid measurement environment. Internally, it includes a medical-grade silent vacuum pump, a negative pressure buffer tank, an electronically controlled proportional solenoid valve, a high-precision differential pressure sensor, and a thermal mass flow meter. When the probe contacts the tissue, the system employs closed-loop feedback control logic; the electronically controlled proportional solenoid valve adjusts the conduit opening based on real-time feedback from the sensor to maintain a stable adsorption pressure. The flow balance equation for the pneumatic circuit is as follows: ; in, This indicates the gas flow rate entering the negative pressure circuit through the electronically controlled proportional solenoid valve. Indicates the flow coefficient of an electronically controlled proportional solenoid valve; This indicates the current opening cross-sectional area of ​​the electronically controlled proportional solenoid valve. Indicates standard atmospheric pressure; This indicates the absolute pressure inside the negative pressure buffer storage tank; This indicates air density.

[0032] The rheological parameter calculation module within the main control computer terminal calculates the dynamic rheological impedance index based on instantaneous pressure difference data provided by the pressure sensor and fluid volumetric flow rate data provided by the flow sensor, used to distinguish different tissues. Its calculation model is as follows: ; in, The dynamic rheological impedance index; This refers to the instantaneous air pressure difference; This refers to the fluid volumetric flow rate; A fluid behavior index characterizing the non-Newtonian properties of a fluid.

[0033] Optical coherence tomography and elasticity measurement: The swept-frequency optical coherence tomography (OCT) system is connected to an integrated optical flow acoustic probe via a single-mode fiber to perform high-sensitivity microstructure imaging and elastic wave detection. Its core components include a wavelength-scanning laser source (center wavelength 1310 nm), a Mach-Zehnder fiber interferometer, and a balanced photodetector. After the probe illuminates the tissue, the collected backscattered light interferes with the reference light, and the balanced photodetector receives this interference signal. The expression for the output photocurrent signal is as follows: in, Indicates wave number Photocurrent intensity at that time; Indicates the responsivity of a balanced photodetector; The power spectral density function representing a wavelength-scanning laser source; Indicates the reflectivity of the reference arm; Indicates the depth in the sample arm is The reflectivity of the scattering body at that location; This indicates the depth of the scatterer relative to the position of zero optical path difference; This represents the additional phase determined by the properties of the scatterer.

[0034] The swept-frequency optical coherence tomography (OCT) host analyzes the phase term in the signal. Changes in these changes enable the detection of micron-level displacements in tissues.

[0035] To perform elasticity measurements, the system also includes a waveform generator and drive amplifier circuit synchronized with the optical scanning clock. The waveform generator produces a signal of a specific frequency (0.5-2 kHz), which, after amplification, drives the piezoelectric actuator within the probe to excite micro-shear waves within the tissue. The elasticity imaging inversion module in the main control computer terminal calculates the Young's modulus of the tissue based on the propagation velocity of the shear waves detected optically.

[0036] The image fusion and rendering module within the main control computer terminal integrates rheological features from the rheological parameter calculation module, stiffness features from the elasticity imaging inversion module, and spatial coordinates from the optical positioning and tracking device. This multimodal data is fused to generate an augmented reality image containing probability information about tumor invasion boundaries. This image is then overlaid in real-time on the surgical microscope's field of view or an external monitor by the augmented reality display terminal, providing surgeons with intuitive navigation and boundary determination information.

[0037] The integrated optical flow acoustic detection probe mainly consists of a proximal handle housing, a mid-section rigid extension sleeve assembly, and a distal optical flow coupling detection head. The mid-section rigid extension sleeve assembly employs a double-layer coaxial metal tube structure, including an outer stainless steel sleeve and an inner protective sleeve. A continuous annular negative pressure channel is formed between the two sleeves, with a hydraulic diameter of... The following geometric relationship must be satisfied: ; in, Indicates the hydraulic diameter of the annular negative pressure channel; Indicates the inner diameter of the outer stainless steel sleeve; This indicates the outer diameter of the inner protective sleeve.

[0038] In this embodiment, The value is set between 0.2 mm and 0.5 mm to generate sufficient capillary adsorption effect and prevent large bone fragments from clogging the channel.

[0039] The remote optical flow coupling probe adopts a non-flush end face design, with the end face of the outer sleeve extending axially and exceeding the end face of the inner protective sleeve by a predetermined distance. This creates a gas-liquid coupling cavity between the optical exit surface of the inner sleeve and the contacting tissue. This cavity not only adsorbs and stabilizes soft tissue under negative pressure but also constructs an immersion optical path with refractive index matching. Its static volume... The definition is as follows: ; in, This represents the theoretical volume of the gas-liquid coupling cavity; Indicates the inner diameter of the outer stainless steel sleeve; This indicates the axial protrusion distance (i.e., the recess depth) of the distal end face of the outer stainless steel sleeve relative to the distal end face of the inner protective sleeve. In this embodiment, It is set to 0.5 mm to 1.0 mm.

[0040] The optical detection assembly is coaxially encapsulated within an inner protective sleeve and consists sequentially of a single-mode optical fiber, a coreless fiber expander, and a gradient refractive index (GRIN) lens. The GRIN lens is responsible for focusing the beam emitted from the fiber onto the tissue interior, where its internal refractive index distribution... Follow the model below: ; in, Indicates the distance from the center of the optical axis The refractive index at that point; This represents the refractive index at the central axis of a gradient refractive index lens; This represents the gradient constant, expressed in units of square millimeters, which determines the focusing ability of the lens. This indicates the radial distance from the central optical axis of the lens.

[0041] To ensure the focal point can penetrate the gas-liquid coupling cavity and reach deep into the trabecular bone space, the working distance... The following inequality constraints must be satisfied: ; in, Indicates the working distance of the optical detection component in the medium; This indicates the axial protrusion distance (i.e., the depth of the gas-liquid coupling cavity) of the distal end face of the outer stainless steel sleeve relative to the distal end face of the inner protective sleeve as defined in the foregoing embodiments. This indicates the preset tissue detection depth (e.g., 0.5 mm to 2.0 mm). This represents the average refractive index of human tissue. The lateral resolution of the optical detection component determines the system's ability to identify minute tumor infiltration lesions. This lateral resolution is determined by the beam waist diameter of the focused spot. According to Gaussian beam propagation theory, the beam waist radius of the optical detection component at the focal point... Determined by the following formula: The lateral resolution of the optical detection component determines the system's ability to identify minute tumor infiltration lesions. This lateral resolution is determined by the beam waist diameter of the focused spot. According to Gaussian beam propagation theory, the beam waist radius of the optical detection component at the focal point is... Determined by the following formula: ; in, This indicates that the electric field amplitude at the focal point decreases to the central value. The horizontal radius at that time; Indicates the center wavelength of the light source; Indicates the effective focal length of a gradient refractive index lens; It represents the refractive index of the fluid (such as physiological saline or tissue fluid) filling the gas-liquid coupling cavity; This represents the radius of the light spot incident on the near end face of the gradient refractive index lens.

[0042] Rayleigh length of optical detection components It is given by the following formula: ; in, This represents the Rayleigh length, which is the axial distance when the beam cross-sectional area doubles. The integrated optical flow acoustic detection probe adjusts the length of the coreless fiber extender to achieve this. to control and This makes Rayleigh length The typical thickness range (e.g., 1 mm to 2 mm) covering the invasive boundary of a chordoma ensures complete recording of the shear wave propagation process within a single scan depth without the need for mechanical focusing.

[0043] The annular negative pressure channel, as a core component of precision rheological measurement, has its geometry optimized to ensure the Reynolds number when the fluid is drawn in. If the Reynolds number is less than 2300, maintain laminar flow and avoid turbulent interference. The Reynolds number calculation formula is: ; in, Represents the Reynolds number; This indicates the density of inhaled fluids (such as blood, saline solution, or liquefied tumor tissue); This indicates the average flow velocity within the annular negative pressure channel; This indicates the hydraulic diameter of the annular negative pressure channel (i.e., the difference between the inner diameter of the outer stainless steel sleeve and the outer diameter of the inner protective sleeve). This indicates the apparent viscosity of the fluid being drawn in.

[0044] This channel provides a constant shear environment for the flowing tissue fluid or liquefied tumor tissue. The characteristic shear rate generated by the channel walls... Defined by the following formula: ; in, This represents the wall shear rate, measured in reciprocal seconds. This represents the real-time volumetric flow rate measured by the flow sensor; Indicates the inner diameter of the outer stainless steel sleeve; This represents the outer diameter of the inner protective sleeve. The formula indicates that the pneumatic negative pressure and flow control components, through fixed geometric parameters, will affect the measured flow rate. This is directly mapped to the shear rate applied to the tissue. The main control computer terminal utilizes this shear rate. By combining the measured pressure difference, the non-Newtonian viscosity coefficient of the fluid can be inverted according to the power-law fluid model.

[0045] The pneumatic negative pressure and flow control components create a gas-liquid dynamic sealing effect at the distal optical-fluid coupling probe. When the integrated optical-fluid acoustic probe is not in contact with tissue, the annular negative pressure channel draws in air with minimal flow resistance. When the end face of the integrated optical-fluid acoustic probe contacts soft tissue, the negative pressure adsorption pulls the tissue into the gas-liquid coupling cavity. At this point, the pneumatic negative pressure and flow control components utilize the inner wall of the outer stainless steel sleeve as the fluid boundary, the end face of the inner protective sleeve as the optical window boundary, and the surface of the drawn-in soft tissue as the elastic boundary, together sealing off a micro-level fluid control domain.

[0046] Within this fluid control domain, pneumatic negative pressure and the flow control components perform continuous self-cleaning cycles. Due to the coaxial gap between the outer stainless steel sleeve and the inner protective sleeve, the negative pressure airflow forms a uniform centripetal suction flow field in the circumferential direction of the gas-liquid coupling cavity. This suction flow field rapidly draws turbid blood and free bone fragments from the gas-liquid coupling cavity into the annular negative pressure channel and expels them from the body, while simultaneously guiding transparent interstitial fluid or artificially dripped physiological saline to fill the gas-liquid coupling cavity. This fluid displacement mechanism ensures that the optical path in front of the optical detection components is always filled with a low-scattering medium.

[0047] In addition, the pneumatic negative pressure and flow control assembly is equipped with an anti-backflow check valve. This anti-backflow check valve is located between the negative pressure distribution chamber and the Luer conical connector, and employs a duckbill or umbrella-shaped silicone valve design. The anti-backflow check valve is configured to automatically close the air passage in the event of a stoppage of the precision negative pressure pump or an unexpected power outage, preventing contaminated waste fluid accumulated in the negative pressure tubing from flowing back into the integrated photoacoustic detection probe, thus ensuring the aseptic safety of the surgical procedure.

[0048] The core of the acoustic / mechanical excitation assembly is a tubular piezoelectric ceramic actuator coaxially sleeved and fixed to the outer wall of the inner protective sleeve. When an AC drive voltage is applied, the actuator generates axial extension and contraction based on the inverse piezoelectric effect, driving the inner protective sleeve to perform a piston-like reciprocating motion.

[0049] Under negative pressure adsorption, the tissue closely adhering to the end face of the inner sheath experiences a micron-level displacement disturbance perpendicular to the surface, thereby exciting radially propagating shear waves within the tissue. The instantaneous axial displacement of the inner sheath tip... It follows the following dynamic equations: ; in, This indicates the instantaneous axial displacement at the top of the inner protective sleeve; This indicates the effective number of stacks in the tubular piezoelectric ceramic actuator (1 if a single-layer tube structure is used). This represents the longitudinal piezoelectric strain constant of a piezoelectric material, expressed in meters per volt. This represents the applied instantaneous driving voltage function, typically . This represents the reaction force of the tissue on the probe tip, and this force is related to the tissue's viscoelasticity and the degree of negative pressure adsorption. It represents the overall equivalent axial stiffness of the acoustic / mechanical excitation assembly.

[0050] This formula reveals that the design of acoustic / mechanical excitation components must balance the driving voltage with system stiffness. In this embodiment, the system monitors... (Based on the aforementioned rheological parameters) dynamically adjusted The amplitude, ensure It always maintains a micro-perturbation range of 1 micrometer to 10 micrometers.

[0051] To ensure the accuracy of the shear wave phase, the natural frequency of the excitation component... It is designed to operate at frequencies well above the operating range (0.5-2 kHz). Its first-order longitudinal natural frequency is calculated as follows: ; in, This represents the first-order longitudinal natural frequency of the inner protective sleeve assembly; Indicates the effective cantilever length of the inner protective sleeve; This indicates the Young's modulus of the inner protective sheath material (e.g., medical stainless steel or titanium alloy). This indicates the density of the inner protective sleeve material. In this embodiment, by shortening... Alternatively, high-rigidity materials can be selected, making The frequency range is set above 20 kHz, far exceeding the 0.5 kHz to 2 kHz band commonly used for shear wave excitation. This design strategy ensures that the acoustic / mechanical excitation components operate in a flat frequency response region within the working frequency band, avoiding shear wave phase distortion caused by the system's own resonance, thereby guaranteeing the accuracy of phase difference measurement in the Young's modulus inversion algorithm.

[0052] See attached document Figure 2 The negative pressure rheological feedback subsystem of the present invention is deployed on the main control computer terminal to realize closed-loop regulation of the pneumatic negative pressure control device and to calculate the rheological characteristics of the contact tissue in real time.

[0053] This subsystem receives signals from pressure and flow sensors in real time and outputs control signals to adjust the vacuum level. Its operating logic includes: Adaptive cleaning of the optical path: When the signal-to-noise ratio of the optical coherence tomography (OCT) signal is detected to be lower than a preset threshold, the subsystem determines that the optical path is blocked. At this time, the system drives the electronically controlled valve to generate an instantaneous high-flow-rate negative pressure pulse to forcibly remove blood or debris from the gas-liquid coupling chamber until the optical signal is restored.

[0054] Rheological feature scanning status: After the probe forms a closed adsorption with the tissue, the subsystem applies a time-varying pressure excitation (such as a triangular wave) and simultaneously records the pressure-flow dataset.

[0055] The rheological parameter calculation unit within the subsystem uses the generalized Hagen-Poiseuille equation to analyze this dataset: ; in, Indicates time Volumetric flow rate; Indicates time The driving pressure difference; This represents the equivalent hydraulic radius of the annular negative pressure channel; Indicates the effective length of the annular negative pressure channel; Fluid behavior indexes characterizing the non-Newtonian properties of fluids; This represents the fluid consistency coefficient.

[0056] By taking the logarithm of the above equation and performing linearization, the computing unit can process the collected data. and The data points are fitted linearly using the least squares method to solve the problem. and .according to Values ​​are organized and categorized: It is determined to be a Newtonian fluid (blood, cerebrospinal fluid, etc.).

[0057] (e.g., 0.3-0.6): It is determined to be a pseudoplastic non-Newtonian fluid with shear-thinning properties, which is a typical characteristic of chordoma mucus.

[0058] and High saturation value: identified as high impedance solid (bone tissue).

[0059] The calculated rheological eigenvectors ( , (etc.) will be transmitted to the multimodal feature fusion unit.

[0060] The optical coherence elastography subsystem of the present invention is used to coordinate the control of acoustic excitation and OCT host, and to solve the micromechanical properties of tissues from the interference spectrum.

[0061] The subsystem first synchronously triggers the acoustic excitation and the M-mode acquisition of the OCT host. In M-mode, the OCT system continuously acquires A-scan spectral data at fixed locations at a high line scan rate (≥50kHz) to capture the complete shear wave propagation process.

[0062] The phase demodulation unit within the subsystem processes the spectral data and extracts the axial vibrational velocity of tissue particles based on the phase difference between adjacent A-scan signals. Calculated using the following formula: ; in, Indicates depth as The organization particles at time axial vibration velocity; Indicates the center wavelength of the swept frequency light source; The average group refractive index of human tissue; This represents the time interval between two consecutive A-scan scans (i.e., the reciprocal of the line scan cycle). Indicates time In depth The phase value obtained by demodulation.

[0063] Subsequently, the wave velocity tracking computing unit uses algorithms such as cross-correlation analysis to identify the arrival time of the shear wave front at different depths on the generated spatiotemporal vibration map, and uses a linear regression model to fit the wave crest propagation trajectory to solve for the shear wave velocity. : ; in, Indicates the depth reached by the shear wave peak At that moment; This represents the propagation velocity of the shear wave to be solved; This indicates the initial moment of the shear wave on the tissue surface; Indicates the detection depth. The wave velocity tracking processing unit processes the measurement data points. Perform a least-squares fit, and the reciprocal of the slope of the fitted line is the average shear wave velocity in that region.

[0064] Finally, the elastic modulus inversion unit is based on the assumption of local isotropy and incorporates the real-time negative pressure value provided by the negative pressure rheological feedback subsystem. Calculate the effective Young's modulus of the tissue. : ; in, This represents the effective Young's modulus of the microstructure under the current negative pressure prestressing state, expressed in Pascals. This indicates the density of human tissue (usually taken as 1050 kg per cubic meter). Indicates negative pressure The shear wave velocity measured under the action.

[0065] The multimodal feature fusion unit of the present invention is used to fuse fluid dynamics and tissue mechanics data to generate tissue identification results with high confidence.

[0066] This unit first timestamps the data from different subsystems and then constructs a timescale containing the effective Young's modulus for each probe contact event. Fluid behavior index Consistency coefficient and optical signal-to-noise ratio Multidimensional feature vectors : ; in, This represents the effective Young's modulus obtained by inversion from the optical coherent elastic imaging subsystem; Fluid behavior indexes characterizing the non-Newtonian properties of fluids; Indicates the fluid consistency coefficient; The signal-to-noise ratio (SNR) of the optical signal is used to evaluate the validity of the detection data.

[0067] The classification decision engine within each unit uses a probability density-based method to calculate the statistical distance between the current feature vector and a pre-set chordoma benchmark feature library, thereby obtaining the chordoma similarity index. (Values ​​range from 0 to 1). The mathematical model for this index is as follows: ; in, The normalized chordoma similarity index ranges from 0 to 1. , , These represent the mean Young's modulus, mean fluid behavior index, and mean consistency coefficient of chordoma tissue in the baseline characteristic parameter library, respectively. , , , representing the standard deviations of the corresponding parameters, are used to define the tolerance of the feature distribution. Chordomas typically exhibit extremely low Young's modulus (soft) and a fluid behavior index less than 1 (shear-thinning mucus), and this formula accurately captures this combination of characteristics.

[0068] The multimodal feature fusion unit is based on the calculated chordoma similarity index. Execute the hierarchical determination logic: Level 1 criterion: When the optical signal-to-noise ratio When the value is below the preset confidence threshold, the multimodal feature fusion unit outputs an invalid / occluded status signal and triggers the aforementioned pneumatic cleaning process, without performing pathological classification.

[0069] Second-level judgment: When Meets the requirements and has an effective Young's modulus When the bone stiffness exceeds a preset threshold (e.g., 1 MPa), regardless of the rheological parameters, the multimodal feature fusion unit forcibly identifies it as bone tissue and outputs a color code (e.g., blue) representing safety. This is based on the physical fact that high-stiffness objects are not soft tissue tumors.

[0070] Level 3 Decision: When not belonging to bone tissue, multimodal feature fusion unit comparison. Compared with the preset judgment threshold (For example, 0.75). If The multimodal feature fusion unit determines that the detection area is a residual chordoma and outputs a color code (e.g., red) representing a warning and a tumor invasion probability value. The multimodal feature fusion unit further analyzes the fluid behavior index. .like If the value is close to 1 (a characteristic of Newtonian fluids), it is identified as cerebrospinal fluid / irrigation fluid; if... Although small If the value is higher than the average for tumors, it is considered inflammatory / fibrotic tissue.

[0071] The multimodal feature fusion unit encapsulates the final generated tissue classification tags, tumor invasion probability values, and original physical parameters (Modulus, Viscosity) into a navigation data packet. This navigation data packet is transmitted to the augmented reality display terminal via an internal high-speed bus, driving the visualization rendering engine to draw real-time tissue property heatmaps within the surgical microscope's field of view. This processing ensures that the navigation system guides surgical resection boundaries based not only on the shape of the imaging images but also on the histological quality.

[0072] See attached document Figure 3 The spatial registration logic unit of the present invention is used to establish a geometric mapping between the physical surgical space and the preoperative digital imaging space, and to track the probe tip pose in real time.

[0073] Probe calibration: The constant offset vector between the optical tracer ball on the probe handle and the probe tip is accurately calculated using the pivot calibration method. .

[0074] Spatial registration: A two-stage strategy, from coarse to fine, is employed. First, rigid point-to-point registration based on anatomical landmarks is performed. Then, the Iterative Closest Point (ICP) algorithm is used to finely fit the patient's surface point cloud acquired by the probe with the preoperative 3D model, yielding the final registration matrix. .

[0075] Real-time tracking: During navigation, the unit uses a reference frame matrix provided in real time by the optical positioning system. and probe matrix The virtual coordinates of the probe tip in the preoperative image coordinate system are calculated in real time using the following coordinate chain transformation formula. : ; Where all matrices are Homogeneous transformation matrix. The purpose of this device is to eliminate the slight movement of the patient's head relative to the optical positioning and tracking device during surgery, thereby achieving dynamic reference tracking. The function of this term is to determine the position of the probe tip in the world coordinate system. The function of this term is to ultimately map the position in the world coordinate system back to the CT / MRI image space.

[0076] In addition, the spatial registration logic unit also includes real-time target registration error (TRE) monitoring logic. After registration is completed, the spatial registration logic unit continuously calculates the residual distance of non-collinear feature points used for verification. When the residual distance exceeds the safety threshold of 2.0 mm, the spatial registration logic unit determines that the registration has failed (due to reference frame displacement or severe vibration) and sends an interrupt signal to the main control computer terminal, forcing the system to suspend the navigation function and prompting the user to re-execute the registration process to prevent incorrect navigation guidance from causing surgical risks.

[0077] The visualization rendering logic unit is embedded in the graphics processing pipeline of the main control computer terminal. It is configured to convert the abstract numerical results output by the multimodal feature fusion unit and the coordinate information output by the spatial registration logic unit into an intuitive three-dimensional navigation image, and drive the augmented reality display terminal to display it in real time.

[0078] The visualization and rendering logic unit first performs dynamic updates to the virtual scene graph. It maintains a real-time rendering scene in graphics memory, containing a preoperative 3D anatomical model of the patient and a virtual model of the integrated optical flow acoustic probe. The visualization and rendering logic unit then updates the virtual coordinates of the probe tip received from the spatial registration logic unit. The probe attitude matrix is ​​updated at a refresh rate of no less than 60 frames per second to update the position and orientation of the integrated optical flow acoustic probe virtual model in the virtual scene, ensuring that the movement of the virtual probe and the physical probe are visually strictly synchronized.

[0079] The core of the visualization rendering logic unit includes a tissue attribute texture mapping engine. This engine is configured to provide intuitive visualization of pathological features. When the integrated photoacoustic probe slides across or contacts the tissue surface, the engine receives the chordoma similarity index output by the multimodal feature fusion unit. And tissue classification labels. The tissue attribute texture mapping engine maps these pathological feature data to the corresponding mesh vertices on the surface of the preoperative 3D anatomical model, dynamically modifying the surface texture color of that area.

[0080] To transform the abstract chordoma similarity index into a visually recognizable warning signal for physicians, the tissue attribute texture mapping engine employs a non-linear color transfer function for color encoding. The tissue attribute texture mapping engine calculates the color vector of the rendered pixels. The mathematical model is as follows: ; in, This represents the output three-channel RGB color vector; The preset bone tissue identifier color (e.g., dark blue RGB[0, 0, 139]) is used to indicate safe high-hardness boundaries; Indicator colors for normal soft tissue or cerebrospinal fluid (e.g., green RGB[0, 255, 0]). Warning colors for chordoma tissue (e.g., bright red RGB[255, 0, 0]); This represents the normalized tumor probability value calculated by the multimodal feature fusion unit. This formula achieves a smooth color transition from normal tissue to tumor tissue, making the tumor invasion boundary appear as a gradual color band rather than an abrupt line, objectively reflecting the diffuse characteristics of pathological infiltration.

[0081] The visualization rendering logic unit also includes an augmented reality viewport compositing engine. This engine is configured to process real-time video streams from the surgical microscope camera and computer-generated virtual image streams. The engine first reads the optical magnification, focal length, and field of view parameters of the surgical microscope and sets the projection matrix of the virtual camera accordingly, ensuring that the perspective of the virtual scene perfectly matches the optical perspective of the microscope.

[0082] The augmented reality viewport synthesis engine then performs alpha blending and overlay operations on the images. To avoid the virtual image completely obscuring key anatomical structures (such as microvessels or nerves) in the field of view, the augmented reality viewport synthesis engine employs an importance-based adaptive transparency control algorithm. The final output to the augmented reality display terminal is the pixel intensity of the synthesized image. Determined by the following mixing equation: ; in, Represents pixel coordinates; This represents the pixel values ​​of the final composite image displayed. This represents the pixel values ​​of the virtual color layer generated by the organization attribute texture mapping engine; This represents the actual pixel values ​​of the video layer captured by the surgical microscope; This represents the composite opacity coefficient of the pixel, with a value ranging from 0 to 1.

[0083] The augmented reality viewport compositing engine dynamically calculates the importance of the information carried by each pixel. When the region corresponding to a certain pixel is determined to be a chordoma remnant (i.e. When the value exceeds a preset threshold, the augmented reality viewport compositing engine enhances the area. Values ​​(e.g., set to 0.6 to 0.8) make red warning blocks clearly stand out on the tissue surface; when the area corresponding to a pixel is determined to be safe bone or cavity, the augmented reality viewport compositing engine reduces... The value (e.g., set to 0.1 to 0.2) allows the surgeon to clearly observe the underlying anatomical texture.

[0084] In addition, the visualization rendering logic unit also provides a slice view auxiliary function. The visualization rendering logic unit captures CT / MRI fusion image slices perpendicular to the axis of the integrated optical flow acoustic probe in real time and displays them in a picture-in-picture window of the augmented reality image. This slice view is overlaid with a microscopic Young's modulus distribution curve generated by the optical coherence elastography subsystem, assisting the surgeon in determining the deep tissue structures in front of the probe tip, thereby achieving full-dimensional visualization guidance from macroscopic anatomical localization to microscopic pathological analysis.

[0085] The auxiliary alarm mechanism of this invention is used to provide the surgeon with graded, multi-channel feedback based on pathological analysis results and spatial location data.

[0086] Auditory alarm: A frequency-converting auditory synthesizer is used to generate audio similar to a Geiger counter. The sound frequency varies with the tumor similarity index. The risk level increases non-linearly and rapidly, providing a non-invasive risk level indication.

[0087] Tactile warning: When the distance between the probe tip and high-risk anatomical structures such as the internal carotid artery and optic nerve is calculated... When the distance is below the preset safety value, the linear resonant actuator (LRA) built into the handle will vibrate. The vibration intensity is inversely proportional to the distance; the closer the distance, the stronger the vibration.

[0088] Status Alarm: To ensure signal stability, alarm judgment uses Schmitt trigger logic with hysteresis. In addition, the system can self-check and report hardware faults such as pipe leaks and probe blockages.

[0089] This stage is performed by a medical image processing workstation, which first acquires thin-slice computed tomography (CT) scans (slice thickness less than 0.75 mm) and high-field magnetic resonance imaging (MRI) data of the patient's head. Subsequently, the spatial transformation matrix is ​​calculated using the maximized normalized mutual information (NMI) algorithm to achieve rigid registration between CT and MRI. The calculation formula is as follows: ; ; in, and These represent the random variables of the input CT and MRI images, respectively. and These represent the edge Shannon entropy of CT and MRI images, respectively; Represents the joint Shannon entropy of two images; It is a joint probability distribution; The grayscale value in a CT image And the gray value at the corresponding location in the MRI image is The joint probability distribution of ; and These are the sets of gray levels of the image.

[0090] After image fusion, the workstation extracts chordoma volume data (target area) and vascular structures (high-risk avoidance zone) using a level set segmentation algorithm, and constructs a bony boundary model and a three-dimensional safety envelope based on CT thresholds. The workstation calculates and generates a virtual cylindrical surgical channel according to the anatomical limitations of the transsphenoidal approach.

[0091] Finally, the reference parameters of the integrated optical flow acoustic detection probe were calibrated. The reference pressure was recorded under no-load conditions. Calculate the system deviation correction factor under load (contact with the standard phantom). Corrected effective Young's modulus for: ; in, It is the modulus value actually measured by the probe; It is the nominal modulus value of the calibration phantom; It is the original modulus value measured by the probe on the calibration phantom.

[0092] After calibration, a surgical engineering file package containing the anatomical model and calibration parameters is generated and transmitted to the main control computer terminal.

[0093] The main control computer terminal activates the optical positioning and tracking device to calculate the six-degree-of-freedom pose of the integrated optical flow acoustic detection probe in real time, and maps it to the image coordinate system using a registration matrix. After verifying the accuracy of the dissecting markers (error less than 1.5 mm), navigation guidance is initiated.

[0094] The main control computer terminal calculates the lateral deviation distance of the probe tip relative to the center axis of the planned path in real time. : ; in, Indicates time The vertical distance between the tip of the integrated optical flow acoustic detection probe and the center line of the planned path; Indicates time Real-time three-dimensional coordinate vector of the tip of the integrated optical flow acoustic detection probe; Represents the coordinate vector of the surgical access point planned before surgery; Represents the unit direction vector along the central axis of the surgical channel; This represents the vector product operation; Represents the magnitude (length) of a vector.

[0095] Simultaneously, calculate the depth distance along the path direction. and the remaining distance from the target area ; Where · represents the vector dot product operation; the main control computer terminal utilizes Calculate the remaining distance between the probe tip and the target area (i.e., the anterior surface of the chordoma). ,when When the distance is reduced to a preset proximity threshold (e.g., 5 mm), the main control computer terminal automatically switches the interface layout, magnifies the local anatomical structure of the target area, and highlights the outlines of the internal carotid artery and optic canal, prompting the operator to slow down the advancement speed. This prevents accidental puncture injury caused by excessive force.

[0096] When the integrated optical flow acoustic detection probe comes into contact with tissue, causing a change in the pressure sensor reading, the main control computer terminal freezes the macroscopic navigation, activates the pneumatic negative pressure and flow control components, and the system switches to the microscopic detection stage.

[0097] The main control computer terminal monitors flow resistance data. When effective contact is confirmed, a negative pressure lock-in mode is triggered to maintain a constant adsorption pressure (e.g., -10 kPa). A synchronous detection sequence is then initiated: the acoustic / mechanical excitation component outputs swept-frequency vibration (500 to 2000 Hz); the swept-frequency optical coherence tomography host performs M-mode acquisition; and the pneumatic negative pressure and flow control component performs low-frequency microrheological testing.

[0098] To prevent excessively fast scanning, the main control computer terminal limits the maximum transverse scanning speed based on the following formula. : ; in, This indicates the maximum permissible movement rate of the integrated optical flow acoustic probe on the tissue surface, expressed in millimeters per second. This indicates the effective spot diameter (lateral resolution) of the optical detection component on the focusing plane. This indicates the effective elastic imaging frame rate of the swept-frequency optical coherence tomography (OCT) host (i.e., the number of complete shear wave tracking cycles completed per second). This represents the spatial overlap coefficient, which must be greater than 2.0 to satisfy the Nyquist sampling theorem, ensuring sufficient overlap between adjacent probe points to prevent small tumor infiltrates from being missed.

[0099] The system uses an organizational heterogeneity boundary enhancement algorithm to calculate the boundary index. To identify tumor boundaries: ; in, Indicates position The effective Young's modulus at the location; Indicates position Fluid behavior index at the location; These are the weighting coefficients for rheological characteristics. When the boundary index... When a local maximum occurs, the main control computer terminal determines that the integrated photoacoustic probe is crossing the interface between different tissue types. At this point, the system automatically increases the redundancy of data acquisition, performing multiple repeated measurements at the same location and averaging the results to suppress noise interference and accurately depict the invasion front of the tumor.

[0100] The main control computer terminal constructs a dynamic virtual resection guidance channel based on the probe data, and colors the tumor infiltration area (red) and the safe area (blue) in the augmented reality display terminal. The dynamic surgical boundary calculation engine calculates the recommended resection margin distance in real time. : ; in, Indicates the additional depth or extent of resection recommended at the current operation point; This indicates the baseline resection safety margin (e.g., 1 mm) based on the anatomical location. This indicates the similarity index of detected chordomas; The effective Young's modulus (normalized value) of the tissue is used to characterize the looseness of tumor tissue (the looser the tissue, the stronger its invasiveness). This represents the tumor invasiveness weighting coefficient; This indicates the closest distance from the current point to a high-risk anatomical structure (such as the internal carotid artery).

[0101] It is a distance-based security penalty function, defined as follows: ; The penalty function works as follows: when the probe point is far from the dangerous structure ( When it is relatively large, Approaching 1, the system allows for larger resection margins to completely remove the tumor; when the probe point approaches a dangerous structure ( When it approaches 0, The system rapidly decays to zero, forcibly contracting the resection area to protect the nerves and blood vessels, achieving a mathematical balance between thoroughness and safety.

[0102] Auxiliary alarm mechanism in case of operational deviation When the area is within range, an audible and tactile alarm is triggered. After resection, the system performs residual cavity verification and iterative cleaning procedures, generates a digital biopsy residual map, and guides remedial resection until all sampling point indicators return to the safe threshold.

Claims

1. A real-time navigation system for chordoma invasion boundaries based on multimodal image fusion, characterized in that, include: Integrated optical flow acoustic detection probe; An optical positioning and tracking device is used to acquire the position of the integrated optical flow acoustic detection probe; A pneumatic negative pressure control device is connected to the integrated optical flow acoustic detection probe; A swept-frequency optical coherence tomography host is connected to the integrated optical flow acoustic detection probe; Augmented reality display terminals and main control computer terminals establish data communication connections with various devices and the host computer; The main control computer terminal is configured as follows: Based on the data obtained from the pneumatic negative pressure control device, the rheological characteristics of the tissue were analyzed; Based on the data obtained from the swept-frequency optical coherence tomography host, the hardness characteristics of the tissue are analyzed; The rheological features and the hardness features are fused to generate tissue identification results that characterize the tumor invasion boundary; Based on the tissue identification results and the probe position provided by the optical positioning and tracking device, an augmented reality navigation image is generated, driving the augmented reality display terminal to display it in real time.

2. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The integrated optical flow acoustic detection probe includes: The mid-section rigid extension sleeve assembly adopts a double-layer coaxial metal tube structure, including an outer stainless steel sleeve and an inner protective sleeve, with an annular negative pressure channel formed between the two sleeves. The remote optical flow coupling probe adopts a non-flush end face design. The end face of the outer stainless steel sleeve extends axially and exceeds the end face of the inner protective sleeve, thereby forming a gas-liquid coupling cavity between the optical emission surface of the inner protective sleeve and the contact tissue.

3. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The pneumatic negative pressure control device includes a high-precision differential pressure sensor and a thermal mass flow meter; The main control computer terminal is configured to calculate the dynamic rheological impedance index based on the instantaneous air pressure difference data provided by the high-precision differential pressure sensor and the fluid volume flow rate data provided by the thermal mass flow meter through a rheological parameter calculation module, and determine the tissue type based on the fluid behavior index.

4. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The integrated optical flow acoustic detection probe includes a piezoelectric actuator for exciting micro-shear waves; The swept-frequency optical coherence tomography host is configured to detect micron-level displacement of tissue by analyzing phase term changes in the interference signal; The main control computer terminal is configured to calculate the Young's modulus of the tissue based on the shear wave propagation velocity detected by optical detection through an elastic imaging inversion module.

5. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The main control computer terminal also includes: The image rigid registration algorithm unit is used to map the coordinate space of magnetic resonance imaging data to the coordinate space of computed tomography data using an automatic registration algorithm based on mutual information, and to construct a preoperative virtual three-dimensional model. The spatial registration logic unit is used to solve the rotation matrix and translation vector using the least squares point cloud matching algorithm.

6. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 5, characterized in that, The main control computer terminal is configured to mark the tumor outline on the preoperative virtual three-dimensional model and expand outward to generate a microscopic detection activation area; When the optical positioning and tracking device tracks the tip of the integrated optical flow acoustic detection probe into the microscopic detection activation area, the main control computer terminal automatically triggers a signal to switch the system from macroscopic geometric navigation to microscopic physical detection mode.

7. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The main control computer terminal performs the fusion through a multimodal feature fusion unit, which is configured as follows: Construct a multidimensional feature vector containing effective Young's modulus, fluid behavior index, and consistency coefficient; The classification decision engine calculates the statistical distance between the current feature vector and the pre-set chordoma baseline feature library to obtain the normalized chordoma similarity index. Based on the normalized chordoma similarity index, the hierarchical judgment logic is executed to output the tissue classification label.

8. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 1, characterized in that, The main control computer terminal generates the augmented reality navigation image through a visualization rendering logic unit, which includes: The tissue attribute texture mapping engine is configured to use a non-linear color transfer function to map the chordoma similarity index onto the corresponding mesh vertices on the surface of the preoperative 3D anatomical model, thereby smoothly transitioning the rendered pixel color between the identifier color representing normal soft tissue and the warning color representing chordoma tissue.

9. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 8, characterized in that, The visualization rendering logic unit also includes: The augmented reality viewport compositing engine is configured to use an adaptive transparency control algorithm to dynamically calculate the compositing opacity coefficient based on the chordoma similarity index corresponding to each pixel, perform an alpha blending and overlay operation on the image, and composite the virtual color layer generated by the tissue attribute texture mapping engine with the real video layer acquired by the surgical microscope.

10. The real-time navigation system for chordoma invasion boundaries based on multimodal image fusion according to claim 7, characterized in that, The main control computer terminal also includes: The dynamic surgical boundary calculation engine is configured to calculate and recommend a resection margin distance in real time based on the chordoma similarity index, the effective Young's modulus of the tissue, and the nearest distance between the current probe point and the high-risk anatomical structure. The calculation of the excision margin includes a distance-based safety penalty function, which attenuates and shrinks the excision range when the probe point approaches a high-risk anatomical structure.

Images