Laser bone knife intelligent maxillofacial plastic surgery robot system and method based on ai and multi-modal navigation

Abstract

The application discloses a kind of laser bone cutter intelligent maxillofacial plastic surgery robot systems and methods based on AI and multimodal navigation, system includes: composite laser generation unit, micro magnetic control flexible laser working head, mechanical arm execution unit, multimodal perception unit, AI surgery planning unit, real-time navigation control unit and man-machine interaction unit, the composite laser generation unit and micro magnetic control flexible laser working head interact, the mechanical arm execution unit and micro magnetic control flexible laser working head interact, the mechanical arm execution unit, multimodal perception unit, AI surgery planning unit, man-machine interaction unit all with real-time navigation control unit interact.Compared with prior art, the present application can cut the deep jaw and face bone in oral cavity, through the multimodal perception unit, can provide multimodal data reference for surgery planning, with AI surgery planning unit, realize intelligent planning of surgery, reduce the dependence on doctor's experience, improve standardization.

Description

Technical Field

[0001] This invention belongs to the fields of medical robots, laser bone cutting, oral and maxillofacial surgery, artificial intelligence image navigation, and minimally invasive surgical instruments, and specifically relates to an intelligent maxillofacial plastic surgery robot system and method based on AI and multimodal navigation using a laser bone scalpel. Background Technology

[0002] Maxillofacial plastic surgery and orthognathic surgery have extremely high requirements for osteotomy precision, minimal invasiveness, and safety. Currently, the traditional method is to cut the bone using a mechanical osteotome and oscillating saw, which usually has an osteotomy error of 1-2mm and has the problems of large trauma, more bleeding, and high risk of postoperative complications.

[0003] Compared to traditional mechanical grinding, laser bone scalpel technology uses laser ablation to better preserve bone tissue structure, reduce pain and mechanical damage, and achieve a cutting precision of 0.2 to 0.3 millimeters.

[0004] However, while existing laser bone cutters have improved precision, they generally suffer from large working head diameters (>8mm), making it impossible to access narrow and curved anatomical areas deep within the oral cavity; they mostly use a single laser source, making it difficult to balance deep cutting capabilities with fine trimming accuracy; their navigation systems are mostly CT / optical dual-modality, lacking closed-loop safety control for real-time tissue recognition, temperature monitoring, and depth monitoring; and the surgery is highly dependent on the surgeon's experience, with low levels of intelligence and standardization. Summary of the Invention

[0005] To address the aforementioned issues, the present invention aims to provide an AI-based and multimodal navigation-based intelligent maxillofacial surgery robot system and method using a laser bone scalpel. This system can cut deep maxillofacial bones in the oral cavity. Through a multimodal sensing unit, it can provide multimodal data references for surgical planning. Combined with an AI surgical planning unit, it enables intelligent surgical planning, reduces reliance on physician experience, and improves standardization.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] This invention provides an AI-based and multimodal navigation-based intelligent maxillofacial plastic surgery robot system using a laser bone scalpel, comprising:

[0008] Composite laser generating unit for outputting flat-top femtosecond composite laser;

[0009] A miniature magnetically controlled flexible laser working head that can rotate flexibly within the oral cavity under the drive of an external magnetic field to achieve osteotomy and cutting in maxillofacial plastic surgery.

[0010] A robotic arm execution unit used to drive a miniature magnetically controlled flexible laser working head to accurately reach the preset surgical site and move along the planned path;

[0011] A multimodal sensing unit used to collect anatomical and physiological information of patients in all dimensions, providing multimodal data reference for surgical planning;

[0012] Based on deep learning, an AI surgical planning unit automatically generates osteotomy paths, safety boundaries, and movement vectors to achieve intelligent surgical planning.

[0013] A real-time navigation control unit used to fuse multimodal data to achieve human-machine collaborative control;

[0014] And a human-computer interaction unit for doctors to monitor the surgical process in real time, adjust parameters, and handle emergencies;

[0015] The composite laser generating unit interacts with the miniature magnetically controlled flexible laser working head, the robotic arm execution unit interacts with the miniature magnetically controlled flexible laser working head, and the robotic arm execution unit, multimodal perception unit, AI surgical planning unit, and human-computer interaction unit all interact with the real-time navigation control unit.

[0016] Furthermore, the composite laser generating unit includes:

[0017] Used for outputting continuous laser light Laser module;

[0018] Beam shaping module for shaping continuous laser beams into flat-top laser beams;

[0019] Femtosecond laser module for outputting femtosecond lasers;

[0020] A beam combining module used to combine flat-top lasers and femtosecond lasers into a flat-top femtosecond composite laser;

[0021] The The laser module and the beam shaping module are connected. Both the beam shaping module and the femtosecond laser module are connected to the beam combining module, and the beam combining module is connected to the micro-magnetically controlled flexible laser working head.

[0022] In this application, The laser module outputs continuous or quasi-continuous laser light, which is shaped by the beam shaping module (spatial light modulator) to form a flat-top beam with uniform energy distribution (energy uniformity ≥90%) for efficient cutting of deep bone tissue. The femtosecond laser module outputs ultrashort pulse laser light, which is coupled to the flat-top beam through the beam combining module for micron-level fine trimming of the osteotomy edge, thereby realizing the integration of deep cutting and high-precision processing under a single optical path.

[0023] Furthermore, the miniature magnetron-controlled flexible laser working head includes:

[0024] Multi-core mid-infrared optical fiber used as a laser transmission channel;

[0025] An aspherical focusing lens is used to focus composite lasers emitted from multi-core mid-infrared optical fibers with high precision, forming a small spot of concentrated energy and clear edges on the bone surface.

[0026] NdFeB magnetic ring used to rotate the front end of a multi-core mid-infrared optical fiber in conjunction with an external magnetic field, thereby changing the laser output direction.

[0027] An OCT sensor used to emit near-infrared detection light and receive echo signals to acquire tomographic images of bone tissue in real time and measure cutting depth and interface changes.

[0028] An infrared temperature sensor used to detect the infrared radiation intensity of the cut area surface in real time and convert it into a surface temperature value.

[0029] And a Raman spectroscopy probe for emitting probe light into oral tissues and collecting scattering spectra, enabling real-time identification of tissue components based on differences in characteristic peaks, thus achieving soft tissue protection;

[0030] The NdFeB magnetic ring, OCT sensor, infrared temperature sensor, and Raman spectroscopy probe are all mounted on a multi-core mid-infrared optical fiber, and the aspherical focusing lens is mounted at the very front end of the multi-core mid-infrared optical fiber.

[0031] In this application, a multi-core mid-infrared optical fiber serves as the laser transmission channel, transmitting the laser output from the composite laser generating unit. Laser and femtosecond lasers are efficiently transmitted to the front end of the working head, and multi-core beam combining is achieved at the output end to form the desired spot shape. Mid-infrared fiber optic adapter. The laser wavelength and multi-core structure facilitate beam homogenization and energy distribution control, which is crucial for achieving high-power laser transmission within a small-diameter working head. The aspherical focusing lens, with its aspherical design, reduces spherical aberration and astigmatism, ensuring high energy density and a small focal spot even with a working head inner diameter of ≤3mm, thus achieving micron-level precision cutting. The NdFeB magnetic ring, acting as a magnetodynamic conversion element, generates torque under the influence of an external controllable magnetic field, driving the front end of the multi-core mid-infrared fiber to rotate or deflect, thereby changing the laser output direction. This allows the laser working head to conform to curved bone surfaces such as the mandibular angle and zygomatic arch, solving the problem of traditional linear working heads being "reachable but inaccurate." The OCT sensor emits near-infrared detection light and receives echo signals. Real-time acquisition of tomographic images of bone tissue and measurement of cutting depth and interface changes are crucial closed-loop sensing methods for controlling osteotomy depth and preventing overcutting and nerve damage. Infrared thermometers can detect the infrared radiation intensity of the cutting area surface in real time and convert it into a surface temperature value for thermal safety closed-loop control, ensuring that the bone surface temperature is ≤47℃ and the heat-affected zone is controlled within ≤100μm to avoid bone necrosis and thermal damage to surrounding tissues. Raman spectroscopy probes emit probe light into the tissue and collect the scattered spectrum, distinguishing different tissue components such as bone, nerves, and blood vessels based on differences in characteristic peaks, enabling real-time identification of tissue components. When the tissue type changes (such as near nerves), it triggers an early warning or automatic shutdown, making it a key sensor for neurovascular protection.

[0032] Furthermore, the robotic arm execution unit includes: a robotic arm with degrees of freedom, wherein a force sensor is provided at the end of the robotic arm to drive a miniature magnetically controlled flexible laser working head to accurately reach a preset surgical site and move along a planned path.

[0033] Furthermore, the multimodal sensing unit includes:

[0034] Medical imaging equipment used to create three-dimensional bone models before surgery;

[0035] Ultrasound equipment used to monitor soft tissue deformation during surgery;

[0036] Infrared optical navigation equipment used for spatial positioning;

[0037] Infrared thermal imager used for regional temperature field monitoring;

[0038] Raman spectrometer for tissue composition analysis;

[0039] And hemodynamic and neurophysiological monitoring equipment for real-time assessment of neurovascular status;

[0040] The medical imaging equipment, ultrasound equipment, infrared optical navigation equipment, infrared thermal imager, Raman spectrometer, and Raman spectrometer all interact with the real-time navigation control unit.

[0041] Furthermore, the AI ​​surgical planning unit incorporates a maxillofacial surgery knowledge base and reinforcement learning training framework built on a large language model. This unit receives preoperative CT / MRI data, automatically identifies key anatomical structures such as the mandibular canal and infraorbital nerve, and generates personalized osteotomy paths, safety boundaries (safe distance ≥2mm from the infraal alveolar nerve), and robotic arm movement vectors, replacing traditional manual experience-based planning.

[0042] Furthermore, the real-time navigation control unit deploys a geometric, visual, and spectral three-modal fusion navigation (GVSN) algorithm. This algorithm deeply fuses features from geometric modalities (CT 3D model and optical navigation pose), visual modalities (intraoperative ultrasound soft tissue deformation), and spectral modalities (Raman spectroscopy tissue identification) to achieve dynamic deformation compensation and real-time tissue classification. Simultaneously, it integrates a human-machine collaborative controller, supporting switching between manual fine-tuning by the physician and autonomous execution by the robot, and constructs a safety closed-loop protection logic.

[0043] Furthermore, the human-computer interaction unit is equipped with a touch screen, a foot switch, a three-dimensional visual navigation interface, and an emergency stop and manual takeover module, which facilitates doctors to monitor the surgical process, adjust parameters, and handle emergencies in real time.

[0044] The working process of this system is as follows:

[0045] 1. The multimodal perception unit collects CT / MRI data of the patient's maxillofacial region, constructs a three-dimensional skeletal and neurovascular model, and sends the anatomical information to the AI ​​surgical planning unit;

[0046] 2. The AI ​​surgical planning unit, based on Large Language Model (LLM) and reinforcement learning, automatically identifies key structures such as the inferior alveolar nerve and infraorbital nerve, generates osteotomy path, safety boundary (≥2mm from the nerve), and laser parameter preset scheme, and sends them to the real-time navigation and control unit.

[0047] 3. The multimodal sensing unit completes the spatial registration of preoperative images and intraoperative optical / ultrasound navigation, aligning the patient's actual pose with the three-dimensional model, with a registration error ≤0.5mm;

[0048] 4. The real-time navigation and control unit initiates GVSN trimodal fusion navigation (geometric + visual + spectral), establishes a dynamic reference coordinate system, and waits for instructions to be executed;

[0049] 5. The real-time navigation and control unit sends trajectory and pose commands to the robotic arm execution unit and laser power, pulse width and mode switching commands to the composite laser generation unit according to the planned path.

[0050] 6. The robotic arm execution unit drives the miniature magnetically controlled flexible laser working head to reach the target bone region through the intraoral approach. The external magnetic field controls the working head to achieve 360° rotation and fit the curved surface.

[0051] 7. The composite laser generator unit outputs a flat-top femtosecond composite laser;

[0052] 8. While outputting laser light, the miniature magnetically controlled flexible laser head uses built-in OCT, infrared thermometry, and Raman spectroscopy sensors to collect cutting depth, surface temperature, and tissue type in real time, and transmits them back to the multimodal sensing unit.

[0053] 9. The multimodal perception unit sends real-time perception data to the real-time navigation and control unit.

[0054] The present invention also provides an intelligent maxillofacial plastic surgery method based on the above system, comprising the following steps:

[0055] S1: Preoperative AI 3D planning, collecting CT and MRI data of the patient's maxillofacial region, constructing a high-precision 3D skeletal and neurovascular model; the AI ​​surgical planning unit automatically generates osteotomy lines based on reinforcement learning strategies, sets safety margins, and outputs initial laser parameters and robotic arm motion trajectories;

[0056] S2: Multimodal registration and navigation initialization: Spatial registration of preoperative image data with intraoperative optical navigation coordinate system, with registration error controlled within ≤0.5mm; Start GVSN trimodal fusion navigation to complete the initial alignment of geometric, visual, and spectral features;

[0057] S3: Minimally invasive intraoral approach and positioning, controlling the robotic arm to drive the miniature magnetically controlled flexible laser working head through the oral cavity, using an external magnetic field to control the front end of the working head to rotate so that it conforms to the curved shape of the target bone;

[0058] S4: Composite laser collaborative cutting and adaptive control. First, the flat-top laser is started to cut deep bone. After reaching the preset depth, the femtosecond laser is switched to refine the edges. The AI ​​control unit dynamically adjusts the laser power, pulse width and scanning speed according to the cutting depth fed back by OCT in real time and the tissue type identified by Raman spectroscopy.

[0059] S5: Multi-dimensional safety closed-loop monitoring. During the operation, the infrared temperature measurement module scans the temperature of the cutting area at a frequency of 50ms / time to ensure that the surface temperature is ≤47℃; OCT monitors the depth of the heat-affected zone to ensure that it is ≤100μm; Raman spectroscopy distinguishes bone and nerve and blood vessel tissue in real time. When the working head is detected to be approaching the danger zone, the system automatically triggers an audible and visual warning and controls the robotic arm to brake.

[0060] S6: Postoperative accuracy verification and evaluation. After the surgery, the system automatically compares the deviation between the actual osteotomy line and the planned path, and generates a report that includes osteotomy accuracy, facial symmetry and safety assessment to assist doctors in making postoperative judgments.

[0061] Compared with the prior art, the present invention has the following significant advantages:

[0062] First, breakthroughs in miniaturization and flexibility: With a magnetically controlled flexible working head with an outer diameter of ≤3mm, the problem of traditional instruments being unable to enter the narrow areas deep in the oral cavity has been solved, achieving 360° operation without blind spots.

[0063] Second, the accuracy is greatly improved: the combination of flat-top femtosecond laser and high-precision navigation can control the average deviation of osteotomy, improve the accuracy of navigation and positioning, and improve the facial symmetry after surgery.

[0064] Third, safety is significantly enhanced: based on trimodal fusion navigation and multidimensional sensing monitoring, the risk of neurovascular injury and thermal injury can be reduced.

[0065] Fourth, intelligent and standardized: AI-powered automatic planning can significantly shorten surgery time, reduce reliance on doctors' experience, and improve the predictability of surgical outcomes.

[0066] Fifth, the minimally invasive approach has significant advantages: the intraoral approach leaves no external scars, reduces intraoperative bleeding, shortens the patient's hospital stay, and results in higher quality bone healing. Attached Figure Description

[0067] Fig. 1 This is a system framework diagram of this embodiment.

[0068] Fig. 2 This is a flowchart of the intelligent maxillofacial plastic surgery method.

[0069] Fig. 3 This is a schematic diagram of the structure of a miniature magneto-controlled flexible laser working head.

[0070] In the diagram: 1. Multi-core mid-infrared optical fiber; 2. Aspherical focusing lens; 3. NdFeB magnetic ring; 4. OCT sensor; 5. Infrared temperature sensor; 6. Raman spectroscopy probe; Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0072] To achieve the above objectives, the technical solution of the present invention is as follows:

[0073] SeeFigs. 1-3 As shown:

[0074] This provides an AI-based and multimodal navigation-enabled intelligent maxillofacial plastic surgery robot system using a laser bone scalpel, comprising:

[0075] Composite laser generating unit for outputting flat-top femtosecond composite laser;

[0076] A miniature magnetically controlled flexible laser working head that can rotate flexibly within the oral cavity under the drive of an external magnetic field to achieve osteotomy and cutting in maxillofacial plastic surgery.

[0077] A robotic arm execution unit used to drive a miniature magnetically controlled flexible laser working head to accurately reach the preset surgical site and move along the planned path;

[0078] A multimodal sensing unit used to collect anatomical and physiological information of patients in all dimensions, providing multimodal data reference for surgical planning;

[0079] Based on deep learning, an AI surgical planning unit automatically generates osteotomy paths, safety boundaries, and movement vectors to achieve intelligent surgical planning.

[0080] A real-time navigation control unit used to fuse multimodal data to achieve human-machine collaborative control;

[0081] And a human-computer interaction unit for doctors to monitor the surgical process in real time, adjust parameters, and handle emergencies;

[0082] The composite laser generating unit interacts with the miniature magnetically controlled flexible laser working head, the robotic arm execution unit interacts with the miniature magnetically controlled flexible laser working head, and the robotic arm execution unit, multimodal perception unit, AI surgical planning unit, and human-computer interaction unit all interact with the real-time navigation control unit.

[0083] Furthermore, the composite laser generating unit includes:

[0084] Used for outputting continuous laser light Laser module;

[0085] Beam shaping module for shaping continuous laser beams into flat-top laser beams;

[0086] Femtosecond laser module for outputting femtosecond lasers;

[0087] A beam combining module used to combine flat-top lasers and femtosecond lasers into a flat-top femtosecond composite laser;

[0088] The The laser module and the beam shaping module are connected. Both the beam shaping module and the femtosecond laser module are connected to the beam combining module, and the beam combining module is connected to the micro-magnetically controlled flexible laser working head.

[0089] In this application, The laser module outputs continuous or quasi-continuous laser light, which is shaped by the beam shaping module (spatial light modulator) to form a flat-top beam with uniform energy distribution (energy uniformity ≥90%) for efficient cutting of deep bone tissue. The femtosecond laser module outputs ultrashort pulse laser light, which is coupled to the flat-top beam through the beam combining module for micron-level fine trimming of the osteotomy edge, thereby realizing the integration of deep cutting and high-precision processing under a single optical path.

[0090] Furthermore, the miniature magnetron-controlled flexible laser working head includes:

[0091] Multi-core mid-infrared optical fiber used as a laser transmission channel;

[0092] An aspherical focusing lens is used to focus composite lasers emitted from multi-core mid-infrared optical fibers with high precision, forming a small spot of concentrated energy and clear edges on the bone surface.

[0093] NdFeB magnetic ring used to rotate the front end of a multi-core mid-infrared optical fiber in conjunction with an external magnetic field, thereby changing the laser output direction.

[0094] An OCT sensor used to emit near-infrared detection light and receive echo signals to acquire tomographic images of bone tissue in real time and measure cutting depth and interface changes.

[0095] An infrared temperature sensor used to detect the infrared radiation intensity of the cut area surface in real time and convert it into a surface temperature value.

[0096] And a Raman spectroscopy probe for emitting probe light into oral tissues and collecting scattering spectra, enabling real-time identification of tissue components based on differences in characteristic peaks, thus achieving soft tissue protection;

[0097] The NdFeB magnetic ring, OCT sensor, infrared temperature sensor, and Raman spectroscopy probe are all mounted on a multi-core mid-infrared optical fiber, and the aspherical focusing lens is mounted at the very front end of the multi-core mid-infrared optical fiber.

[0098] In this application, a multi-core mid-infrared optical fiber serves as the laser transmission channel, transmitting the laser output from the composite laser generating unit. Laser and femtosecond lasers are efficiently transmitted to the front end of the working head, and multi-core beam combining is achieved at the output end to form the desired spot shape. Mid-infrared fiber optic adapter. The laser wavelength and multi-core structure facilitate beam homogenization and energy distribution control, which is crucial for achieving high-power laser transmission within a small-diameter working head. The aspherical focusing lens, with its aspherical design, reduces spherical aberration and astigmatism, ensuring high energy density and a small focal spot even with a working head inner diameter of ≤3mm, thus achieving micron-level precision cutting. The NdFeB magnetic ring, acting as a magnetodynamic conversion element, generates torque under the influence of an external controllable magnetic field, driving the front end of the multi-core mid-infrared fiber to rotate or deflect, thereby changing the laser output direction. This allows the laser working head to conform to curved bone surfaces such as the mandibular angle and zygomatic arch, solving the problem of traditional linear working heads being "reachable but inaccurate." The OCT sensor emits near-infrared detection light and receives echo signals. Real-time acquisition of tomographic images of bone tissue and measurement of cutting depth and interface changes are crucial closed-loop sensing methods for controlling osteotomy depth and preventing overcutting and nerve damage. Infrared thermometers can detect the infrared radiation intensity of the cutting area surface in real time and convert it into a surface temperature value for thermal safety closed-loop control, ensuring that the bone surface temperature is ≤47℃ and the heat-affected zone is controlled within ≤100μm to avoid bone necrosis and thermal damage to surrounding tissues. Raman spectroscopy probes emit probe light into the tissue and collect the scattered spectrum, distinguishing different tissue components such as bone, nerves, and blood vessels based on differences in characteristic peaks, enabling real-time identification of tissue components. When the tissue type changes (such as near nerves), it triggers an early warning or automatic shutdown, making it a key sensor for neurovascular protection.

[0099] Furthermore, the robotic arm execution unit includes: a robotic arm with degrees of freedom, wherein a force sensor is provided at the end of the robotic arm to drive a miniature magnetically controlled flexible laser working head to accurately reach a preset surgical site and move along a planned path.

[0100] Furthermore, the multimodal sensing unit includes:

[0101] Medical imaging equipment used to create three-dimensional bone models before surgery;

[0102] Ultrasound equipment used to monitor soft tissue deformation during surgery;

[0103] Infrared optical navigation equipment used for spatial positioning;

[0104] Infrared thermal imager used for regional temperature field monitoring;

[0105] Raman spectrometer for tissue composition analysis;

[0106] And hemodynamic and neurophysiological monitoring equipment for real-time assessment of neurovascular status;

[0107] The medical imaging equipment, ultrasound equipment, infrared optical navigation equipment, infrared thermal imager, Raman spectrometer, and Raman spectrometer all interact with the real-time navigation control unit.

[0108] Furthermore, the AI ​​surgical planning unit incorporates a maxillofacial surgery knowledge base and reinforcement learning training framework built on a large language model. This unit receives preoperative CT / MRI data, automatically identifies key anatomical structures such as the mandibular canal and infraorbital nerve, and generates personalized osteotomy paths, safety boundaries (safe distance ≥2mm from the infraal alveolar nerve), and robotic arm movement vectors, replacing traditional manual experience-based planning.

[0109] Furthermore, the real-time navigation control unit deploys a geometric, visual, and spectral three-modal fusion navigation (GVSN) algorithm. This algorithm deeply fuses features from geometric modalities (CT 3D model and optical navigation pose), visual modalities (intraoperative ultrasound soft tissue deformation), and spectral modalities (Raman spectroscopy tissue identification) to achieve dynamic deformation compensation and real-time tissue classification. Simultaneously, it integrates a human-machine collaborative controller, supporting switching between manual fine-tuning by the physician and autonomous execution by the robot, and constructs a safety closed-loop protection logic.

[0110] Furthermore, the human-computer interaction unit is equipped with a touch screen, a foot switch, a three-dimensional visual navigation interface, and an emergency stop and manual takeover module, which facilitates doctors to monitor the surgical process, adjust parameters, and handle emergencies in real time.

[0111] The working process of this system is as follows:

[0112] 1. The multimodal perception unit collects CT / MRI data of the patient's maxillofacial region, constructs a three-dimensional skeletal and neurovascular model, and sends the anatomical information to the AI ​​surgical planning unit;

[0113] 2. The AI ​​surgical planning unit, based on Large Language Model (LLM) and reinforcement learning, automatically identifies key structures such as the inferior alveolar nerve and infraorbital nerve, generates osteotomy path, safety boundary (≥2mm from the nerve), and laser parameter preset scheme, and sends them to the real-time navigation and control unit.

[0114] 3. The multimodal sensing unit completes the spatial registration of preoperative images and intraoperative optical / ultrasound navigation, aligning the patient's actual pose with the three-dimensional model, with a registration error ≤0.5mm;

[0115] 4. The real-time navigation and control unit initiates GVSN trimodal fusion navigation (geometric + visual + spectral), establishes a dynamic reference coordinate system, and waits for instructions to be executed;

[0116] 5. The real-time navigation and control unit sends trajectory and pose commands to the robotic arm execution unit and laser power, pulse width and mode switching commands to the composite laser generation unit according to the planned path.

[0117] 6. The robotic arm execution unit drives the miniature magnetically controlled flexible laser working head to reach the target bone region through the intraoral approach. The external magnetic field controls the working head to achieve 360° rotation and fit the curved surface.

[0118] 7. The composite laser generator unit outputs a flat-top femtosecond composite laser;

[0119] 8. While outputting laser light, the miniature magnetically controlled flexible laser head uses built-in OCT, infrared thermometry, and Raman spectroscopy sensors to collect cutting depth, surface temperature, and tissue type in real time, and transmits them back to the multimodal sensing unit.

[0120] 9. The multimodal perception unit sends real-time perception data to the real-time navigation and control unit.

[0121] This embodiment also provides an intelligent maxillofacial plastic surgery method based on the above system, including the following steps:

[0122] S1: Preoperative AI 3D planning, collecting CT and MRI data of the patient's maxillofacial region, constructing a high-precision 3D skeletal and neurovascular model; the AI ​​surgical planning unit automatically generates osteotomy lines based on reinforcement learning strategies, sets safety margins, and outputs initial laser parameters and robotic arm motion trajectories;

[0123] S2: Multimodal registration and navigation initialization: Spatial registration of preoperative image data with intraoperative optical navigation coordinate system, with registration error controlled within ≤0.5mm; Start GVSN trimodal fusion navigation to complete the initial alignment of geometric, visual, and spectral features;

[0124] S3: Minimally invasive intraoral approach and positioning, controlling the robotic arm to drive the miniature magnetically controlled flexible laser working head through the oral cavity, using an external magnetic field to control the front end of the working head to rotate so that it conforms to the curved shape of the target bone;

[0125] S4: Composite laser collaborative cutting and adaptive control. First, the flat-top laser is started to cut deep bone. After reaching the preset depth, the femtosecond laser is switched to refine the edges. The AI ​​control unit dynamically adjusts the laser power, pulse width and scanning speed according to the cutting depth fed back by OCT in real time and the tissue type identified by Raman spectroscopy.

[0126] S5: Multi-dimensional safety closed-loop monitoring. During the operation, the infrared temperature measurement module scans the temperature of the cutting area at a frequency of 50ms / time to ensure that the surface temperature is ≤47℃; OCT monitors the depth of the heat-affected zone to ensure that it is ≤100μm; Raman spectroscopy distinguishes bone and nerve and blood vessel tissue in real time. When the working head is detected to be approaching the danger zone, the system automatically triggers an audible and visual warning and controls the robotic arm to brake.

[0127] S6: Postoperative accuracy verification and evaluation. After the surgery, the system automatically compares the deviation between the actual osteotomy line and the planned path, and generates a report that includes osteotomy accuracy, facial symmetry and safety assessment to assist doctors in making postoperative judgments.

[0128] The aforementioned intelligent maxillofacial plastic surgery methods can be applied to mandibular angle reduction surgery, zygomatic bone and arch reduction surgery, genioplasty, and orthognathic surgery for skeletal deformities. Their specific implementation methods are as follows:

[0129] 1. Jaw angle reduction surgery:

[0130] (1) Collect the patient’s maxillofacial CT data. The AI ​​generates the osteotomy path within 3 minutes, with a safety margin ≥2mm from the inferior alveolar nerve.

[0131] (2) The system completes the registration of CT and optical navigation with a registration error of ≤0.5mm.

[0132] (3) The robotic arm drives the 2.8mm magnetic control head through the intraoral approach to reach the inner side of the mandibular angle.

[0133] (4) A flat-top laser is used for deep cutting, a femtosecond laser is used to trim the edges, and OCT is used to monitor the depth in real time.

[0134] (5) Infrared temperature measurement 50ms / scan, temperature ≤47℃, heat-affected zone ≤100μm.

[0135] (6) Raman spectroscopy identifies bone / nerves with an accuracy rate of ≥98% and no neurovascular damage throughout the process.

[0136] (7) The operation time is about 25 minutes, the osteotomy accuracy is ≤0.2mm, and the postoperative symmetry is ≥95%.

[0137] 2. Zygomatic bone and zygomatic arch reduction surgery:

[0138] (1) CT reconstruction of the zygomatic arch and zygomatic bone shape, AI planning of the arc osteotomy line.

[0139] (2) The flexible working head is magnetically controlled to steer and fits the zygomatic arch surface for operation.

[0140] (3) Composite laser precision cutting, real-time deformation compensation error ≤0.8mm.

[0141] (4) The osteotomy surface is smooth after surgery, and the asymmetry rate is ≤3%.

[0142] 3. Genioplasty:

[0143] (1) Collect three-dimensional CT images of the patient's mandible and chin. AI automatically designs the range of chin osteotomy, forward / backward / upward adjustment based on surgical knowledge base, and reserves a safe distance of ≥2.2mm for the mental nerve.

[0144] (2) After multimodal registration is completed, the 2.8mm magnetically controlled flexible working head is used for minimally invasive access through the intraoral vestibule and is adapted to the narrow surgical area of ​​the chin bone by bending at multiple angles using the built-in micro magnetic ring.

[0145] (3) The main body of the chin bone is cut open by a flat-top laser, the osteotomy surface is finely polished by a femtosecond laser, the mental nerve and bone tissue are identified by real-time Raman spectroscopy, and the local temperature is continuously controlled to be below 47°C by infrared thermal imaging.

[0146] (4) The depth of osteotomy was dynamically monitored by OCT during the operation to avoid damage to the tooth root and nerve. The overall operation took about 20 minutes, the osteotomy error was ≤0.2mm, the chin shape adjustment accuracy met the standard, and the postoperative soft tissue swelling and bleeding were greatly reduced.

[0147] 4. Orthognathic surgery for skeletal deformities (comprehensive correction of both upper and lower jaws):

[0148] (1) Complete the three-dimensional reconstruction of the whole jaw CT and MRI for patients with malocclusion such as skeletal underbite and protruding mouth. AI combines occlusal data and maxillofacial anatomical parameters to automatically plan the maxillary Le Fort osteotomy line and mandibular sagittal split osteotomy trajectory based on reinforcement learning, and delineate the safe isolation zone for dental nerves, maxillary sinus and inferior alveolar vessels.

[0149] (2) GVSN trimodal navigation completes real-time registration of preoperative images and intraoperative tissues. The seven-degree-of-freedom robotic arm, equipped with a miniature flexible working head, enters the periphery of the maxillary sinus and the deep narrow cavity of the mandibular ramus, and flexibly changes direction by relying on the magnetic field.

[0150] (3) Layered switching of composite laser parameters: The flat-top laser completes the deep splitting of the thick jawbone, and the femtosecond laser performs micron-level fine cutting on the area near the nerve and tooth root. The system synchronously links the nerve electrophysiology and blood flow monitoring modules, and triggers the device to self-lock and stop instantly when the parameters are abnormal.

[0151] (4) Real-time compensation for intraoperative jaw soft tissue deformation, deformation compensation error <0.8mm, the time for a single orthognathic surgery is shortened by 50% compared with the traditional method, osteotomy accuracy ≤0.2mm, postoperative occlusal alignment accuracy ≥96%, and no major nerves or blood vessels are accidentally injured.

[0152] Compared with the prior art, the present invention has the following significant advantages:

[0153] First, breakthroughs in miniaturization and flexibility: With a magnetically controlled flexible working head with an outer diameter of ≤3mm, the problem of traditional instruments being unable to enter the narrow areas deep in the oral cavity has been solved, achieving 360° operation without blind spots.

[0154] Second, the accuracy is greatly improved: the combination of flat-top femtosecond laser and high-precision navigation can control the average deviation of osteotomy, improve the accuracy of navigation and positioning, and improve the facial symmetry after surgery.

[0155] Third, safety is significantly enhanced: based on trimodal fusion navigation and multidimensional sensing monitoring, the risk of neurovascular injury and thermal injury can be reduced.

[0156] Fourth, intelligent and standardized: AI-powered automatic planning can significantly shorten surgery time, reduce reliance on doctors' experience, and improve the predictability of surgical outcomes.

[0157] Fifth, the minimally invasive approach has significant advantages: the intraoral approach leaves no external scars, reduces intraoperative bleeding, shortens the patient's hospital stay, and results in higher quality bone healing.

[0158] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A laser-based bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation, characterized in that, include: Composite laser generating unit for outputting flat-top femtosecond composite laser; A miniature magnetically controlled flexible laser working head that can rotate flexibly within the oral cavity under the drive of an external magnetic field to achieve osteotomy and cutting in maxillofacial plastic surgery. A robotic arm execution unit used to drive a miniature magnetically controlled flexible laser working head to accurately reach the preset surgical site and move along the planned path; A multimodal sensing unit used to collect anatomical and physiological information of patients in all dimensions, providing multimodal data reference for surgical planning; Based on deep learning, an AI surgical planning unit automatically generates osteotomy paths, safety boundaries, and movement vectors to achieve intelligent surgical planning. A real-time navigation control unit used to fuse multimodal data to achieve human-machine collaborative control; And a human-computer interaction unit for doctors to monitor the surgical process in real time, adjust parameters, and handle emergencies; The composite laser generating unit interacts with the miniature magnetically controlled flexible laser working head, the robotic arm execution unit interacts with the miniature magnetically controlled flexible laser working head, and the robotic arm execution unit, multimodal perception unit, AI surgical planning unit, and human-computer interaction unit all interact with the real-time navigation control unit.

2. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The composite laser generating unit includes: Used for outputting continuous laser light Laser module; Beam shaping module for shaping continuous laser beams into flat-top laser beams; Femtosecond laser module for outputting femtosecond lasers; A beam combining module used to combine flat-top lasers and femtosecond lasers into a flat-top femtosecond composite laser; The The laser module and the beam shaping module are connected. Both the beam shaping module and the femtosecond laser module are connected to the beam combining module, and the beam combining module is connected to the micro-magnetically controlled flexible laser working head.

3. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The miniature magnet-controlled flexible laser working head includes: Multi-core mid-infrared optical fiber used as a laser transmission channel; An aspherical focusing lens is used to focus composite lasers emitted from multi-core mid-infrared optical fibers with high precision, forming a small spot of concentrated energy and clear edges on the bone surface. NdFeB magnetic ring used to rotate the front end of a multi-core mid-infrared optical fiber in conjunction with an external magnetic field, thereby changing the laser output direction. An OCT sensor used to emit near-infrared detection light and receive echo signals to acquire tomographic images of bone tissue in real time and measure cutting depth and interface changes. An infrared temperature sensor used to detect the infrared radiation intensity of the cut area surface in real time and convert it into a surface temperature value. And a Raman spectroscopy probe for emitting probe light into oral tissues and collecting scattering spectra, enabling real-time identification of tissue components based on differences in characteristic peaks, thus achieving soft tissue protection; The NdFeB magnetic ring, OCT sensor, infrared temperature sensor, and Raman spectroscopy probe are all mounted on a multi-core mid-infrared optical fiber, and the aspherical focusing lens is mounted at the very front end of the multi-core mid-infrared optical fiber.

4. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The robotic arm execution unit includes: a robotic arm with degrees of freedom, the end of which is equipped with a force sensor to drive a miniature magnetically controlled flexible laser working head to accurately reach a preset surgical site and move along a planned path.

5. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The multimodal sensing unit includes: Medical imaging equipment used to create three-dimensional bone models before surgery; Ultrasound equipment used to monitor soft tissue deformation during surgery; Infrared optical navigation equipment used for spatial positioning; Infrared thermal imager used for regional temperature field monitoring; Raman spectrometer for tissue composition analysis; And hemodynamic and neurophysiological monitoring equipment for real-time assessment of neurovascular status; The medical imaging equipment, ultrasound equipment, infrared optical navigation equipment, infrared thermal imager, and Raman spectrometer all interact with the real-time navigation control unit.

6. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The AI ​​surgical planning unit incorporates a maxillofacial surgery knowledge base built on a large language model and a reinforcement learning training framework.

7. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The real-time navigation control unit deploys a geometric, visual, and spectral three-modal fusion navigation algorithm.

8. A laser bone scalpel intelligent maxillofacial plastic surgery robot system based on AI and multimodal navigation as described in claim 1, characterized in that, The human-computer interaction unit is equipped with a touch screen, foot switch, three-dimensional visualization navigation interface, and emergency stop and manual takeover module, which facilitates doctors to monitor the surgical process, adjust parameters, and handle emergencies in real time.

9. A smart maxillofacial plastic surgery method based on the system described in any one of claims 1-8, comprising the following steps: S1: Preoperative AI 3D planning, collecting CT and MRI data of the patient's maxillofacial region, constructing a high-precision 3D skeletal and neurovascular model; the AI ​​surgical planning unit automatically generates osteotomy lines based on reinforcement learning strategies, sets safety margins, and outputs initial laser parameters and robotic arm motion trajectories; S2: Multimodal registration and navigation initialization: Spatial registration of preoperative image data with intraoperative optical navigation coordinate system, with registration error controlled within ≤0.5mm; Start GVSN trimodal fusion navigation to complete the initial alignment of geometric, visual, and spectral features; S3: Minimally invasive intraoral approach and positioning, controlling the robotic arm to drive the miniature magnetically controlled flexible laser working head through the oral cavity, using an external magnetic field to control the front end of the working head to rotate so that it conforms to the curved shape of the target bone; S4: Composite laser collaborative cutting and adaptive control. First, the flat-top laser is started to cut deep bone. After reaching the preset depth, the femtosecond laser is switched to refine the edges. The AI ​​control unit dynamically adjusts the laser power, pulse width and scanning speed according to the cutting depth fed back by OCT in real time and the tissue type identified by Raman spectroscopy. S5: Multi-dimensional safety closed-loop monitoring. During the operation, the infrared temperature measurement module scans the temperature of the cutting area at a frequency of 50ms / time to ensure that the surface temperature is ≤47℃; OCT monitors the depth of the heat-affected zone to ensure that it is ≤100μm; Raman spectroscopy distinguishes bone and nerve and blood vessel tissue in real time. When the working head is detected to be approaching the danger zone, the system automatically triggers an audible and visual warning and controls the robotic arm to brake. S6: Postoperative accuracy verification and evaluation. After the surgery, the system automatically compares the deviation between the actual osteotomy line and the planned path, and generates a report that includes osteotomy accuracy, facial symmetry and safety assessment to assist doctors in making postoperative judgments.

Images