A multi-wavelength AI control device, method, and system for minimally invasive spinal surgery

Abstract

The present invention is applicable to the field of spinal endoscope technology, and provides a multi-wavelength AI control device, method and system for minimally invasive spinal surgery, wherein the device comprises: a light source module for generating a plurality of irradiation lights; a working pipe, wherein the hollow interior of the working pipe is used to pass through a fiber optic imaging module to provide lighting for the imaging lens of the fiber optic imaging module; an AI analysis module for performing AI real-time analysis on the imaging image acquired by the fiber optic imaging module to obtain the tissue type in the imaging image; and a logic control module, which is preset with a plurality of light source control schemes. The advantage of this application is that, through the three-in-one method of multi-wavelength optical imaging, AI real-time analysis and adaptive light source control, the device significantly improves the accuracy of tissue identification, imaging stability of the surgical field and operational safety, and is particularly suitable for complex scenarios requiring high-precision anatomical separation.

Description

Technical Field

[0001] The present invention belongs to the technical field of spinal endoscopes, and in particular relates to a multi-wavelength AI control device, method and system for minimally invasive spinal surgery. Background Art

[0002] A spinal endoscope is a key tool used in minimally invasive spinal surgery. Its structural design is intended to enter the spinal canal or intervertebral foramen through a small incision, enabling precise observation and operation. A spinal endoscope generally includes an optical imaging system, a lighting system, an instrument channel, an irrigation / suction channel, an outer sheath and expansion system, and related auxiliary support systems. Through the interaction of the above structures, a spinal endoscope can complete complex operations such as disc removal, nerve decompression, and vertebral fusion under minimally invasive conditions, significantly reducing postoperative recovery time.

[0003] Imaging quality is a key factor in measuring the performance of spinal endoscopy systems. Current cutting-edge systems feature a wide range of auxiliary imaging methods. For example, they incorporate visual navigation systems, autonomous navigation technology based on computer vision guidance, and robotic propulsion mechanisms, using force sensors to monitor interventional resistance, reducing reliance on operator experience. Another example is the integration of deep learning algorithms for image enhancement, and AI-based optimization of original image color restoration to eliminate image defects such as artifacts and dark areas.

[0004] However, existing spinal endoscope imaging devices or methods often perform post-processing and data processing on the original imaging images, which cannot effectively improve the quality of the original imaging images. Therefore, it is necessary to further improve the existing spinal endoscopes. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a multi-wavelength AI control method for minimally invasive spinal surgery, aiming to solve the problem that existing spinal endoscopic imaging devices or methods often focus on post-processing of original imaging images and cannot effectively improve the quality of original imaging images.

[0006] The embodiment of the present application is implemented by providing a multi-wavelength AI control device for minimally invasive spinal surgery, the device comprising:

[0007] A light source module, used for generating a plurality of irradiation lights;

[0008] A working pipe, wherein the interior of the working pipe is hollow and is used to pass through the optical fiber imaging module. The inner wall of the working pipe is integrated with an optical fiber array, and the optical fiber array is used to guide the irradiation light to the end outlet of the working pipe and provide illumination for the imaging lens of the optical fiber imaging module;

[0009] An AI analysis module, configured to perform AI real-time analysis on the imaging images acquired by the optical fiber imaging module to obtain the tissue type in the imaging images; and

[0010] A logic control module is pre-configured with a plurality of light source control schemes, each of which is used to control the light source module to generate illumination light of a single wavelength or a combination of multiple wavelengths;

[0011] The logic control module is further used to obtain analysis results of the tissue types and select corresponding light source control schemes based on different analysis results.

[0012] Another object of the present invention is to provide a multi-wavelength AI control method for minimally invasive spinal surgery, the method comprising:

[0013] Obtaining a preset optical configuration, controlling the light source module to output a preset light based on the preset optical configuration, wherein the preset light is an illumination light of a single wavelength or a combination of multiple wavelengths, and adjusting the illumination light to a preset light intensity;

[0014] Acquire imaging images acquired by the imaging module in real time during surgery, and analyze tissue types in the imaging images in real time based on AI;

[0015] Obtaining a light source control scheme corresponding to the tissue type, wherein the light source control scheme is preset with several options, each of which is used to control a light source module to generate irradiation light of a single wavelength or a combination of multiple wavelengths;

[0016] The light source module is controlled to output illumination light based on the light source control scheme.

[0017] Another object of the embodiments of the present application is to provide a multi-wavelength AI control system for minimally invasive spinal surgery, the system comprising the multi-wavelength AI control device for minimally invasive spinal surgery as described above, and further comprising:

[0018] The optical fiber imaging module is used to collect imaging images and output the imaging images to the display and the AI analysis module at the same time.

[0019] The embodiment of the present application provides a multi-wavelength AI control device for minimally invasive spinal surgery, which has the following outstanding advantages: the light source module can provide a combination irradiation mode of near-infrared, visible light, ultraviolet light, etc. of different wavelengths, making full use of the differences in the absorption / reflection characteristics of different tissues for different wavelengths, and significantly enhancing the contrast of the image from the source. By dynamically switching the wavelength of the light source, layered imaging and automatic adjustment are achieved, which significantly improves the imaging quality and reduces the risk of misjudgment by medical staff. The AI analysis module can identify tissue types in real time and provide adaptive real-time light source switching control based on the actual scenario and needs of the operation. The trinity of multi-wavelength optical imaging, AI real-time analysis, and adaptive light source control significantly improves the accuracy of tissue identification, imaging stability of the surgical field, and operational safety, and is especially suitable for complex scenarios requiring high-precision anatomical separation. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of a multi-wavelength AI control device for minimally invasive spinal surgery provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of an imaging image of vascular tissue provided in an embodiment of the present application;

[0022] Figure 3 This is a flowchart of a multi-wavelength AI control method for minimally invasive spinal surgery provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present 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 only used to explain the present invention and are not intended to limit the present invention.

[0024] It will be understood that the terms "first," "second," etc., used herein may be used to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first unit or module from another unit or module. For example, a first light source may be referred to as a second light source, and similarly, a second light source may be referred to as a first light source without departing from the scope of this application.

[0025] In one embodiment of the present application, a multi-wavelength AI control device for minimally invasive spinal surgery is provided, which includes at least the following modules:

[0026] A light source module is used to generate a number of illumination lights; a working pipe, the interior of which is hollow for the fiber optic imaging module to pass through, and the inner wall of the working pipe is integrated with an optical fiber array, which is used to guide the illumination light to the end outlet of the working pipe and provide illumination for the imaging lens of the fiber optic imaging module; an AI analysis module is used to perform AI real-time analysis on the imaging image acquired by the fiber optic imaging module to obtain the tissue type in the imaging image; a logic control module is preset with several light source control schemes, each of which is used to control the light source module to generate illumination light of one wavelength or a combination of multiple wavelengths; the logic control module is also used to obtain the analysis results of the tissue type and select the corresponding light source control scheme based on different analysis results.

[0027] Those skilled in the art will recognize that current mainstream endoscope systems often use a single white light or single wavelength near-infrared light source. However, the researchers of this application considered that due to the different absorption characteristics of different tissues in the body, such as the yellow ligament, nerves, and blood vessels, traditional endoscope systems using a single light source, even when combined with various advanced auxiliary image processing methods, still find it difficult to effectively improve the image quality.

[0028] In this embodiment, the light source module uses a combination of near-infrared, visible light, ultraviolet, and other wavelengths of irradiation to fully utilize the differences in absorption / reflection characteristics of different tissues for different wavelengths, significantly enhance the image contrast from the source, dynamically switch the light source wavelength, achieve layered imaging and automatic adjustment, significantly improve the imaging quality, reduce the risk of misjudgment by medical staff, and avoid the problem of missing surgical field information caused by a single light source in traditional equipment.

[0029] In addition, traditional methods rely on the surgeon's naked eye judgment and are easily limited by experience. The AI analysis module provided by the device in this application identifies tissue types in real time and can provide adaptive real-time light source switching control based on the actual scenario and needs of the operation. Compared with traditional minimally invasive spinal technology, this application significantly improves the accuracy of tissue identification, imaging stability of the surgical field, and operational safety through the three-in-one design of multi-wavelength optical imaging, AI real-time analysis, and adaptive light source control. It is especially suitable for complex scenarios that require high-precision anatomical separation.

[0030] In this embodiment, if Figure 1 As shown in the figure, a schematic diagram of the whole and part of a multi-wavelength AI control device for minimally invasive spinal surgery is given. Figure 1Part a is a structural model of the entire device, while parts b, c, and d show the effects of the light source in white, blue, and yellow light modes, respectively. The working pipe is hollow and inserted into the patient's body during surgery. Its distal end is located near the surgical site and accommodates operating instruments and an optical imaging module. The optical imaging module can be a fiber optic imaging module for imaging, with an imaging lens at the end of the optical fiber. The AI analysis module can obtain images captured by the imaging lens.

[0031] In order to further reduce the time delay of AI analysis, the AI analysis model in the AI analysis module can be locally deployed and run in a general-purpose processor or a dedicated AI processing chip. The AI analysis model can use a pre-trained lightweight Transformer model to identify the types of nerves, blood vessels, fat and other tissues in the surgical field in real time. For example, the model can automatically identify and mark the following: Figure 2 The AI analysis model can analyze endoscopic images in real time and detect key anatomical structures in the field of view.

[0032] The logic control module's preset light source control schemes can be a logic control table that pre-stores the optimal wavelength combinations for each tissue type. This table switches wavelengths in real time based on signals output by the AI analysis module. It also adjusts the intensity of the illumination light. Specifically, after obtaining the tissue type output by the AI analysis module, it selects the light source control scheme corresponding to that tissue type. For example, when the AI identifies neural tissue in an image, it automatically disables the 405nm wavelength and activates the 540nm and 660nm wavelengths to prevent tissue damage and improve image resolution. Simultaneously, the light intensity is increased to a safe threshold—for example, adjusting the 540nm wavelength to 500mW and the 660nm wavelength to 300mW. For another example, when the ligamentum flavum is identified in an image, the wavelengths are adjusted to a combination of 540nm (the absorption peak of ligamentum flavum collagen) and 660nm (the low absorption region of adipose tissue). This improves edge contrast through differential absorption, and based on experimental data, this can increase display contrast by approximately 40%. For example, when a blood vessel is identified in an image, the system adjusts to a dual-wavelength combination of 760nm (deoxyhemoglobin absorption peak) and 850nm (oxyhemoglobin absorption peak), leveraging differences in hemoglobin absorption to distinguish between arteries and veins. This is particularly useful in special situations such as intraoperative bleeding. The core purpose of the light source control solution is to select a wavelength combination that maximizes the contrast between the target tissue and the background based on the optical properties of the tissue type.

[0033] Preferably, the device in the present application may also include other optical structures, such as a fiber coupler: using a multi-channel fiber bundle, preferably with a core diameter of 200 μm and NA = 0.22, for coupling different light sources in the light source module, such as an LED light source and a laser light source output light, to the same transmission fiber to reduce optical path crosstalk. Another example is a collimating lens group: an aspheric lens is provided at the front end of each light source to collimate the divergent light into a parallel beam, thereby improving light energy utilization and achieving higher coupling efficiency.

[0034] In a preferred embodiment, the light source module includes: a tunable laser and several fixed wavelength light sources;

[0035] The tunable laser can emit light in the wavelength range of 280nm-1600nm;

[0036] The fixed wavelength light source includes at least one of 405 nm, 540 nm, 660 nm, 760 nm, 850 nm and 940 nm wavelength light sources.

[0037] In an embodiment of the present application, a fixed wavelength light source can be an LED light source group, and the LED light source group and the laser are arranged in a ring array and symmetrically distributed around the endoscope optical channel to ensure that light is evenly projected onto the surgical area. The LED light source group can include 6 groups of high-power LEDs at the same time, namely 405nm, 540nm, 660nm, 760nm, 850nm, and 940nm, arranged in ascending order of wavelength, each group is equipped with an independent heat sink, and is connected to the thermoelectric cooler through a copper substrate. The tunable laser can be located at the center of the array, using a distributed feedback laser with a wavelength coverage of 280-1600nm. The wavelength is precisely tuned by a fiber Bragg grating, and the accuracy can be controlled within ±1nm.

[0038] In a preferred embodiment, the working pipe is transparent or translucent and is made of a composite material of polypropylene and polyamide; the inner wall of the working pipe is coated with an anti-reflection film.

[0039] In this embodiment, the working pipe is a 7:3 blend of polypropylene (PP) and polyamide (PA6), with 1 wt% nano-SiO2 added for improved wear resistance. The entire structure is injection molded at 220°C and 80 MPa, and the inner wall is coated with a 100 nm thick MgF2 anti-reflection coating.

[0040] In a preferred embodiment, Figure 3 As shown, a multi-wavelength AI control method for minimally invasive spinal surgery is proposed. It can be understood that this method can be applied to the multi-wavelength AI imaging device for minimally invasive spinal surgery. The method can be run in a general-purpose processor or a dedicated AI processing chip or module in the device, and specifically may include the following steps:

[0041] Step S10, obtaining a preset optical configuration, and controlling the light source module to output a preset light based on the preset optical configuration, wherein the preset light is an illumination light of a single wavelength or a combination of multiple wavelengths, and adjusting the illumination light to a preset light intensity.

[0042] In this embodiment, the preset optical configuration is obtained by preloading tissue optical parameters based on the patient's CT / MRI data. For example, based on preoperative CT / MRI data, the tissue's absorption capacity for specific wavelengths is quantified, and diffuse optical tomography (DOT) is used to assess tissue scattering properties, optimize light source penetration depth, and generate an initial wavelength plan.

[0043] In an embodiment of the present application, the method for calculating the initial light intensity can be: estimating the amount of bleeding (i.e., depth) based on preoperative CT / MRI data, and then calculating the penetration depth that different incident light intensities can have according to the Beer-Lambert law to obtain the initial light intensity.

[0044] Step S20: Obtain an imaging image acquired by the imaging module in real time during the operation, and analyze the tissue type in the imaging image in real time based on AI.

[0045] In this embodiment, the AI real-time analysis model significantly improves the accuracy and safety of spinal endoscopic surgery through anatomical structure recognition and dynamic wavelength matching.

[0046] Step S30 , obtaining a light source control scheme corresponding to the tissue type. The light source control scheme is preset with several items, each of which is used to control the light source module to generate irradiation light of one wavelength or a combination of irradiation light of multiple wavelengths.

[0047] Step S40: controlling the light source module to output illumination light based on the light source control scheme.

[0048] In an embodiment of the present application, the AI model identifies tissue types through real-time image analysis, dynamically matches the optimal light source solution, and combines multi-wavelength combined irradiation to simultaneously enhance the development effects of different tissues. The light source module can provide a combination of irradiation modes of near-infrared, visible light, ultraviolet light, etc. of different wavelengths, making full use of the differences in the absorption / reflection characteristics of different tissues for different wavelengths, and significantly enhancing the contrast of the image from the source. By dynamically switching the wavelength of the light source, layered imaging and automatic adjustment are achieved, which significantly improves the imaging quality and reduces the risk of misjudgment by medical staff. The AI analysis module can identify tissue types in real time and provide adaptive real-time light source switching control based on the actual scenario and needs of the operation. The trinity of multi-wavelength optical imaging, AI real-time analysis, and adaptive light source control significantly improves the accuracy of tissue identification, imaging stability of the surgical field, and operational safety, and is especially suitable for complex scenarios requiring high-precision anatomical separation.

[0049] In a preferred embodiment, the light source control scheme includes:

[0050] When the tissue type is identified as nerve, the light source module outputs an irradiation light with a wavelength of 660nm and disables irradiation light with a wavelength in the ultraviolet band;

[0051] When the tissue type is identified as ligamentum flavum, the light source module outputs irradiation light with wavelengths of 540nm and 660nm;

[0052] When the tissue type is identified as blood vessels, the light source module outputs irradiation light with wavelengths of 760nm and 860nm;

[0053] When the tissue type is identified as an intervertebral disc, the wavelength of the illumination light output by the light source module is 980nm;

[0054] When the tissue type is identified as a collagen denaturation area, the light source module outputs irradiation light with wavelengths of 280 and 980 nm;

[0055] When the tissue type is identified as adipose tissue, the wavelength of the irradiation light output by the light source module is 660nm;

[0056] When the tissue type is identified as bony structure, the wavelength of the illumination light output by the light source module is 850nm.

[0057] In an embodiment of the present application, corresponding control schemes for irradiation light are set for different tissue types to further improve the level of intelligence and lighting safety. For example, if neural tissue is detected, the ultraviolet band is automatically shielded, and the safe wavelength of 660nm can be selected to enable.

[0058] As a preferred embodiment of the present application, when there are multiple types of tissues in the field of view, after determining the tissue type, a contrast judgment is also required. Only when the contrast is lower than the threshold is the illumination light switched to avoid frequent switching.

[0059] In this embodiment, intelligent decision-making is implemented. First, AI-based real-time analysis identifies the most important or focal tissue types in the image. Contrast evaluation is then performed, with the contrast index, such as Weber contrast, calculated in real time for the current wavelength combination. If the contrast meets the required level, switching is temporarily suspended even if the tissue type in the field of view changes. When the contrast falls below a preset contrast threshold, a switch command is triggered to switch to a different light source illumination mode to achieve greater contrast.

[0060] In a preferred embodiment, the system also includes overheat protection. For example, if the proportion of saturated pixels in any wavelength image is greater than 5%, the light intensity is automatically reduced to 80% of the original value; if the temperature sensor is high in temperature, the power reduction mode is triggered.

[0061] In a preferred embodiment, different light source control schemes are preset with the same or different control priorities; the light source control scheme corresponding to arteriovenous bleeding of the tissue type has the highest priority;

[0062] When multiple different tissue types are simultaneously imaged, the light source control scheme with the highest control priority is selected first.

[0063] In this embodiment, different priorities are set to address emergencies such as sudden bleeding or prioritize the visualization of critical tissue types. For example, if a disc is to be removed, penetrating imaging at 980nm (water absorption peak) is enabled, combined with 280nm (collagen characteristic absorption) to locate the lesion. If vascular bleeding is detected, the system automatically switches to dual-wavelength mode at 760nm (deoxyhemoglobin absorption peak) and 850nm (oxyhemoglobin absorption peak) to enhance vascular visualization.

[0064] In a preferred embodiment, the method for analyzing the tissue type in the imaging image is:

[0065] Analyzing the tissue type based on a pre-trained lightweight Transformer model;

[0066] The encoder of the lightweight Transformer model is used to extract local features of multi-wavelength images, the attention mechanism is used to weightedly focus on key areas in the imaging image, and the classification head is used to output tissue type and judgment probability.

[0067] In the embodiments of this application, a lightweight Transformer model is used to identify tissue types such as nerves, blood vessels, and fat in the surgical field of view in real time with extremely low latency. This model can determine tissue type through the following feature analyses: Texture analysis, which extracts tissue texture such as the fibrous structure of the yellow ligament and the uniformity of fat through the gray-level co-occurrence matrix; Edge gradient, which calculates the image gradient amplitude to identify the sharpness of the boundary between the yellow ligament and surrounding tissue; Color distribution, which statistically analyzes the distribution of pixel intensities at different wavelengths in RGB or multispectral images.

[0068] In the embodiments of this application, based on the above model, the system can capture dynamic features: for example, real-time bleeding detection can be performed by monitoring pixel mutations in the red channel of the image to determine the bleeding area. The attenuation rate of light in the 940nm near-infrared band can be used to assess the dynamic changes in tissue water content and identify edema changes in key areas.

[0069] In a preferred embodiment, the method further comprises:

[0070] The tissue type in the imaging image acquired in real time is obtained, the thermal damage threshold corresponding to the tissue type is obtained, and a safe range of the irradiation light intensity is set based on the thermal damage threshold.

[0071] In the embodiment of the present application, according to the ISO15004-2 standard, the optical power density on the tissue surface is ≤1W / cm². For example, when the spot diameter of a 760nm beam is 3mm, 800mW corresponds to a power density of ≈1.13W / cm², which is close to the upper limit of safety.

[0072] In a preferred embodiment, the method further comprises:

[0073] Acquiring contrast data and signal-to-noise ratio data of the imaging image acquired in real time, and controlling the illumination intensity of the illumination light;

[0074] The control modes for controlling the illumination light intensity include a contrast priority mode and a signal-to-noise ratio priority mode;

[0075] In the signal-to-noise ratio priority mode, a signal-to-noise ratio histogram of the imaging image is obtained based on the signal-to-noise ratio data, and the illumination light intensity is adjusted based on the signal-to-noise ratio histogram to reduce the image signal-to-noise ratio;

[0076] In the contrast priority mode, the illumination light intensity is adjusted using a gradient descent algorithm based on the contrast data to improve the contrast of the image.

[0077] In this embodiment, the Weber contrast of a dual-wavelength image can be calculated. If the contrast is less than a certain value, the light intensity is adjusted using a gradient descent algorithm. For example, at 760nm, the intensity is increased by 50mW per gradient, while at 850nm, the intensity is decreased by 30mW per gradient. The signal-to-noise ratio can be determined by analyzing the image histogram. The ultimate goal of these adjustments is to improve image clarity.

[0078] In the embodiment of the present application, a curve corresponding to the relationship between light intensity and contrast can be obtained based on experiments. For example, an in vitro blood vessel experiment shows that when the 760nm light intensity increases from 500mW to 800mW, the contrast increases from about 0.4 to about 0.75.

[0079] In a preferred embodiment, the system includes the multi-wavelength AI control device for minimally invasive spinal surgery as described above, and further includes:

[0080] The optical fiber imaging module is used to collect imaging images and output the imaging images to the display and the AI analysis module at the same time.

[0081] In this embodiment, the system integrates both an imaging system and an imaging AI lighting system. The advantage of this system is that the light source module can provide a combination of irradiation modes of near-infrared, visible light, ultraviolet light, etc. of different wavelengths, making full use of the differences in the absorption / reflection characteristics of different tissues for different wavelengths, and significantly enhancing the contrast of the image from the source. By dynamically switching the wavelength of the light source, layered imaging and automatic adjustment are achieved, which significantly improves the imaging quality and reduces the risk of misjudgment by medical staff. The AI analysis module can identify tissue types in real time and provide adaptive real-time light source switching control based on the actual scenario and needs of the surgery. The trinity of multi-wavelength optical imaging, AI real-time analysis, and adaptive light source control significantly improves the accuracy of tissue identification, imaging stability of the surgical field, and operational safety, and is especially suitable for complex scenarios requiring high-precision anatomical separation.

[0082] It should be understood that, although each step in the flow chart of each embodiment of the present application is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is clear explanation in this article, the execution of these steps does not have strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0083] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0084] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0085] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A multi-wavelength AI control method for minimally invasive spinal surgery, characterized in that: The method comprises: Obtaining a preset optical configuration, controlling the light source module to output a preset light based on the preset optical configuration, wherein the preset light is an illumination light of a single wavelength or a combination of multiple wavelengths, and adjusting the illumination light to a preset light intensity; Acquire imaging images acquired by the imaging module in real time during surgery, and analyze tissue types in the imaging images in real time based on AI; Obtaining a light source control scheme corresponding to the tissue type, wherein the light source control scheme is preset with several options, each of which is used to control a light source module to generate irradiation light of a single wavelength or a combination of multiple wavelengths; Control the light source module to output illumination light based on the light source control scheme; Different light source control schemes are preset with the same or different control priorities; the light source control scheme corresponding to arteriovenous bleeding as the tissue type has the highest priority; When multiple different tissue types are simultaneously imaged, the light source control scheme with the highest control priority is selected first; The method further comprises: Obtaining a tissue type in an imaging image acquired in real time, obtaining a thermal damage threshold corresponding to the tissue type, and setting a safe range of irradiation light intensity based on the thermal damage threshold; The method further comprises: Acquiring contrast data and signal-to-noise ratio data of the imaging image acquired in real time, and controlling the illumination intensity of the illumination light; The control modes for controlling the illumination light intensity include a contrast priority mode and a signal-to-noise ratio priority mode; In the signal-to-noise ratio priority mode, a signal-to-noise ratio histogram of the imaging image is obtained based on the signal-to-noise ratio data, and the illumination light intensity is adjusted based on the signal-to-noise ratio histogram to reduce the image signal-to-noise ratio; In the contrast priority mode, the illumination light intensity is adjusted using a gradient descent algorithm based on the contrast data to improve the contrast of the image.

2. The multi-wavelength AI control method for minimally invasive spinal surgery according to claim 1 is characterized in that: The light source control scheme includes: When the tissue type is identified as nerve, the light source module outputs an irradiation light with a wavelength of 660nm and disables irradiation light with a wavelength in the ultraviolet band; When the tissue type is identified as ligamentum flavum, the light source module outputs irradiation light with wavelengths of 540nm and 660nm; When the tissue type is identified as blood vessels, the light source module outputs irradiation light with wavelengths of 760nm and 860nm; When the tissue type is identified as an intervertebral disc, the wavelength of the illumination light output by the light source module is 980nm; When the tissue type is identified as a collagen denaturation area, the light source module outputs irradiation light with wavelengths of 280 and 980 nm; When the tissue type is identified as adipose tissue, the wavelength of the irradiation light output by the light source module is 660nm; When the tissue type is identified as bony structure, the wavelength of the illumination light output by the light source module is 850nm.

3. The multi-wavelength AI control method for minimally invasive spinal surgery according to claim 1 is characterized in that: The method for analyzing the tissue type in the imaging image is: Analyzing the tissue type based on a pre-trained lightweight Transformer model; The encoder of the lightweight Transformer model is used to extract local features of multi-wavelength images, the attention mechanism is used to weightedly focus on key areas in the imaging image, and the classification head is used to output tissue type and judgment probability.

4. A multi-wavelength AI control device for minimally invasive spinal surgery, characterized in that: The device is controlled and operated based on the method according to any one of claims 1 to 3, and the device comprises: A light source module, used for generating a plurality of irradiation lights; A working pipe, wherein the interior of the working pipe is hollow and is used to pass through the optical fiber imaging module. The inner wall of the working pipe is integrated with an optical fiber array, and the optical fiber array is used to guide the irradiation light to the end outlet of the working pipe and provide illumination for the imaging lens of the optical fiber imaging module; An AI analysis module, configured to perform AI real-time analysis on the imaging images acquired by the optical fiber imaging module to obtain the tissue type in the imaging images; and A logic control module is pre-configured with a plurality of light source control schemes, each of which is used to control the light source module to generate illumination light of a single wavelength or a combination of multiple wavelengths; The logic control module is further used to obtain analysis results of the tissue types and select corresponding light source control schemes based on different analysis results.

5. The multi-wavelength AI control device for minimally invasive spinal surgery according to claim 4 is characterized in that: The light source module includes: a tunable laser and several fixed wavelength light sources; The tunable laser can emit light in the wavelength range of 280nm-1600nm; The fixed wavelength light source includes at least one of 405 nm, 540 nm, 660 nm, 760 nm, 850 nm and 940 nm wavelength light sources.

6. The multi-wavelength AI control device for minimally invasive spinal surgery according to claim 4, characterized in that: The working pipe is transparent or translucent and is made of a composite material of polypropylene and polyamide; the inner wall of the working pipe is coated with an anti-reflection film.

7. A multi-wavelength AI control system for minimally invasive spinal surgery, characterized by: The system includes a multi-wavelength AI control device for minimally invasive spinal surgery according to any one of claims 4 to 6, and further includes: The optical fiber imaging module is used to collect imaging images and output the imaging images to the display and the AI analysis module at the same time.

Images