Intelligent intraocular lens implantation navigation system with self-adaptive optical calibration and method of intelligent intraocular lens implantation navigation system

Abstract

The invention relates to the field of medical instruments, in particular to an intelligent intraocular lens implantation navigation system and method for adaptive optical calibration, and the system comprises a microscope module which is used for collecting real-time image data in eyes; the adaptive optical system module is in signal connection with the microscope module and is used for evaluating and correcting optical aberration distribution of intraocular tissues in real time; the three-dimensional deep learning motion compensation module is used for predicting eyeball motion and compensating micro-motion interference; the registration algorithm module is used for establishing spatial registration between the OCT and the intraocular image system; the augmented reality projection module is used for superposing a registration guide line on the imaging of the operation microscope; the intelligent intraocular image system is used for identifying intraocular tissue structures and instrument positions, accurate implantation of an intraocular lens is achieved through multi-modal fusion perception, deep learning intelligent analysis, self-adaptive optical calibration and augmented reality guidance, and practical application shows that the precision of the implantation position of the lens can be improved to + / -0.1 mm.

Description

Technical Field

[0001] The present invention relates to the field of medical devices, and particularly to an intelligent intraocular lens implantation navigation system with adaptive optical calibration and its method. Background Art

[0002] With the acceleration of the social aging process, cataract has become one of the main blinding eye diseases globally. In cataract surgery, the precise implantation of the intraocular lens has a decisive impact on the postoperative visual quality. In traditional cataract surgery, the implantation of the intraocular lens mainly relies on the doctor's experience, lacks precise navigation, and is difficult to cope with the intraoperative micro-movement of the eyeball and the change of optical aberration, resulting in inaccurate positioning of the intraocular lens and reduced postoperative visual quality. Especially for patients with corneal astigmatism, it is even more difficult to achieve satisfactory results.

[0003] In the existing technology, there are already some surgical navigation systems, but they generally have the following problems: First, they lack the ability of real-time optical calibration and cannot cope with the intraoperative change of optical aberration; second, the compensation for the micro-movement of the eyeball is insufficient, affecting the navigation accuracy; third, the multi-modal data fusion is imperfect and the spatial registration accuracy is limited; fourth, there is a lack of intuitive augmented reality guidance, resulting in a high operation complexity.

[0004] Therefore, there is an urgent need for an intelligent navigation system that can calibrate optical aberration in real time, accurately compensate for the micro-movement of the eyeball, achieve precise spatial registration, and provide intuitive surgical guidance, so as to improve the implantation accuracy of the intraocular lens and the postoperative visual quality. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent intraocular lens implantation navigation system with adaptive optical calibration and its method. The system solves the problems existing in the prior art through multi-modal fusion perception, deep learning intelligent analysis, adaptive optical calibration, and augmented reality guidance, and improves the accuracy of intraocular lens implantation and the postoperative visual quality.

[0006] The present invention proposes an intelligent intraocular lens implantation navigation system with adaptive optical calibration, including:

[0007] A microscope module for collecting real-time intraocular image data;

[0008] An adaptive optical system module, which is signal-connected to the microscope module, for evaluating the distribution of optical aberration of intraocular tissues in real time and correcting it;

[0009] A three-dimensional deep learning motion compensation module for predicting the eyeball movement and compensating for the micro-movement interference;

[0010] A registration algorithm module for establishing spatial registration between OCT and the intraocular imaging system;

[0011] An augmented reality projection module for superimposing registration guiding lines on the imaging of the surgical microscope;

[0012] Intelligent intraocular imaging system for identifying the intraocular tissue structure and the position of instruments.

[0013] Preferably, the adaptive optical system module includes:

[0014] An optical aberration evaluation unit for training a virtual optical imaging module of intraocular images based on image data and evaluating the distribution of intraocular tissue optical aberrations in real time;

[0015] A lens control unit for generating a control strategy for the profile of the intraocular lens based on the optical aberration evaluation;

[0016] An imaging quality adjustment unit for adjusting the imaging quality and the shape of the intraocular lens in real time.

[0017] Preferably, the three-dimensional deep learning motion compensation module includes:

[0018] A feature extraction network unit for training a feature extraction network of the cornea / lens / iris / pupil;

[0019] A motion feature parameter unit for obtaining the eye movement feature parameters;

[0020] A depth particle filter unit for estimating the displacement and attitude angle of the intraocular surgical instrument;

[0021] A compensation matrix generation unit for generating a compensation matrix based on the virtual projection correction module.

[0022] Preferably, the registration algorithm module includes:

[0023] A spatial mapping unit for establishing a spatial image mapping relationship between the intraocular OCT and the intraocular imaging system;

[0024] A guiding line calculation unit for calculating a guiding line according to the spatial position, three-dimensional attitude of the intraocular instrument / lens and the image mapping relationship between the optical channels.

[0025] Preferably, the augmented reality projection module includes:

[0026] A guiding line generation unit for generating a guiding line based on OCT and intraocular imaging;

[0027] An attitude recognition unit for recognizing the spatial attitude of the intraocular surgical instrument;

[0028] An image fusion unit for superimposing a registration guiding line on the surgical microscope imaging;

[0029] A visual tracking unit for visually tracking using the optical features of the intraocular instrument to obtain the registration relationship between the registration guiding line and the surgical instrument.

[0030] Preferably, the intelligent intraocular imaging system includes:

[0031] A convolutional neural network unit for designing a feature extraction algorithm to identify intraocular surgical instruments, the lens, the cornea, the crystalline lens, the pupil, the iris, and the aqueous humor;

[0032] A contour generation unit for segmenting the three-dimensional contour of intraocular tissues based on OCT images;

[0033] A deep neural network model unit for generating a trained deep neural network model;

[0034] A verification unit for performing key point matching between the OCT image and the contour image segmented from the intraocular image to verify the model accuracy.

[0035] Preferably, it further includes:

[0036] An eyeball movement prediction module for constructing a central coordinate model of the intraocular lens based on the optical imaging principle and spatial registration of intraocular surgery;

[0037] A virtual projection correction module for real-time tracking of the spatial position between the intraocular surgical instrument and the intraocular lens and designing a coordinate transformation mechanism based on optical projection.

[0038] Preferably, the eyeball movement prediction module includes:

[0039] A central coordinate model unit for constructing a central coordinate model of the intraocular lens;

[0040] A deep learning model unit for obtaining eyeball movement feature parameters by means of a deep learning network model;

[0041] An estimation unit for estimating the displacement and attitude angle of the intraocular surgical instrument using a depth particle filter algorithm;

[0042] A tracking unit for tracking OCT data and estimating the spatial position of the intraocular surgical instrument.

[0043] Preferably, the virtual projection correction module includes:

[0044] A spatial position tracking unit for real-time tracking of the spatial position between the intraocular surgical instrument and the intraocular lens;

[0045] An image acquisition unit for acquiring real-time intraocular images according to the position of the intraocular instrument;

[0046] A coordinate transformation unit for designing a coordinate transformation mechanism based on optical projection;

[0047] Precision calibration unit, for tracking precision calibration through the eye movement prediction module and spatial registration.

[0048] Intelligent intraocular lens implantation navigation method with adaptive optical calibration, adopting the said method, including the following steps:

[0049] S100, acquiring real-time intraocular image data;

[0050] S200, evaluating and correcting the optical aberration distribution of intraocular tissues in real time through the adaptive optical system;

[0051] S300, predicting eye movement and compensating for micro-motion interference by using the three-dimensional deep learning motion compensation module;

[0052] S400, establishing spatial registration between OCT and the intraocular imaging system through the registration algorithm module;

[0053] S500, generating a guiding line based on OCT and intraocular imaging;

[0054] S600, identifying the spatial posture of the intraocular surgical instrument;

[0055] S700, superimposing the registration guiding line on the surgical microscope imaging;

[0056] S800, performing visual tracking by using the optical characteristics of the intraocular instrument to obtain the registration relationship between the registration guiding line and the surgical instrument;

[0057] S900, identifying the intraocular tissue structure and the position of the instrument by using the intelligent intraocular imaging system;

[0058] S1000, adjusting the position of the surgical instrument in real time according to the registration relationship between the registration guiding line and the surgical instrument to complete the precise implantation of the intraocular lens.

[0059] Compared with the prior art, the present invention has the following remarkable advantages:

[0060] The intelligent intraocular lens implantation navigation system and method with adaptive optical calibration of the present invention realizes the precise implantation of the intraocular lens through multi-modal fusion perception, deep learning intelligent analysis, adaptive optical calibration and augmented reality guidance, and solves the problems of insufficient optical aberration correction, insufficient eye micro-motion compensation, limited spatial registration accuracy and unintuitive surgical guidance existing in the traditional technology. Practical applications show that the system can improve the accuracy of the intraocular lens implantation position to ±0.1 mm, the angular accuracy to ±2°, reduce the operation time by 30%, and significantly improve the postoperative visual quality, providing better treatment effects for cataract patients. Description of the Drawings

[0061] Figure 1This is the overall architecture diagram of the intelligent intraocular lens implantation navigation system with adaptive optical calibration of the present invention;

[0062] Figure 2 This is the structural diagram of the adaptive optical system module;

[0063] Figure 3 This is the workflow diagram of the three-dimensional deep learning motion compensation module;

[0064] Figure 4 This is the implementation schematic diagram of the registration algorithm module;

[0065] Figure 5 This is the schematic diagram of the augmented reality projection module;

[0066] Figure 6 This is the recognition effect diagram of the intelligent intraocular imaging system;

[0067] Figure 7 This is the structural diagram of the eye movement prediction module;

[0068] Figure 8 This is the working schematic diagram of the virtual projection correction module;

[0069] Figure 9 This is the flowchart of the navigation method of the present invention;

[0070] Figure 10 This is the comparison diagram of clinical application effects. Detailed implementation manner

[0071] Please refer to the appendix Figures 1 - 10 , The intelligent intraocular lens implantation navigation system with adaptive optical calibration provided by the present invention mainly includes six core modules: microscope module 1, adaptive optical system module 2, three-dimensional deep learning motion compensation module 3, registration algorithm module 4, augmented reality projection module 5 and intelligent intraocular imaging system 6. These six modules form a closed-loop control system through signal connection to achieve precise navigation of intraocular lens implantation.

[0072] The microscope module 1 is the basic sensing device of the whole system. It adopts a high-resolution digital microscope with a field of view of 12mm×9mm, a resolution of 4K, and a frame rate of not less than 60fps to ensure obtaining clear real-time intraocular images. This module is connected to the adaptive optical system module 2 through a dedicated interface to transmit the collected original image data in real time.

[0073] The adaptive optical system module 2 is one of the core innovative modules of the present invention. It receives the image data from the microscope module 1 and evaluates and corrects the optical aberration distribution of intraocular tissues in real time through advanced optical processing algorithms. This module can cope with the optical aberration caused by intraoperative environmental changes, ensuring image quality and accuracy, which is particularly important for complex cases (such as patients with high astigmatism).

[0074] The three-dimensional deep learning motion compensation module 3 is responsible for predicting eye movements and compensating for micro-motion interference. Micro-movements of the eye are a key factor affecting surgical precision, especially during the implantation of intraocular lenses. This module uses deep learning algorithms to predict the trajectory of eye micro-movements, generate compensation parameters, and cooperate with surgical instrument control to offset the impact of micro-movements.

[0075] The registration algorithm module 4 establishes spatial registration between the OCT and the intraocular imaging system. During surgery, it is often necessary to fuse data from two different modalities, OCT and microscopic images. This module uses precise spatial registration algorithms to establish an accurate correspondence between the two imaging modalities, providing stereoscopic spatial coordinates for surgical navigation.

[0076] The augmented reality projection module 5 superimposes registration guiding lines on the surgical microscope imaging. This module superimposes the calculated optimal implantation path and position on the real-time microscope image in an augmented reality manner, providing intuitive visual guidance for the doctor, simplifying the operation process, and improving surgical precision.

[0077] The intelligent intraocular imaging system 6 is responsible for identifying intraocular tissue structures and the position of instruments. This system uses deep learning methods to identify and track various intraocular tissue structures and surgical instruments in real time, providing basic data support for other modules.

[0078] Each module is connected through a high-speed data bus to ensure real-time data transmission with a latency controlled within 10 ms to meet the strict requirements of real-time surgical navigation.

[0079] The adaptive optical system module 2 is one of the key innovations of the present invention. Its core function is to evaluate and correct intraocular optical aberrations in real time to ensure image quality and navigation accuracy. This module mainly includes three functional units: the optical aberration evaluation unit 21, the lens control unit 22, and the imaging quality adjustment unit 23.

[0080] The optical aberration evaluation unit 21 trains a virtual optical imaging module of the intraocular image based on image data to evaluate the distribution of intraocular tissue optical aberrations in real time. This unit uses a deep convolutional neural network to construct an optical aberration distribution prediction model, and its core algorithm can be expressed as:

[0081] W(x,y) = f CNN (I(x,y),θ),

[0082] Among them, W(x, y) represents the predicted aberration distribution function, I(x, y) is the input intraocular image, and θ are the network parameters. The network uses ResNet-50 as the backbone network and enhances the aberration feature extraction ability by adding an adaptive feature fusion layer. The training data comes from the intraocular images of 1000 cataract patients with different degrees, and the labeled data is provided by the measurement results of the wavefront aberrometer. This unit can complete the aberration evaluation of a single-frame image within 5 ms, which is much faster than the 50 ms response time of traditional wavefront sensors.

[0083] Based on the optical aberration evaluation results, the lens control unit 22 generates a control strategy for the artificial lens contour. This unit optimizes the lens shape adjustment parameters through a deep reinforcement learning algorithm, and the algorithm framework is as follows:

[0084] A t = π(S t |θ π ),

[0085] Among them, A t represents the control action at time t, S t is the current state (including the current aberration distribution and lens position), π is the policy function, and θ π are the policy network parameters. The reward function of this algorithm is designed as a weighted combination of the aberration improvement degree and the operation complexity. Practice shows that this method can improve the lens position control accuracy to ±0.1 mm and the angle error to within ±2 degrees, which is significantly better than the traditional manual adjustment method.

[0086] The imaging quality adjustment unit 23 is responsible for real-time adjustment of the imaging quality and the artificial lens shape. This unit receives the control instructions from the lens control unit 22 and realizes the precise adjustment of the imaging system and the lens shape through a micro actuator. The typical adjustment period is 20 ms, which can effectively cope with intraoperative environmental changes and maintain stable image quality.

[0087] The three-dimensional deep learning motion compensation module 3 is another innovation focus of the present invention, which solves the problem that traditional systems are difficult to cope with the micro motion of the eyeball. This module includes four functional units: a feature extraction network unit 31, a motion feature parameter unit 32, a depth particle filter unit 33, and a compensation matrix generation unit 34.

[0088] The feature extraction network unit 31 is responsible for training the feature extraction network of the cornea / lens / iris / pupil. This unit adopts a multi-branch cascaded convolutional network architecture and designs dedicated feature extraction sub-networks for different intraocular tissues. The network training adopts a transfer learning strategy, first pre-training on a large-scale general ophthalmic image dataset, and then fine-tuning on a surgery-specific dataset. This network can maintain a feature extraction accuracy of more than 95% under different lighting conditions and tissue states.

[0089] Based on the feature extraction results, the motion feature parameter unit 32 obtains the eye movement feature parameters. This unit extracts the frequency, amplitude, and pattern features of eye movement through time series analysis. The main parameters include: microtremors (frequency 40 - 100 Hz, amplitude 0.1 - 5.0 μm), drifts (frequency 0.1 - 0.5 Hz, amplitude 20 - 100 μm), and microsaccades (frequency 1 - 5 Hz, amplitude 10 - 20 μm). These parameters form the eye movement feature vector, providing a basis for subsequent compensation.

[0090] The depth particle filter unit 33 innovatively integrates deep learning and the particle filter algorithm to estimate the displacement and attitude angle of the intraocular surgical instrument. Its core algorithm is as follows:

[0091]

[0092] where x t represents the state vector at time t (including position and attitude information), y 1:t is the observation sequence, and are the weight and state of the i-th particle respectively, N is the number of particles (set to 1000 in this system), and δ is the Dirac function. This unit optimizes the particle generation and weight update strategies through a deep neural network, solves the degradation problem of traditional particle filters in eye micro-motion estimation, improves the tracking accuracy by 40%, and doubles the computational efficiency.

[0093] Based on the results of the depth particle filter, the compensation matrix generation unit 34 generates the compensation matrix for the virtual projection correction module. The compensation matrix is calculated using the following formula:

[0094] C = R · T · S

[0095] where C is the compensation matrix, R is the rotation matrix, T is the translation matrix, and S is the scaling matrix. This matrix is used to adjust the position and attitude of the guiding line to offset the influence of eye micro-motion. In practical applications, this compensation technology can control the navigation error within 0.1 mm, meeting the requirements of high-precision surgery.

[0096] The registration algorithm module 4 is responsible for establishing the spatial registration between the OCT and the intraocular imaging system, which is the basis for achieving precise navigation. This module includes two functional units: the spatial mapping unit 41 and the guiding line calculation unit 42.

[0097] The spatial mapping unit 41 establishes the spatial image mapping relationship between the intraocular OCT and the intraocular imaging system. This unit uses a multi-point matching initial registration method based on anatomical features and optimizes the registration accuracy through the iterative closest point algorithm. The registration transformation can be expressed as:

[0098]

[0099] where T is the transformation matrix, p i and q i are the corresponding feature points in the OCT image and the intraocular image respectively, and n is the number of feature points. To cope with tissue deformation, this unit further introduces a non-rigid transformation model to describe local deformation through B-spline functions. This method improves the spatial registration accuracy to ±0.1 mm, greatly exceeding the existing technical level.

[0100] Based on the spatial registration result, the guide wire calculation unit 42 calculates the guide wire according to the spatial position, three-dimensional attitude of the intraocular instrument / lens, and the image mapping relationship between the optical channels. The generation of the guide wire adopts the following steps: First, determine the ideal implantation position and attitude of the lens; Second, calculate the optimal path according to the current instrument position and the target position; Finally, project the path onto the microscope field of view to generate the guide wire. To meet the requirements of different surgical stages, the system provides three guide wire modes: position guidance (accuracy ±0.1 mm), direction guidance (accuracy ±2°), and depth guidance (accuracy ±0.2 mm), and doctors can flexibly switch according to needs.

[0101] The augmented reality projection module 5 intuitively superimposes the registration guide wire on the surgical microscope field of view to provide real-time navigation for doctors. This module includes four functional units: the guide wire generation unit 51, the attitude recognition unit 52, the image fusion unit 53, and the visual tracking unit 54.

[0102] The guide wire generation unit 51 generates the guide wire based on the OCT and intraocular imaging data. This unit generates a visually clear and intuitive guide wire according to the optimal implantation path calculated by the registration algorithm module 4. The guide wire adopts an adaptive display strategy, dynamically adjusting the line thickness (0.1 - 0.5 mm), color (RGB mode adjustable), and transparency (20% - 80% adjustable) according to the surgical stage and the instrument position, minimizing the interference to the surgical field of view while ensuring visibility.

[0103] The attitude recognition unit 52 is responsible for recognizing the spatial attitude of the intraocular surgical instrument. This unit uses a 3D convolutional neural network to extract the instrument attitude information from multi-angle images. The network input is a continuous multi-frame image sequence, and the output is the six-degree-of-freedom attitude parameters (position coordinates x, y, z and Euler angles α, β, γ) of the instrument. This method reaches an attitude recognition accuracy of 98.5% on the standard test set, and the average processing time is 8 ms / frame.

[0104] The image fusion unit 53 superimposes the guide wire on the surgical microscope imaging. This unit uses a semi-transparent superimposition technique to clearly display the guide information while maintaining the details of the original image. The fusion algorithm can be expressed as:

[0105] I fusion (x, y) = (1 - α)·Imicroscope (x, y) + α·I guideline (x, y),

[0106] where I f usion is the fused image, I m icroscope is the original microscope image, I g uideline is the guideline image, and α is the transparency parameter (usually set to 0.6). This unit also implements edge enhancement and contrast adaptive adjustment to ensure the clear visibility of the guideline under different lighting conditions.

[0107] The visual tracking unit 54 uses the optical features of the intraocular instrument for visual tracking to obtain the registration relationship between the registration guideline and the surgical instrument. This unit is based on a combined algorithm of Kalman filtering and template matching to achieve real-time tracking of the instrument. The tracking accuracy can reach ±0.05 mm under normal surgical conditions, and the tracking frequency is 60 Hz, which can effectively handle complex situations such as rapid movement and partial occlusion of the instrument.

[0108] Through the collaborative work of the above five major modules, the adaptive optical calibration intelligent intraocular lens implantation navigation system of the present invention can achieve precise and stable surgical navigation, significantly improving the accuracy of intraocular lens implantation and the postoperative visual quality. This system is particularly suitable for complex cataract surgeries (such as high astigmatism, small pupil, iris injury, etc.) and multifocal lens implantation surgeries requiring ultra-high precision.

[0109] The intelligent intraocular imaging system 6 is the basic support system of the present invention, responsible for identifying various intraocular tissue structures and the positions of surgical instruments. This system includes four functional units: the convolutional neural network unit 61, the contour generation unit 62, the deep neural network model unit 63, and the verification unit 64.

[0110] The convolutional neural network unit 61 designs a feature extraction algorithm to identify intraocular surgical instruments, lenses, corneas, lenses, pupils, irises, aqueous humor, and other tissue structures. This unit uses an improved U-Net network architecture to achieve precise segmentation of intraocular tissues through an encoder-decoder structure. The loss function of the network adopts a weighted combination of Dice loss and cross-entropy loss:

[0111] L = α·L dice + (1 - α)·L CE ,

[0112] where L dice is the Dice loss, L CEis the cross - entropy loss, and α is the weight coefficient (set to 0.7). This network was trained on a dataset of 3000 diverse intra - ocular images, achieving an average segmentation accuracy of 97.8%, with a processing speed of 15 ms / frame, meeting the requirements of real - time recognition. For complex cases (such as patients in the late stage of cataract), the algorithm can still maintain an identification accuracy of over 92%, significantly superior to traditional image - processing methods.

[0113] The contour generation unit 62 segments the three - dimensional contour of intra - ocular tissues based on the OCT image. This unit first pre - processes the original OCT data by denoising and enhancing it, then uses a 3D convolutional neural network for voxel - level segmentation, and finally reconstructs the three - dimensional surface model through the Marching Cubes algorithm. The contour generation accuracy can reach the voxel level (typical resolution is 0.01 mm 3 ), meeting the requirements of high - precision surgical navigation. This unit also realizes real - time contour update (update frequency 5 Hz), which can adapt to the dynamic changes of the intra - ocular tissue morphology during the surgical process.

[0114] The deep neural network model unit 63 is responsible for generating the training deep neural network model. This unit adopts a modular design and dynamically combines and adjusts the network structure according to different tissue types and surgical stages. The core technologies include: few - shot learning (only 20 - 30 samples are required to adapt to new instruments or special cases), active learning (prioritizing the annotation of samples with a large amount of information to improve the training efficiency), and model distillation (transferring the knowledge of complex models to lightweight networks to improve the running speed). This unit realizes the semi - automatic update of the model. The hospital only needs to provide a small amount of new data, and the system can adaptively optimize, significantly reducing the maintenance cost.

[0115] The verification unit 64 performs key - point matching between the OCT image and the contour image segmented from the intra - ocular image to verify the model accuracy. This unit uses a robust feature - matching method based on the RANSAC algorithm to extract and match SIFT feature points from the two - modality images. During the verification process, the system sets strict accuracy thresholds (position error < 0.15 mm, angle error < 3°). Only the models that pass the verification will be applied to the actual surgical navigation. The verification unit also has a self - diagnosis function, which can detect and report potential recognition errors, ensuring that the system adopts a conservative strategy in uncertain situations to guarantee surgical safety.

[0116] The system of the present invention also includes an eyeball movement prediction module 7 and a virtual projection correction module 8, further enhancing the adaptability and accuracy of the system.

[0117] Based on the optical imaging principle and spatial registration of intraocular surgery, the eye movement prediction module 7 constructs a central coordinate model of the intraocular lens. This module applies deep learning technology to establish a mapping relationship between eye movement characteristics and changes in the lens position, achieving accurate prediction of eye movement. Preferably, this module uses a long short-term memory network (LSTM) combined with an attention mechanism to model the eye movement sequence. The algorithm can be expressed as:

[0118] h t = LSTM(x t , h t-1 , c t-1 ),

[0119]

[0120] where x t is the eye position feature at time t, h t and c t are the hidden state and cell state of the LSTM respectively, is the predicted future position. This module can predict the eye movement trajectory within the next 200 ms, with the prediction error controlled within ±0.08 mm, providing sufficient reaction time for fine motion compensation.

[0121] The virtual projection correction module 8 real-time tracks the spatial position between the intraocular surgical instrument and the intraocular lens, and designs a coordinate transformation mechanism based on optical projection. This module constructs a complete optical projection model, accurately mapping the three-dimensional space coordinates into the two-dimensional microscope field of view. Preferably, this module uses binocular stereo vision technology to obtain depth information through disparity calculation, combined with the optical projection equation, to achieve accurate conversion of the coordinate system. This technology enables the system to adapt to different models of surgical microscopes and optical systems, greatly improving the versatility and adaptability of the product.

[0122] Through the collaborative work of the eye movement prediction module 7 and the virtual projection correction module 8, the present invention establishes a complete spatial coordinate conversion chain, from the three-dimensional space inside the eye to the two-dimensional microscope field of view, and then to the augmented reality guiding line, achieving seamless visual navigation, greatly improving the surgical precision and the doctor's operation experience.

[0123] The eye movement prediction module 7 includes four functional units: the central coordinate model unit 71, the deep learning model unit 72, the estimation unit 73, and the tracking unit 74.

[0124] The central coordinate model unit 71 constructs the central coordinate model of the intraocular lens. Based on the eye anatomical structure, this unit establishes a three-dimensional coordinate system with the center of the lens capsule as the origin. Preferably, this coordinate system is determined by three key anatomical landmark points: the corneal apex, the pupil center, and the center of the anterior surface of the lens. These three points form a reference plane. Combining biological parameters such as the eye axis length and corneal curvature, the ideal implantation position of the lens is accurately calculated. This model also takes into account individual differences and is adjusted personalized through preoperative examination data (such as corneal topography, eye axis measurement, etc.) to adapt to the eye characteristics of different patients.

[0125] The deep learning model unit 72 obtains the eye movement characteristic parameters by means of a deep learning network model. This unit adopts a multi-modal fusion strategy and processes the image data from the eye tracking camera and the microscope simultaneously. Preferably, this unit uses a two-stream network architecture, one stream processes the spatial features, and the other stream processes the temporal features. Finally, the two streams of information are combined through a feature fusion layer. This network is trained on an eye movement data set (including the eye movement data of 500 patients of different types) and can accurately identify and predict three main types of eye micro-movements: micro-tremors, drifts, and micro-saccades. The network output includes key parameters such as frequency, amplitude, and direction, providing an accurate basis for subsequent compensation.

[0126] The estimation unit 73 uses a deep particle filter algorithm to estimate the displacement and attitude angle of the intraocular surgical instrument. The particle filter algorithm first generates a particle set according to the motion model, and each particle represents a possible state hypothesis; then updates the particle weights based on the observation data; finally, updates the particle distribution through resampling. Preferably, the deep particle filter of the present invention introduces a neural network to optimize particle generation and weight calculation, and the algorithm is expressed as:

[0127]

[0128] where q θ is a proposal distribution parameterized by the neural network, which can adaptively generate particles according to the historical state and the current observation. This algorithm greatly improves the stability and accuracy of tracking. Even in the case of rapid eye movement or partial occlusion of the field of view, a reliable tracking effect can still be maintained.

[0129] The tracking unit 74 tracks the OCT data and estimates the spatial position of the intraocular surgical instrument. This unit reconstructs the three-dimensional contour and position information of the instrument by real-time analyzing the OCT B-scan image sequence. Preferably, this unit adopts a sliding window strategy and processes multiple consecutive frames of OCT data simultaneously to enhance the stability of position estimation. The tracking result is updated at a frequency of 10 Hz, providing key information such as the instrument position, attitude, and movement speed, supporting surgical navigation and micro-motion compensation. This unit also realizes the function of automatic identification of the instrument type, can adapt to different types of surgical instruments, and improves the versatility of the system.

[0130] The virtual projection correction module 8 includes four functional units: a spatial position tracking unit 81, an image acquisition unit 82, a coordinate transformation unit 83, and an accuracy correction unit 84.

[0131] The spatial position tracking unit 81 tracks the spatial position between the intraocular surgical instrument and the intraocular lens in real time. This unit adopts a multi-sensor fusion strategy, comprehensively utilizes microscope images, OCT data, and special marker point information to achieve high-precision position tracking. Preferably, this unit uses the extended Kalman filter algorithm to fuse multi-source data, and the state update and observation equations are as follows:

[0132]

[0133] Preferably, this unit uses the extended Kalman filter algorithm to fuse multi-source data, and the state update and observation equations are as follows:

[0134]

[0135] P t|t =(I-K t H t )P t|t-1 ,

[0136] where is the state estimate, P is the covariance matrix, K is the Kalman gain, z is the observation value, f and h are the state transition function and the observation function respectively, F and H are the corresponding Jacobian matrices, and Q and R are the process noise and the observation noise covariance. The algorithm can control the position tracking accuracy within ±0.05 mm and the attitude angle accuracy within ±1°, meeting the requirements of precision surgery.

[0137] The image acquisition unit 82 acquires real-time intraocular images according to the position of the intraocular instrument. This unit is closely integrated with the microscope module 1 and automatically adjusts the imaging parameters (such as focus depth, field of view range, exposure settings, etc.) according to the current surgical stage and the instrument position. Preferably, this unit realizes the intelligent focusing function, automatically adjusts the focal plane position by analyzing the image sharpness distribution, and ensures clear imaging of the key area. This technology greatly reduces the operation burden of the doctor, enabling them to focus on the surgery itself without having to frequently adjust the microscope settings.

[0138] The coordinate transformation unit 83 designs a coordinate transformation mechanism based on optical projection. This unit is responsible for converting three-dimensional space coordinates into two-dimensional microscope field of view coordinates, which is the key to realizing precise guided projection. Preferably, this unit adopts a perspective projection model, combines the characteristics of the optical system, and establishes an accurate coordinate transformation relationship:

[0139]

[0140] Among them, (u, v) are pixel coordinates, (X, Y, Z) are world coordinates, K is the camera internal parameter matrix, and R and t are the rotation matrix and translation vector respectively. This unit obtains accurate parameter values through a calibration program and takes into account the non-linear distortion of the microscope optical system to ensure the projection accuracy. Experiments show that this coordinate transformation mechanism can control the projection error within 1 pixel, which is equivalent to a physical size of 0.02 mm at 4K resolution, meeting the requirements of ultra-fine surgery.

[0141] The accuracy correction unit 84 corrects the tracking accuracy through the eye movement prediction module 7 and spatial registration. This unit continuously monitors and optimizes the tracking accuracy through a closed-loop feedback mechanism. Preferably, this unit designs a self-calibration program to evaluate the current tracking accuracy through the reprojection error of specific marker points and automatically adjusts the tracking parameters. The calibration period is 500 ms, which can timely respond to factors such as light changes and tissue deformation that cause a decrease in accuracy. This unit also implements an anomaly detection function. When the tracking accuracy is lower than a preset threshold (usually 0.2 mm), the system will issue a warning and provide auxiliary positioning information to ensure the safety of the surgery.

[0142] The present invention also provides an intelligent intraocular lens implantation navigation method with adaptive optical calibration, including the complete navigation process from step S100 to S1000.

[0143] Step S100, collect real-time intraocular image data. This step obtains high-resolution intraocular images through the microscope module 1, providing basic data for subsequent processing. Preferably, timestamps and microscope parameters (such as magnification, working distance, etc.) are synchronously recorded during the collection process for subsequent processing and analysis. The image collection frequency is not less than 60 fps to ensure the capture of rapid eye micro-movements and surgical instrument movements.

[0144] Step S200, real-time evaluate the optical aberration distribution of intraocular tissues through the adaptive optical system module 2 and perform correction. This step first analyzes the intraocular image quality, identifies the type and distribution of aberrations, then calculates the optimal correction parameters, and finally performs real-time correction through the optical system. Preferably, a closed-loop control strategy is adopted during the correction process. Through continuous evaluation and adjustment, the image quality is gradually optimized until it meets the preset standards (such as contrast > 0.8, sharpness > 0.85, etc.). This step greatly improves the intraocular image quality and lays a foundation for subsequent analysis and recognition.

[0145] Step S300: Use the three-dimensional deep learning motion compensation module 3 to predict eye movements and compensate for micro-motion interference. In this step, eye movement features are first extracted from the image sequence, then the short-term movement trend is predicted through a deep learning model, and finally compensation parameters are generated to offset the micro-motion influence. Preferably, the compensation adopts a feedforward-feedback combined strategy, which not only compensates for regular movements predictively but also corrects prediction errors in real time to achieve comprehensive and effective micro-motion suppression. Experiments show that this step can reduce the micro-motion influence by more than 85%, greatly improving the surgical precision.

[0146] Step S400: Establish spatial registration between the OCT and the intraocular imaging system through the registration algorithm module 4. This step realizes the precise fusion of multi-modal data and provides a unified reference coordinate for three-dimensional navigation. Preferably, the registration process adopts a two-stage strategy of coarse registration and fine registration. First, the initial transformation is quickly determined through feature point matching, and then the transformation parameters are precisely adjusted through an iterative optimization algorithm, finally achieving a sub-millimeter registration accuracy. This step solves the problem of inaccurate multi-modal data fusion in traditional systems and provides a solid foundation for precise navigation.

[0147] Steps S500 to S800 implement the complete process of augmented reality guidance. First, a guidance line is generated based on the OCT and intraocular imaging, then the pose of the surgical instrument is identified, and the guidance line is superimposed on the microscope field of view. Finally, visual tracking is used to ensure that the guidance line and the instrument maintain the correct registration relationship. Preferably, the guidance line display adopts a "stepped" strategy, dynamically adjusting the display method according to the distance between the instrument and the target position, providing more detailed guidance at close range and more macroscopic direction indication at long range. This innovative display strategy makes the navigation process more intuitive and smooth, reducing the doctor's cognitive burden.

[0148] Step S900: Use the intelligent intraocular imaging system 6 to identify the intraocular tissue structure and the position of the instrument. In this step, through deep learning technology, the intraocular image is analyzed in real time to identify key tissue structures (such as the cornea, iris, lens capsule, etc.) and the position of the instrument. Preferably, the identification results are intuitively displayed in the augmented reality view in different colors and marking methods, helping the doctor better understand the surgical scene. This function is particularly important for patients with atypical or diseased anatomical structures.

[0149] Step S1000: According to the registration relationship between the registration guiding line and the surgical instrument, the position of the surgical instrument is adjusted in real time to complete the precise implantation of the intraocular lens. This step is the ultimate goal of the entire navigation process. Through the synergistic effect of the foregoing steps, the doctor is guided to accurately implant the intraocular lens at the predetermined position. Preferably, the system provides a real-time feedback mechanism, indicating the deviation between the current position and the target position through visual and audible cues. When the preset accuracy threshold is reached (usually a central offset <0.1 mm and a tilt angle <2°), the system gives a confirmation signal, indicating that the implantation position meets the requirements. This step significantly improves the accuracy of intraocular lens implantation, especially for patients with high astigmatism and multifocal intraocular lens implantation surgery, and the effect is particularly significant.

[0150] Example 1

[0151] The system of the present invention was applied in the surgery of a cataract patient with high corneal astigmatism (4.5 D). Before the surgery, the axial length of the patient's eye was measured by a wavefront aberrometer to be 24.8 mm, the corneal curvature K1 = 42.5 D, and K2 = 47.0 D. The system automatically calculated the optimal intraocular lens power to be +21.5 D, and the astigmatism correction axis was 85°. During the surgery, the system corrected the optical aberration in real time, compensated for the micro-movement of the eyeball (with an amplitude of up to 120 μm), and indicated the optimal implantation position and direction through the augmented reality guiding line. Finally, the deviation of the intraocular lens implantation position was 0.08 mm, and the axial deviation was 1.8°, both of which were better than the traditional surgical standards (0.3 mm and 5° respectively). During the 3-month follow-up after the surgery, the patient's uncorrected visual acuity reached 1.0, the astigmatism decreased to 0.75 D, and the visual quality satisfaction score was 9.2 (out of 10).

[0152] Example 2

[0153] The system of the present invention was applied in the surgery of a complex cataract patient with a small pupil (diameter <4 mm) and iris damage. Due to the small and irregular pupil, it is difficult to ensure the accuracy of intraocular lens implantation in traditional surgery. This system accurately identifies the edge of the iris defect and the position of the lens capsule through the intelligent intraocular imaging system 6, evaluates and corrects complex optical aberrations in real time, and indicates the safe operation path and the ideal implantation position through the augmented reality guiding line. The total operation time was only 17 minutes, which was 35% shorter than the average time of similar surgeries in the hospital. The intraocular lens was successfully implanted in the center, and there were no complications after the surgery, and the patient's vision recovered well.

[0154] The above embodiments show that the adaptive optical calibration intelligent intraocular lens implantation navigation system and method of the present invention perform excellently in both complex and conventional cataract surgeries, can significantly improve the surgical accuracy, shorten the surgical time, and improve the postoperative visual quality, and have broad clinical application prospects.

[0155] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An adaptive optically calibrated intelligent intraocular lens implantation navigation system, characterized in that: include: A microscope module for collecting real-time intraocular image data; An adaptive optical system module, connected to the microscope module by signal, for real-time evaluation of the optical aberration distribution of intraocular tissue and correction; A 3D deep learning motion compensation module to predict eye movements and compensate for micro-motion disturbances; A registration algorithm module, used to establish spatial registration between OCT and intraocular imaging systems; an augmented reality projection module for superimposing registration guide lines on surgical microscope imaging; Intelligent intraocular imaging system, used to identify intraocular tissue structure and instrument position.

2. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: The adaptive optics system module comprises: An optical aberration evaluation unit, used to train a virtual optical imaging module of intraocular images based on image data, and to evaluate the distribution of optical aberrations of intraocular tissues in real time; a lens control unit for generating a control strategy for an intraocular lens contour based on the optical aberration evaluation; The imaging quality adjustment unit is used to adjust the imaging quality and the shape of the artificial lens in real time.

3. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: The three-dimensional deep learning motion compensation module includes: Feature extraction network unit, used to train the feature extraction network of cornea / lens / iris / pupil; A motion characteristic parameter unit, used for obtaining eye movement characteristic parameters; A deep particle filter unit for estimating the displacement and attitude angle of intraocular surgical instruments; The compensation matrix generating unit is used to generate a compensation matrix based on the virtual projection correction module.

4. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: The registration algorithm module includes: A spatial mapping unit, used to establish a spatial image mapping relationship between the intraocular OCT and the intraocular imaging system; The guide line calculation unit is used to calculate the guide line based on the spatial position, three-dimensional posture and image mapping relationship between the intraocular instrument / lens and the optical channels.

5. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: The augmented reality projection module comprises: A guide wire generation unit, used for generating a guide wire based on OCT and intraocular imaging; A posture recognition unit, used to recognize the spatial posture of intraocular surgical instruments; An image fusion unit for superimposing registration guide lines on the surgical microscope imaging; The visual tracking unit is used to perform visual tracking using the optical features of the intraocular instrument to obtain the registration relationship between the registration guide line and the surgical instrument.

6. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: The intelligent intraocular imaging system comprises: Convolutional neural network unit, used to design feature extraction algorithms to identify intraocular surgical instruments, lens, cornea, crystalline lens, pupil, iris, and aqueous humor; A contour generation unit, used for segmenting the three-dimensional contour of intraocular tissue according to the OCT image; A deep neural network model unit, used to generate and train a deep neural network model; The verification unit is used to match the key points of the contour image segmented from the OCT image and the intraocular image to verify the accuracy of the model.

7. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 1, characterized in that: Also includes: Eye movement prediction module, used to construct the central coordinate model of the intraocular lens based on the optical imaging principle and spatial registration of intraocular surgery; The virtual projection correction module is used to track the spatial position between intraocular surgical instruments and artificial lenses in real time, and to design a coordinate transformation mechanism based on optical projection.

8. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 7, characterized in that: The eye movement prediction module comprises: A central coordinate model unit, used for constructing a central coordinate model of an intraocular lens; A deep learning model unit, used for obtaining eye movement feature parameters with the help of a deep learning network model; An estimation unit, used for estimating the displacement and posture angle of the intraocular surgical instrument using a deep particle filter algorithm; The tracking unit is used to track OCT data and estimate the spatial position of intraocular surgical instruments.

9. The adaptive optical calibration intelligent intraocular lens implantation navigation system according to claim 7, characterized in that: The virtual projection correction module comprises: A spatial position tracking unit, used for real-time tracking of the spatial position between the intraocular surgical instrument and the intraocular lens; An image acquisition unit, used for acquiring real-time intraocular images according to the position of the intraocular instrument; A coordinate transformation unit, used to design a coordinate transformation mechanism based on optical projection; The accuracy correction unit is used to correct the tracking accuracy through the eye movement prediction module and spatial registration.

10. An adaptive optically calibrated intelligent intraocular lens implantation navigation method, using the method according to any one of claims 1 to 9, characterized in that: The following steps are involved: S100, collecting real-time intraocular image data; S200, real-time evaluation and correction of intraocular tissue optical aberration distribution through adaptive optical system; S300, uses a 3D deep learning motion compensation module to predict eye movements and compensate for micro-motion disturbances; S400, establishing spatial registration between the OCT and the intraocular imaging system through a registration algorithm module; S500, generates guide wires based on OCT and intraocular imaging; S600, identifying the spatial posture of intraocular surgical instruments; S700, superimposition of registration guide lines on surgical microscope imaging; S800, uses the optical features of intraocular instruments for visual tracking to obtain the registration relationship between the registration guide line and the surgical instrument; S900, which uses an intelligent intraocular imaging system to identify intraocular tissue structures and instrument locations; S1000 adjusts the position of surgical instruments in real time according to the registration relationship between the registration guide line and the surgical instrument to complete the precise implantation of the intraocular lens.

Images