Liver operation navigation system based on fluorescence and three-dimensional reconstruction and storage medium

Abstract

The invention belongs to the technical field of surgical navigation, and particularly relates to a liver surgical navigation system based on fluorescence and three-dimensional reconstruction and a storage medium. The system comprises: an input module configured to input an endoscopic image collected during an operation and a preoperative three-dimensional reconstruction model; the segmentation module is configured to segment a liver region in the endoscope picture; a fluorescence segmentation module configured to segment a fluorescence region from the liver region; the three-dimensional reconstruction module is configured to perform intraoperative three-dimensional reconstruction to obtain a liver three-dimensional reconstruction model, and the liver three-dimensional reconstruction model comprises a part corresponding to the fluorescent region; and the point cloud registration module is configured to register the liver three-dimensional reconstruction model obtained by the three-dimensional reconstruction module and the liver area in the preoperative three-dimensional reconstruction model. The liver surgery navigation system can provide more objective, accurate and easily recognized navigation information, can better achieve the target of precise medical treatment, and has a good application prospect.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of surgical navigation, and particularly relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction and a storage medium. BACKGROUND

[0002] Laparoscopic hepatectomy is a minimally invasive surgical procedure that offers significant advantages over traditional open surgery, such as less trauma and faster recovery. During the procedure, the surgeon makes several small incisions in the patient's abdomen and inserts a laparoscope and surgical instruments through these incisions. The laparoscope transmits real-time images of the abdominal cavity to a high-definition display screen, allowing the surgeon to accurately operate the instruments based on the images to remove the liver. This procedure requires high technical requirements and professional team cooperation. Its indications include some benign liver tumors and early-stage liver cancer, but whether it is suitable needs to be evaluated in detail. Patients generally recover quickly after surgery, with shorter hospital stays and less impact on quality of life.

[0003] In laparoscopic hepatectomy, the application of surgical navigation technology greatly improves the accuracy and safety of the procedure. In the prior art, common surgical navigation methods include augmented reality (AR) navigation, image-guided surgery (IGS) systems, real-time image registration technology, and indocyanine green (ICG) fluorescence staining technology. Among them, real-time image registration technology precisely aligns the preoperative three-dimensional reconstruction model with the intraoperative laparoscopic image, providing more accurate navigation information. Indocyanine green (ICG) fluorescence staining technology can be used during surgery to locate the tumor boundary in real time and guide precise resection. This technology is particularly suitable for complex cases, such as tumors located in the center of the liver or adjacent to important blood vessels.

[0004] In the prior art, real-time image registration technology and indocyanine green (ICG) fluorescence staining technology are two different navigation technologies. If both technologies are used simultaneously during surgery, they will provide surgical navigation reference information to the surgeon performing the surgery in different ways (three-dimensional model and two-dimensional image). The surgeon then needs to integrate the information based on experience and self-awareness, which has the disadvantage of being too subjective. Therefore, there is an urgent need in the art to develop new navigation methods and systems that can integrate real-time image registration technology and indocyanine green (ICG) fluorescence staining technology to provide unified and accurate navigation information, making the surgeon's decision-making during surgery more objective and accurate. SUMMARY

[0005] To solve the problems in the prior art, the present application provides a liver surgery navigation system based on fluorescence and three-dimensional reconstruction and a storage medium.

[0006] A liver surgery navigation system based on fluorescence and three-dimensional reconstruction, comprising:

[0007] The input module is configured to input an intraoperative endoscopic image;

[0008] The segmentation module is configured to segment a liver region in the endoscopic image by a segmentation model;

[0009] The fluorescence segmentation module is configured to segment a fluorescence region from the liver region by a fluorescence region segmentation algorithm;

[0010] The three-dimensional reconstruction module is configured to perform intraoperative three-dimensional reconstruction on images in the liver region by a three-dimensional reconstruction model to obtain a liver three-dimensional reconstruction model, wherein the liver three-dimensional reconstruction model comprises a part corresponding to the fluorescence region;

[0011] The point cloud registration module is configured to register the liver region in the liver three-dimensional reconstruction model obtained by the three-dimensional reconstruction module and a preoperative three-dimensional reconstruction model by a point cloud registration model.

[0012] Preferably, the endoscopic image is collected by a monocular camera in an endoscope.

[0013] Preferably, training image data for training the three-dimensional reconstruction model comprises 2D-3D registration data and 2D video under an endoscope, and the 2D-3D registration data is collected by using a depth camera to collect internal organs of an animal in an animal experiment, and retaining 2D RGB images and corresponding point cloud information.

[0014] Preferably, in the segmentation module, the algorithm of the segmentation model is selected from DeepLab, U-Net, AttentionU-Net, U-Net++, TransUNet, CPAD-Net, TD-Net, MS-FANet or MCT-Net.

[0015] Preferably, in the fluorescence segmentation module, the fluorescence region segmentation algorithm comprises the following steps:

[0016] Step 1: calculate a feature map according to the following formula:

[0017] gr gray =2I G -I R -I B

[0018] wherein I R , I G , I B represent RGB channel values of an image respectively, and gr gray represents a feature map calculated by image color;

[0019] Step 2, the optimal threshold value is determined by maximizing the variance value of two types of pixels after image segmentation by Otsu algorithm, so that the image is binarized, and a segmentation graph gr is obtained, and the expression is:

[0020] gr = Otsu((gr gray -min(gr gray )) / (max(gr gray )-min(gr gray )))

[0021] Wherein, Otsu represents Otsu algorithm;

[0022] Step 3, remove the noise area in the segmentation graph gr by opening operation;

[0023] Step 4, calculate the maximum connected domain to obtain the fluorescent region.

[0024] Preferably, in the three-dimensional reconstruction module, the algorithm of the three-dimensional reconstruction model is selected from SLAM3R, ORB-SLAM, LSD-SLAM, DVO-SLAM, SVO, VINS-Mono, DeepSLAM and NeuralRecon.

[0025] Preferably, in the point cloud registration module, the algorithm of the point cloud registration model is selected from

[0026] SACF-Net, ICP, RANSAC, Go-ICP, PointNetLK, DCP, PRNet and 3DMatch.

[0027] Preferably, it further comprises:

[0028] The visualization module is configured to render and display the corresponding part of the fluorescent region in the preoperative three-dimensional reconstruction model.

[0029] Preferably, the visualization module is further configured to visualize the direction of the current endoscopic image relative to the liver in the preoperative three-dimensional reconstruction model.

[0030] The application also provides a computer readable storage medium, which stores a computer program for implementing the above-mentioned liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

[0031] The application constructs a new liver surgery navigation system, which achieves the following beneficial technical effects:

[0032] 1. The liver surgery navigation system can provide more objective, accurate and easy to identify navigation information, and a new method for training data acquisition of intracavity monocular three-dimensional reconstruction model is provided.

[0033] 2. The liver surgery navigation system can provide more objective, accurate and easy to identify navigation information, and a new method for training data acquisition of intracavity monocular three-dimensional reconstruction model is provided.

[0034] 3. The most difficult thing in monocular three-dimensional reconstruction of intracavity pictures under minimally invasive surgery is data source. The three-dimensional reconstruction technology in the prior art is carried out in a natural scene, and intracavity three-dimensional data cannot be collected as point cloud data using rich depth cameras in a natural scene. Therefore, a new idea for collecting training data is provided in the preferred embodiment of the present application. The depth camera is used to take pictures at a close distance in animal experiments, and pictures similar to those collected by a laparoscope are collected. The monocular three-dimensional reconstruction model is modeled using these data, so that the three-dimensional reconstruction model constructed can accurately reconstruct a three-dimensional model under a laparoscope.

[0035] In summary, the liver surgery navigation system can provide more objective, accurate and easy to identify navigation information, and a new method for training data acquisition of intracavity monocular three-dimensional reconstruction model is provided. Thus, precise medical treatment is better achieved, and the application prospect is good.

[0036] Obviously, according to the above content of the present application, according to the ordinary technical knowledge and common means in the art, other various forms of modifications, replacements or changes can be made without departing from the above basic technical idea of the present application.

[0037] The above content of the present application is further described in detail through the specific embodiments below. However, it should not be understood that the above subject matter of the present application is limited to the following examples. Any technology realized based on the above content of the present application belongs to the scope of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The result example picture of the liver region segmented by the segmentation module in Example 1 is shown in the following figure;

[0039] Figure 2 The result example picture of the fluorescent region segmented by the fluorescent segmentation module in Example 1 is shown in the following figure;

[0040] Figure 3 The result example picture of the reconstructed point cloud obtained by the three-dimensional reconstruction module in Example 1 is shown in the following figure;

[0041] Figure 4 An example diagram of the result of registering two point clouds by the point cloud registration module in embodiment 1;

[0042] Figure 5 An example diagram of the result of rendering the fluorescent region into the preoperative three-dimensional reconstruction model by the visualization module in embodiment 1. DETAILED DESCRIPTION

[0043] It should be particularly pointed out that the algorithms of data acquisition, transmission, storage and processing steps not specifically explained in the embodiments, and the hardware structure, circuit connection, etc. not specifically explained can be realized through the contents disclosed by the prior art.

[0044] Embodiment 1: Liver surgery navigation system based on fluorescence and three-dimensional reconstruction

[0045] The system of the embodiment comprises:

[0046] An input module configured to input an intraoperative endoscope picture;

[0047] A segmentation module configured to segment a liver region in the endoscope picture through a segmentation model;

[0048] A fluorescence segmentation module configured to segment a fluorescent region from the liver region through a fluorescent region segmentation algorithm;

[0049] A three-dimensional reconstruction module configured to perform intraoperative three-dimensional reconstruction on images in the liver region through a three-dimensional reconstruction model to obtain a liver three-dimensional reconstruction model, wherein the liver three-dimensional reconstruction model contains a portion corresponding to the fluorescent region;

[0050] A point cloud registration module configured to register the liver region in the liver three-dimensional reconstruction model obtained by the three-dimensional reconstruction module and a preoperative three-dimensional reconstruction model through a point cloud registration model.

[0051] A visualization module configured to render and display the portion corresponding to the fluorescent region in the preoperative three-dimensional reconstruction model, and visualize the direction of the current endoscope picture relative to the liver in the preoperative three-dimensional reconstruction model.

[0052] In the embodiment, 2D-3D registration data and 2D videos under the endoscope are collected respectively for model training. The 2D-3D registration data uses a depth camera to collect point cloud data of the internal organs of animals such as monkeys, pigs and rabbits in animal experiments, and retains the 2D RGB images and corresponding point cloud information for use as modeling data later. The 2D videos under the endoscope are directly recorded using an endoscope device, and the frames of the pictures containing the liver are extracted using python, and the liver region is annotated for semantic segmentation using the labelme annotation software.

[0053] The algorithm of the segmentation model is selected from DeepLab, U-Net, Attention U-Net, U-Net++, TransUNet, CPAD-Net, TD-Net, MS-FANet or MCT-Net.

[0054] As a preferred manner, the embodiment uses DeepLab V3+ algorithm to model the liver segmentation model. The algorithm selects to use the dilated convolution network on the selection of convolution, controls the feature resolution calculated by the deep convolutional neural network, and adjusts the field of view of the filter to capture multi-scale information:

[0055]

[0056] where y i is the data of each position i on the output feature map, r is the dilated convolution expansion rate, k is the convolution kernel size, x i+rk is the value of the upper layer feature map at the i+rk position, w k is the network parameter.

[0057] The deep separable convolution is used in the convolution calculation, which reduces the computational complexity by decomposing the standard convolution into depth convolution and pointwise convolution. Specifically, the depth convolution performs spatial convolution independently on each input channel, while the pointwise convolution is used to combine the output of the depth convolution. Finally, combined with the dilated convolution and the deep separable convolution, the encoder-decoder structure is adopted in the overall architecture of the network, Xception is used as the backbone network, and the liver segmentation model is obtained using the liver segmentation data. In the embodiment, an example of the result obtained by using the algorithm is shown in Figure 1 .

[0058] The fluorescent region segmentation algorithm in the fluorescent segmentation module includes the following steps:

[0059] Step 1, calculate the feature map according to the following formula:

[0060] gr gray = 2I G -I R -I B

[0061] where I R , I G , I B represent the numerical values of the RGB channels of the image respectively, gr gray represents the feature map calculated by the image color;

[0062] Step 2, determine the optimal threshold value by maximizing the variance value of the two types of pixels after image segmentation by Otsu algorithm, so as to binarize the image and obtain the segmentation graph gr, the expression is:

[0063] gr = Otsu((gr gray -min(gr gray )) / (max(gr gray )-min(gr gray ))

[0064] where Otsu denotes the Otsu algorithm.

[0065] Step 3, the subsequent use of open operation to remove noise area. That is, first replace the pixel value in the region covered by the structural element with the minimum value in the region, and the erosion operation will make the object boundary in the image shrink inward; then replace the pixel value in the region covered by the structural element with the maximum value in the region. The expansion operation will make the object boundary in the image expand outward. The calculation formula of open operation is represented as:

[0066] Open(gr) = Dilation(Erosion(gr))

[0067] where Dilation denotes the expansion operation, and Erosion denotes the erosion operation.

[0068] Step 4, calculate the maximum connected domain to obtain the subsequent use of fluorescent region.

[0069] The results obtained by this method are shown in the following examples. Figure 2

[0070] In the three-dimensional reconstruction module, the algorithm of the three-dimensional reconstruction model is selected from SLAM3R, ORB-SLAM, LSD-SLAM, DVO-SLAM, SVO, VINS-Mono, DeepSLAM and NeuralRecon.

[0071] As a preferred scheme, in the embodiment, a monocular three-dimensional reconstruction model is constructed using the SLAM3R algorithm architecture. The algorithm consists of two components: an image-to-point (I2P) network that recovers local 3D points from video clips; and a local-to-world (L2W) network that local reconstruction into the global scene coordinate system.

[0072] The I2P model uses ViT as its backbone. It includes a shared encoding E img , two independent decoders D key and D sup , and a point regression head for final prediction. For a given video clip, the image encoder encodes each frame of data into Token representation. The encoder contains m ViT encoder modules, each block containing self-attention and feedforward layers:

[0073] ​

[0074] where, denotes i-frame data encoding Token, T denotes Token quantity, d denotes Token latitude, denotes each frame data, H, W respectively denote the height and width of the image, i denotes the number of frames in the video. Each frame is independently processed in parallel, and the output is divided into F key keyframe features and auxiliary frame features.

[0075] keyframe decoder D key consists of n ViT decoder blocks, each block containing self-attention, cross-attention and feed-forward layers. Multi-view cross-attention is used to combine information from different support frames. Given a feature map F key and keyframe decoder D key takes F key as input for self-attention and performs cross-attention between F key and . Then a max-pooling layer is used to aggregate features after cross-attention to get the decoded keyframe Token G key :

[0076] G key = (F key , F sup1 , F sup2 ,... F supL-1 )

[0077] Finally, a linear head is used to regress dense 3D point clouds in the uniform coordinate system from the decoded Token. In addition to the point cloud, this embodiment also predicts the confidence map of all frames to evaluate their reliability:

[0078] Y H×W×3 , C H×W×1 = H(G i )

[0079] where Y H×W×3 denotes the three-dimensional space coordinates corresponding to each pixel of the image, C H×W×1 denotes the reliability of each pixel of the image, H(G i ) is a linear head, and G i denotes the feature map after cross-attention of the i-th image.

[0080] The 3D point cloud reconstructed by the I2P model is encoded into the L2W model, which is processed using a similar method to image patching in the ViT encoder:

[0081] P i = E pts (Y H×W×3)

[0082] where P i is the encoding result of each point cloud, E pts is the encoder of the L2W model. The registration decoder D reg takes the feature markers P i as input, aiming to convert the local reconstruction of the key frame into the scene coordinate system, and the combination process filters out the unreliable part of the point cloud. The decoding process is represented as:

[0083] g key = D reg (P key , P sup1 , P sup2 , … P supL-1 )

[0084] where g key is the coordinate information of the point cloud, P key and P supi are the encoding results of the key frame point cloud and other auxiliary frame point cloud, respectively.

[0085] Through the above algorithm, an example of the reconstructed point cloud result is shown in FIG. 2. Figure 3

[0086] In the point cloud registration module, the algorithm of the point cloud registration model is selected from SACF-Net, ICP, RANSAC, Go-ICP, PointNetLK, DCP, PRNet, and 3DMatch.

[0087] As a preferred scheme, in the present embodiment, the SACF-Net algorithm is used to register the preoperative three-dimensional reconstruction point cloud and the intraoperative three-dimensional reconstruction point cloud. The algorithm is an encoder-decoder network. A skip attention connection is used between the encoder and the decoder to connect the point features in the corresponding resolution of the encoder and the decoder.

[0088] Given two unaligned point clouds P and Q as input, the shared encoder of SACF-Net produces two initial correspondence matrices and extracts the interaction features to the decoder. These correspondence matrices are composed of L attention matrices and which are obtained in L layers of the encoder, each layer being composed of a feature extraction module and a feature interaction module.

[0089] ​The encoder then extracts interactive features from different resolution levels, and the decoder generates point features in the same way. Here, skip attention is used to establish inter-layer connections between the interactive features extracted by the encoder and the point features generated in the decoder. The decoder skip attention network is designed to provide a connection for each matching point pair. and Output confidence score w pj and w qj The decoder uses skip attention at each resolution level to convey features from the encoder at the same level. Represented as:

[0090]

[0091] Where, p i q j These are the points in point clouds P and Q that can be mapped. In the transformation estimation, the weighted Procrustes method is used with weights w. pj Estimate the rotation and translation from P to Q. The translation and rotation matrices are represented as:

[0092]

[0093] Where Π· represents the hard elimination process, R est t est Let R represent the translation and rotation matrix. pi +t represents p i The result after the points have been transformed. This allows us to register the two point cloud positions using the final translation and rotation matrix. An example of the result is shown below. Figure 4 As shown.

[0094] Finally, the preoperative 3D reconstruction model is rendered directly, and an example of the result is shown below. Figure 5 As shown, the green highlighted area represents the anatomical region.

[0095] As can be seen from the above embodiments, this invention combines image registration technology and indocyanine green fluorescent staining technology, enabling the simultaneous provision of the preoperative 3D reconstruction model and fluorescent staining region information within the preoperative 3D reconstruction model in the navigation image. This provides surgeons with more objective, accurate, and easily identifiable information, thus better achieving precision medicine. Therefore, this invention has excellent application prospects in surgical navigation for laparoscopic liver resection.

Claims

1. A liver surgery navigation system based on fluorescence and three-dimensional reconstruction, characterized by, The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

2. The liver surgery navigation system based on fluorescence and three-dimensional reconstruction according to claim 1, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

3. The liver surgery navigation system based on fluorescence and three-dimensional reconstruction according to claim 1 or 2, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

4. The liver surgery navigation system based on fluorescence and three-dimensional reconstruction according to claim 1, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

5. The fluorescence and three-dimensional reconstruction based navigation system for liver surgery according to claim 1, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. gr gray = 2I G -I R -I B wherein I R , I G , I B respectively represent the image RGB channel values, gr gray represents the feature map calculated by image color. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. gr = Otsu(gr gray -min(gr gray )) / (max(gr gray )-min(gr gray ))) The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

6. The fluorescence and three-dimensional reconstruction based navigation system for liver surgery according to claim 1, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

7. The fluorescence and three-dimensional reconstruction based navigation system for liver surgery according to claim 1, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

8. The fluorescence and three-dimensional reconstruction based navigation system for liver surgery according to claim 1, characterized in that, The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

9. The fluorescence and three-dimensional reconstruction based navigation system for liver surgery according to claim 8, characterized in that: The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction.

10. A computer-readable storage medium, characterized in that, The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation system based on fluorescence and three-dimensional reconstruction. The application relates to a liver surgery navigation

Images