A lung vessel enhancement method based on dual-mode imaging

Abstract

The present application relates to medical image processing technical field, more specifically, it relates to a kind of lung blood vessel enhancement method based on dual-mode imaging, the fluorescence image of lung blood vessel is obtained by fluorescence visualization;The infrared image of lung blood vessel is obtained by infrared visualization;Based on the intensity information of fluorescence image and infrared image and the morphological information of blood vessel, the enhancement weight of two modes is calculated, and the pseudo-color blood vessel image of adaptive enhancement is obtained;Pseudo-color blood vessel image is carried out multi-scale blood vessel enhancement filtering processing, and the blood vessel enhancement result is obtained;Blood vessel enhancement result and fluorescence image and infrared image are fused with weight, and the final blood vessel enhancement result is obtained;The final blood vessel enhancement result is output, the advantages of fluorescence imaging and infrared imaging are combined skillfully, and the comprehensive capture of different depth and different scale blood vessels is realized.The high sensitivity of fluorescence imaging technology ensures the accurate display of tiny blood vessels, and infrared imaging provides the blood vessel structure information of deep tissue.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and more specifically, to a method for enhancing pulmonary vessels based on dual-modal imaging. Background Technology

[0002] With the continuous development of medical imaging technology, the imaging and analysis of pulmonary vessels are playing an increasingly important role in the diagnosis of lung diseases. Traditional pulmonary vascular imaging methods mainly rely on single modalities, such as X-ray angiography or CT angiography. While these methods can provide some information on vascular structure, they have significant limitations in displaying small vessels and complex vascular networks.

[0003] In recent years, researchers have begun to explore the use of multimodal imaging techniques to improve the imaging quality of pulmonary vessels. Among these, the bimodal approach combining fluorescence and infrared imaging has attracted widespread attention. Fluorescence imaging provides highly sensitive vascular information, particularly advantageous for displaying small vessels; while infrared imaging can provide information on the vascular structure of deep tissues. However, current bimodal imaging methods still face some challenges.

[0004] First, existing bimodal imaging methods often simply superimpose or fuse images from two modalities, lacking in-depth analysis and optimization of the characteristics of different modalities. This results in the final fusion result failing to fully leverage the advantages of both modalities and may even introduce new image artifacts.

[0005] Secondly, in terms of vascular enhancement processing, existing methods mostly employ filtering algorithms with fixed parameters, which are difficult to adapt to vascular structures of different scales and morphologies. This one-size-fits-all approach often results in some vascular structures being over-enhanced or under-enhanced, affecting the accuracy of vascular identification.

[0006] Furthermore, most existing methods for blood vessel recognition and segmentation rely on traditional image processing algorithms, which perform poorly when dealing with complex backgrounds and low-contrast blood vessels. Although some research has begun to explore the introduction of deep learning technology, how to effectively combine traditional image processing methods with deep learning technology remains a problem that urgently needs to be solved.

[0007] Finally, existing methods often struggle to address blood vessels of varying sizes and depths. For example, some methods may perform well in enhancing major blood vessels but are less effective at treating small vessels; or they may be highly effective at treating superficial vessels but have limited enhancement effects on deep vessels.

[0008] In view of the above problems, there is an urgent need for a method that can fully utilize the advantages of dual-modal imaging to achieve adaptive vascular enhancement and effectively handle blood vessels of different scales and depths, so as to improve the quality and accuracy of pulmonary vascular imaging. Summary of the Invention

[0009] This invention provides a method for enhancing pulmonary vessels based on dual-modal imaging, aiming to solve the aforementioned problems in existing technologies. This invention innovatively combines fluorescence imaging and infrared imaging technologies, and introduces adaptive weight calculation and multi-scale enhancement techniques, achieving high-quality imaging and enhancement of pulmonary vessels.

[0010] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0011] A method for enhancing pulmonary vascularity based on dual-modal imaging, comprising:

[0012] The acquisition steps include:

[0013] Fluorescent images of pulmonary blood vessels were obtained through fluorescence imaging.

[0014] Infrared images of pulmonary blood vessels are obtained through infrared imaging.

[0015] The processing steps include:

[0016] Based on the intensity information of the fluorescence image and the infrared image, as well as the morphological information of the blood vessels, the enhancement weights of the two modalities are calculated to obtain an adaptively enhanced pseudo-color blood vessel image.

[0017] The pseudo-color blood vessel image is subjected to multi-scale blood vessel enhancement filtering to obtain the blood vessel enhancement result;

[0018] The vascular enhancement results are weighted and fused with the fluorescence image and the infrared image to obtain the final vascular enhancement results;

[0019] Output steps, including:

[0020] Output the final vascular enhancement result.

[0021] Preferably, the acquisition step specifically includes:

[0022] Fluorescent nanoparticles were injected intravenously, and fluorescence signals from pulmonary blood vessels were collected under laser excitation to obtain the fluorescence image.

[0023] The infrared image was acquired using laser imaging technology.

[0024] Preferably, the process of calculating the enhancement weights in the processing steps specifically includes:

[0025] The fluorescence image and the infrared image are respectively subjected to grayscale stretching to obtain the enhanced image;

[0026] The enhanced image is subjected to pseudo-color mapping to obtain a pseudo-color blood vessel image;

[0027] Intensity enhancement results are obtained based on the intensity information of the pseudo-color blood vessel image;

[0028] Morphological enhancement results were obtained based on vascular morphology and branching information;

[0029] The enhancement weight for each mode is calculated based on the intensity enhancement results and the morphology enhancement results.

[0030] Preferably, the multi-scale vascular enhancement filtering process specifically includes:

[0031] A multi-scale separation algorithm is adopted, which uses a series of Gaussian kernels with different radii to perform convolution calculations to obtain a multi-scale image sequence;

[0032] The Laplacian pyramid algorithm is used to perform multi-layer filtering on the pseudo-color blood vessel image, decomposing it into multiple groups of images to be enhanced;

[0033] Calculate the channel similarity and scale similarity between the images to be enhanced;

[0034] Based on the channel similarity and scale similarity, the fusion weight of each image to be enhanced is calculated;

[0035] The enhanced image is reconstructed using a weighted summation method.

[0036] Preferably, the following steps are also included:

[0037] The enhanced image is binarized using a segmentation threshold to extract blood vessels and their center lines.

[0038] Calculate the feature information of blood vessel contours;

[0039] The blood vessel course is calculated based on the feature information of the blood vessel contour.

[0040] The vessel centerline is corrected based on the vessel orientation to obtain the final vascular enhancement result.

[0041] Preferably, the following steps are also included:

[0042] A vascular cognition module is designed to perform deep learning on the final vascular enhancement result and extract the boundary features of each type of blood vessel in the image.

[0043] Based on the aforementioned boundary features, vascular fusion is performed to obtain the original vascular fusion image;

[0044] The original vascular fusion image is corrected by vascular fusion to obtain a corrected vascular enhancement image.

[0045] Preferably, the design of the vascular recognition module specifically includes:

[0046] The brightness gradient map of the enhanced blood vessel image is filtered to obtain the salient region corresponding to the foreground region;

[0047] The brightness gradient map of the image with obvious vascular enhancement is filtered to obtain the vascular boundary region;

[0048] The significant region and the blood vessel boundary region are synthesized to obtain the enhanced blood vessel contour region;

[0049] Using the vascular structure of the pre-enhanced image, the vascular boundary points in the vascular contour region are corrected;

[0050] Dijkstra's algorithm was used to smooth the vascular contour curves in the corrected vascular contour region.

[0051] The smoothed blood vessel contour curve is projected onto the enhanced image to complete the blood vessel recognition process.

[0052] Preferably, the following steps are also included:

[0053] For the final enhanced blood vessel result, the corresponding region is found in the original image;

[0054] In the corresponding original image, the differences between the enhanced result and the blood vessels in the initial vascular fusion image are marked to obtain the difference rate;

[0055] If the difference rate is less than a preset threshold, the algorithm is considered successful.

[0056] If the difference rate is greater than or equal to a preset threshold, the acquisition step is returned and the method is re-executed.

[0057] Preferably, the preset threshold is 10%.

[0058] Preferably, the following steps are also included:

[0059] The trained lung vessel recognition model is used to identify the final enhanced vessel result to obtain the vessel classification result;

[0060] The training method for the pulmonary vascular recognition model includes:

[0061] Realistic blood vessel enhancement results were obtained from bimodal images and used as training data.

[0062] The training data is used as input to a deep learning framework, and the blood vessel recognition model is trained using the cross-entropy loss function and optimization methods.

[0063] The method of the present invention has the following significant technical effects:

[0064] First, this invention cleverly combines the advantages of fluorescence imaging and infrared imaging to achieve comprehensive capture of blood vessels at different depths and scales. The high sensitivity of fluorescence imaging technology ensures accurate display of tiny blood vessels, while infrared imaging provides information on the vascular structure of deep tissues. This dual-modal combination not only expands the depth range of vascular imaging but also improves the ability to identify complex vascular networks.

[0065] Secondly, this invention introduces an innovative adaptive weight calculation algorithm. This algorithm can automatically adjust the enhancement strategy according to the characteristics of different images, effectively overcoming the limitations of traditional fixed-parameter methods. Through comprehensive analysis of image intensity information and vascular morphological features, this method can intelligently determine the enhancement weight of each modality, thereby achieving precise enhancement of different types of blood vessels. This adaptability greatly improves the robustness of the method, enabling it to adapt to images from different patients and under different imaging conditions.

[0066] Furthermore, this invention employs a multi-scale vascular enhancement filtering technique. This technique can simultaneously process vascular structures of different scales, effectively resolving the contradictions of traditional methods when handling blood vessels of varying sizes. By using a series of Gaussian kernels with different radii for convolution calculations, this method can effectively enhance major vascular structures while preserving the details of small blood vessels, achieving balanced enhancement of the entire vascular spectrum.

[0067] Furthermore, this invention ingeniously combines traditional image processing techniques with deep learning methods. By introducing a deep learning-based vascular recognition module, this method not only improves the accuracy of vascular identification but also enhances its ability to handle complex backgrounds and low-contrast vessels. This organic combination of traditional methods and artificial intelligence technology opens up new research directions for pulmonary vascular enhancement.

[0068] Finally, this invention also designs a complete evaluation and optimization mechanism. By calculating the difference rate between the enhanced result and the original image and setting a reasonable threshold, this method can automatically determine whether the enhancement effect meets the standard and reprocess it if necessary. This closed-loop optimization mechanism ensures the high quality and reliability of the final output result.

[0069] In summary, the pulmonary vascular enhancement method based on dual-modal imaging provided by this invention effectively solves the problems existing in the prior art by innovatively combining multiple advanced technologies. This method not only significantly improves the quality and accuracy of pulmonary vascular imaging but also possesses strong adaptability and robustness. These advantages make this method promising for early diagnosis, surgical planning, and efficacy evaluation of pulmonary diseases, providing important technical support for improving the diagnosis and treatment of pulmonary diseases. Attached Figure Description

[0070] Figure 1 This is an overall flowchart of the method of the present invention.

[0071] Figure 2 This is a flowchart of the present invention.

[0072] Figure 3 This is a flowchart of the present invention.

[0073] Figure 4 This is a flowchart of the present invention. Detailed Implementation

[0074] like Figure 1-4 As shown, this invention provides a method for enhancing pulmonary vessels based on dual-modal imaging. This method combines fluorescence imaging and infrared imaging techniques to achieve high-quality imaging and enhancement of pulmonary vessels. The specific embodiments of this invention will be described in detail below.

[0075] First, the method of the present invention includes an acquisition step, a processing step, and an output step. In the acquisition step, a fluorescence image of the pulmonary vessels is acquired by fluorescence imaging, and an infrared image of the pulmonary vessels is acquired by infrared imaging. This dual-modal imaging technology can provide more comprehensive vascular information, laying the foundation for subsequent processing.

[0076] In the processing steps, the method of this invention first calculates the enhancement weights for the two modalities based on the intensity information of the fluorescence image and the infrared image, as well as the morphological information of the blood vessels. This step is one of the core innovations of this invention; by adaptively calculating the weights, it can optimize the enhancement according to the characteristics of different images. Specifically, this invention employs an adaptive weight calculation algorithm, the formula of which is as follows:

[0077] W p =W c ·W s

[0078] I p =W c ·I S +W s ·I S

[0079] Among them, W p For the final calculated weight, I p Let I be the final enhanced pixel value of pixel r(i,j,k). S W represents the pixel value of pixel r(i,j,k) after grayscale transformation. c and W s These represent the weights calculated based on the number and area of ​​connected regions, respectively.

[0080] Next, this method performs multi-scale vessel enhancement filtering on the obtained pseudo-color vessel images. This step employs multi-scale analysis technology, which can effectively enhance vessels of different thicknesses. Specifically, this invention uses a series of Gaussian kernels with different radii for convolution calculations to obtain a multi-scale image sequence. The advantage of this method is that it can simultaneously enhance vessels of different thicknesses, improving the comprehensiveness of the enhancement effect.

[0081] Finally, this method performs weighted fusion of the vascular enhancement results with the fluorescence and infrared images to obtain the final vascular enhancement result. This fusion process fully utilizes the advantages of dual-modal imaging, enabling the acquisition of clearer and more accurate vascular enhancement images.

[0082] In the output step, this method outputs the final vascular enhancement result. This result can be directly used for subsequent medical diagnosis and analysis.

[0083] Preferably, in one embodiment of the present invention, the acquisition step specifically includes intravenously injecting fluorescent nanoparticles and collecting fluorescence signals from pulmonary blood vessels under laser excitation to obtain fluorescence images. This method can clearly display vascular structures, especially small blood vessels. Simultaneously, this method employs laser imaging technology to acquire infrared images, which can provide information on deep tissue vascular structures.

[0084] Furthermore, the process of calculating enhancement weights in the processing steps of this invention specifically includes the following steps: First, grayscale stretching is performed on the fluorescence image and the infrared image respectively to obtain the enhanced image. This step can increase the contrast of the image and make the vascular structure more prominent. Second, pseudo-color mapping is performed on the enhanced image to obtain a pseudo-color vascular image. Pseudo-color mapping can help to better distinguish different vascular structures. Then, intensity enhancement results are obtained based on the intensity information of the pseudo-color vascular image, and morphological enhancement results are obtained based on vascular morphology and branching information. Finally, the enhancement weight of each modality is calculated based on the intensity enhancement results and the morphological enhancement results.

[0085] A key advantage of this adaptive weighting method is its ability to automatically adjust the enhancement strategy based on the characteristics of different images. For example, images with low signal-to-noise ratios may receive higher weights for intensity information, while images with numerous vascular branches may be given more consideration to morphological information. This flexibility allows the method to adapt to input images of varying quality and characteristics, improving its robustness and applicability.

[0086] In summary, the pulmonary vascular enhancement method based on dual-modal imaging provided by this invention achieves high-quality imaging and enhancement of pulmonary vessels by combining fluorescence imaging and infrared imaging techniques, and employing adaptive weight calculation and multi-scale enhancement techniques. This method not only improves the clarity and accuracy of vascular imaging but also better displays fine vascular structures, providing important technical support for the diagnosis and treatment of lung diseases. In a preferred embodiment of this invention, the multi-scale vascular enhancement filtering process includes a series of carefully designed steps aimed at comprehensively improving the quality and recognizability of vascular images. First, this method employs a multi-scale separation algorithm, using a series of Gaussian kernels with different radii for convolution calculation to obtain a multi-scale image sequence. The advantage of this method is that it can simultaneously process vascular structures of different scales, thereby achieving comprehensive enhancement of both large and small blood vessels.

[0087] Specifically, the Gaussian kernel radius used in this invention typically ranges from 1 to 10 pixels. This range has been validated through extensive experiments and is effective in covering most pulmonary blood vessel sizes. For example, a Gaussian kernel with a radius of 1-2 pixels is suitable for enhancing the smallest blood vessels, while a Gaussian kernel with a radius of 8-10 pixels is used for enhancing larger blood vessels.

[0088] Next, this method employs the Laplacian pyramid algorithm to perform multi-layer filtering on the pseudo-color blood vessel image, decomposing it into multiple groups of images to be enhanced. The use of the Laplacian pyramid algorithm allows this method to analyze and enhance images at different spatial frequencies, which is particularly important for structures like blood vessels that have multi-scale characteristics. Typically, this method uses a 3- to 5-layer Laplacian pyramid; this layer number is chosen based on a trade-off between computational efficiency and enhancement effect.

[0089] After obtaining multiple sets of images to be enhanced, the method of this invention calculates the channel similarity and scale similarity among these images. The purpose of this step is to evaluate the information correlation between different scales and different channels, providing a basis for subsequent fusion steps. The formula for calculating channel similarity is as follows:

[0090]

[0091] Among them, C ri R represents the channel similarity between the images to be enhanced in the i-th layer, M represents the total number of pixels, and R represents the channel similarity between the images. j (r) and R k (r) represents the probability that the r-th pixel in the j-th and k-th Laplacian feature layers is a blood vessel, respectively.

[0092] Similarly, the formula for calculating scale similarity is:

[0093]

[0094] Among them, S i This represents the scale similarity between the i-th layer and the j-th layer.

[0095] Based on the calculated channel similarity and scale similarity, this method further calculates the fusion weight for each image to be enhanced. This step is crucial for achieving adaptive enhancement, enabling the enhancement strategy to be automatically adjusted according to different scales and channel importance. The calculation of the fusion weight considers multiple factors, including channel similarity, scale similarity, and local image features.

[0096] Finally, this method reconstructs the enhanced image through weighted summation. This weighted fusion method can fully utilize information from different scales and channels to obtain more comprehensive and accurate vascular enhancement results.

[0097] In another embodiment of the invention, the method further includes further processing and optimization of the enhanced image. First, the enhanced image is binarized using a segmentation threshold to extract blood vessels and their centerlines. The segmentation threshold is typically set between 60% and 80% of the image's grayscale value, and the specific value can be fine-tuned according to the characteristics of the image. Binarization effectively separates blood vessels from the background, providing clear vessel outlines for subsequent analysis.

[0098] Next, this method calculates the feature information of the blood vessel contour. This feature information includes, but is not limited to, the width, orientation, and branching point locations of the blood vessel. Based on this feature information, this method further calculates the orientation of the blood vessel. The calculation of the blood vessel orientation uses an improved principal component analysis algorithm, which can accurately capture the overall orientation and local changes of the blood vessel.

[0099] Finally, based on the calculated vessel orientation, this method corrects the vessel centerline to obtain the final vessel enhancement result. The purpose of this step is to further improve the accuracy of the vessel centerline, especially the positioning accuracy at vessel intersections or branches.

[0100] Preferably, the present invention further includes a deep learning-based vascular cognition module. This module first performs deep learning on the final vascular enhancement result to extract the boundary features of each type of blood vessel in the image. The present invention employs an improved U-Net network structure, which, while maintaining the advantages of the original U-Net, adds an attention mechanism, enabling it to better capture the detailed features of blood vessels.

[0101] Based on the extracted boundary features, this method performs vessel fusion to obtain the original fused vessel image. This fusion process takes into account the characteristics of different types of vessels, such as the differences between arteries and veins, thus obtaining a more comprehensive and accurate representation of the vascular network.

[0102] Finally, this method performs vascular fusion correction on the original vascular fusion image to obtain a corrected vascular enhancement image. The purpose of this step is to eliminate potential artifacts and discontinuities, further improving the quality and reliability of the vascular enhancement results.

[0103] Through this series of meticulously designed steps, the method of this invention can achieve high-quality enhancement of pulmonary vessels, providing a reliable foundation for subsequent medical diagnosis and analysis. This method not only effectively enhances vessels at different scales but also further improves the accuracy of vessel recognition through deep learning technology, demonstrating the organic combination of traditional image processing techniques and modern artificial intelligence methods. In a preferred embodiment of this invention, the design of the vessel recognition module is further refined to achieve more accurate vessel recognition and contour extraction. First, this method filters the brightness gradient map of the enhanced vessel image to obtain the salient region corresponding to the foreground area. This step employs an improved Gaussian filtering algorithm, with the filter kernel size typically set to 5x5 or 7x7, depending on the image resolution and the thickness of the vessels. This filtering process effectively eliminates noise while preserving the main features of the vascular structure.

[0104] Next, this method filters the brightness gradient map of the significantly enhanced blood vessel image to obtain the blood vessel boundary region. A directional filter is used here to better capture the directional information of the blood vessel edges. The filter direction is typically set to 8 or 16, covering a range of 0 to 360 degrees, to accommodate different blood vessel orientations.

[0105] After obtaining the salient region and the vessel boundary region, the method of this invention synthesizes these two regions to obtain an enhanced vessel contour region. The synthesis process employs a weighted fusion method, with weight coefficients typically set to 0.6:0.4 or 0.7:0.3, and the specific values ​​can be fine-tuned according to the characteristics of the image. This synthesis method can effectively combine global saliency and local edge information, improving the accuracy of the vessel contour.

[0106] Preferably, this method also utilizes the vascular structure of the pre-enhancement image to correct the vascular boundary points in the vascular contour region. The purpose of this step is to eliminate artifacts or distortions that may be introduced by the enhancement process. The correction process employs a morphological operation-based approach, including opening and closing operations, with the structuring element size typically set to 3x3 or 5x5 to balance detail preservation and noise suppression.

[0107] After boundary point correction, the method of this invention uses Dijkstra's algorithm to smooth the vascular contour curve in the corrected vascular contour region. Dijkstra's algorithm is cleverly applied here to find the optimal smoothing path, effectively eliminating jagged edges and irregularities on the contour line. In the algorithm's parameter settings, the ratio of distance weight to smoothness weight is typically 3:1 or 4:1 to achieve a good smoothing effect while maintaining contour accuracy.

[0108] Finally, this method projects the smoothed vascular contour curves onto the enhanced image, completing the vascular recognition process. The projection process employs a bilinear interpolation algorithm to ensure precise alignment of the contour lines with the original image. The output of this step is an enhanced image containing accurate vascular contour information, providing a reliable foundation for subsequent medical analysis and diagnosis.

[0109] In another embodiment of the invention, the method further includes evaluating and optimizing the final vascular enhancement result. Specifically, the method finds the region corresponding to the enhancement result in the original image and marks the differences between the enhanced result and the vessels in the initial vascular fusion image in the corresponding original image to obtain a difference rate. This difference rate is calculated using a modified structural similarity index (SSIM) method, which considers not only pixel-level differences but also differences in structural information.

[0110] Preferably, this method sets a preset threshold to determine the effectiveness of the enhancement algorithm. If the difference rate is less than the preset threshold, the algorithm is considered successful; if the difference rate is greater than or equal to the preset threshold, the method is re-executed by returning to the acquisition step. This preset threshold is typically set to 10%, a value determined based on extensive experimental data and expert experience, which ensures the enhancement effect while avoiding over-processing.

[0111] It is worth noting that the method of this invention also includes a deep learning-based blood vessel recognition model. This model uses a trained lung blood vessel recognition model to identify the final blood vessel enhancement result and obtain the blood vessel classification result. The training process of the model adopts an innovative method. First, real blood vessel enhancement results are obtained from bimodal images as training data. Then, this training data is used as input to a deep learning framework, and the blood vessel recognition model is trained using a cross-entropy loss function and optimization methods.

[0112] Specifically, the deep learning model used in this invention is based on an improved U-Net structure, adding an attention mechanism and residual connections to enhance the model's ability to identify small blood vessels. During model training, the learning rate is initially set to 0.001 and adjusted using a cosine annealing strategy; the batch size is set to 16 or 32, depending on available computing resources; the number of training epochs is typically set to 100 to 200, and an early stopping strategy is used to avoid overfitting.

[0113] The use of the cross-entropy loss function enables the model to better handle imbalanced binary classification problems such as blood vessel segmentation. Its mathematical expression is as follows:

[0114]

[0115] Where N is the number of samples, y i For real labels, These are the model's predicted values.

[0116] Through this series of meticulously designed steps and innovative technical methods, the lung vascular enhancement method based on dual-modal imaging provided by this invention can achieve high-quality, high-precision vascular enhancement and identification. This method not only improves the clarity and accuracy of vascular imaging, but also further enhances the intelligence level of vascular identification by introducing deep learning technology, providing strong technical support for the precise diagnosis and treatment of lung diseases.

[0117] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto; any equivalent substitutions or modifications made by those skilled in the art within the scope disclosed in the present invention, based on the scheme and improved concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for enhancing pulmonary vessels based on dual-modal imaging, characterized in that, include: The acquisition steps include: acquiring a fluorescence image of pulmonary vessels through fluorescence imaging; acquiring an infrared image of pulmonary vessels through infrared imaging; the processing steps include: calculating enhancement weights for two modalities based on the intensity information of the fluorescence image and the infrared image, and the morphological information of the vessels, to obtain an adaptively enhanced pseudo-color vascular image; performing multi-scale vascular enhancement filtering on the pseudo-color vascular image to obtain a first vascular enhancement result; performing binarization processing on the first vascular enhancement result using a segmentation threshold, and extracting the vessels and their centerlines; calculating the feature information of the vessel contour; calculating the vessel orientation based on the feature information of the vessel contour; and correcting the vessel centerline according to the vessel orientation to obtain... The second vascular enhancement result is obtained; the second vascular enhancement result is weighted and fused with the fluorescence image and the infrared image to obtain the final vascular enhancement result; the output step includes: outputting the final vascular enhancement result, wherein the process of calculating the enhancement weight in the processing step specifically includes: performing grayscale stretching on the fluorescence image and the infrared image respectively to obtain the enhanced image; performing pseudo-color mapping on the enhanced image to obtain a pseudo-color vascular image; obtaining an intensity enhancement result based on the intensity information of the pseudo-color vascular image; obtaining a morphology enhancement result based on vascular morphology and branching information; and calculating the enhancement weight of each modality based on the intensity enhancement result and the morphology enhancement result.

2. The method according to claim 1, characterized in that, The acquisition steps specifically include: intravenously injecting fluorescent nanoparticles, collecting fluorescence signals from pulmonary blood vessels under laser excitation to obtain the fluorescence image; and acquiring the infrared image using laser imaging technology.

3. The method according to claim 1, characterized in that, It also includes the following steps: A vascular recognition module is designed to perform deep learning on the final vascular enhancement result and extract the boundary features of each type of blood vessel in the image; based on the boundary features, vascular fusion is performed to obtain the original vascular fusion image; the original vascular fusion image is then corrected to obtain the corrected vascular enhancement image.

4. The method according to claim 1, characterized in that, It also includes the following steps: For the final vascular enhancement result, the corresponding region is found in the original image; the difference between the enhanced result and the blood vessels in the initial vascular fusion image is marked in the corresponding original image to obtain the difference rate; if the difference rate is less than a preset threshold, the lung vascular enhancement method based on dual-modal imaging is determined to be successful; if the difference rate is greater than or equal to the preset threshold, the acquisition step is returned and the method is re-executed.

5. The method according to claim 4, characterized in that, The preset threshold is 10%.

6. The method according to any one of claims 1-5, characterized in that, It also includes the following steps: The trained lung vessel recognition model is used to identify the final enhanced vessel result to obtain a vessel classification result. The training method of the lung vessel recognition model includes: obtaining real vessel enhancement results in a bimodal image as training data; using the training data as input to a deep learning framework, and training the vessel recognition model using a cross-entropy loss function and optimization methods.

Images