Lung ICG fluorescence dynamic diffusion quantitative analysis method

Abstract

The invention relates to the technical field of lung blood vessel function analysis methods, in particular to a lung ICG fluorescence dynamic diffusion quantitative analysis method which comprises the following steps: acquiring a lung slice image of a patient; fluorescent tracer agent injection information is received; constructing a lung three-dimensional model based on the lung slice image; establishing a blood vessel model and a blood flow model based on the lung three-dimensional model; according to the fluorescent tracer injection information, a fluorescent tracer diffusion process is simulated in the lung three-dimensional model and the blood vessel model; calculating the flow velocity of all pipelines by adopting an optical flow method; determining a key time point according to a preset condition; calculating a flow value of each pipeline at the key time point of the tracing process based on the key time point; according to the flow value, the maximum flow change rate and the maximum flow difference value are calculated; according to the method, the physiological parameter information corresponding to the lung three-dimensional model is obtained, the blood vessel function evaluation model is generated, the blood vessel recognition accuracy is improved, and particularly the recognition capacity for tiny blood vessels is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pulmonary vascular function analysis methods, and in particular to a pulmonary ICG fluorescence dynamic diffusion quantitative analysis method. Background Art

[0002] With the continuous development of medical imaging technology, the diagnosis and evaluation methods of lung diseases are also constantly improving. Traditional methods for evaluating pulmonary vascular function mainly rely on technologies such as CT enhanced scanning and magnetic resonance angiography. Although these methods can provide certain anatomical information, they still have obvious deficiencies in dynamic blood flow evaluation and microvascular function analysis.

[0003] In recent years, indocyanine green (ICG) fluorescence imaging technology has attracted widespread attention due to its low invasiveness and high sensitivity. This technology injects ICG tracers and uses near-infrared fluorescence imaging equipment to capture the dynamic distribution of ICG in blood vessels, thereby evaluating vascular function. However, existing ICG fluorescence imaging technology still faces many challenges when applied to the evaluation of pulmonary vascular function.

[0004] First, the complex three-dimensional structure of the lungs makes it difficult for traditional two-dimensional fluorescence imaging to accurately reflect the spatial distribution and functional status of blood vessels. Second, existing methods often have difficulty accurately quantifying the dynamic diffusion process of ICG in blood vessels, which limits the accurate assessment of hemodynamic parameters. Furthermore, due to the lack of effective data processing and analysis methods, existing technologies have difficulty extracting valuable vascular function information from massive dynamic image data.

[0005] In addition, existing ICG fluorescence imaging technology has obvious deficiencies in the identification and functional assessment of tiny blood vessels in the lungs. The dysfunction of these tiny blood vessels is often an early sign of a variety of lung diseases, such as pulmonary embolism and pulmonary hypertension. However, due to the limitations of resolution and sensitivity, it is difficult for existing technologies to effectively evaluate the function of these tiny blood vessels.

[0006] Finally, existing methods for assessing pulmonary vascular function generally have problems such as complex operation, long analysis time, and high radiation dose, which not only increase the burden on patients but also limit the widespread application of these methods in clinical practice.

[0007] In view of the above problems, there is an urgent need for a new method that can achieve accurate quantitative analysis of pulmonary vascular function, especially for effective evaluation of microvascular function. Summary of the invention

[0008] In view of the above technical problems, the present invention proposes a method for optimizing the efficiency of multi-band radio frequency energy collection. The method should be able to provide high-resolution three-dimensional vascular function information, accurately capture the dynamic diffusion process of ICG, and extract valuable vascular function parameters from complex image data. At the same time, the method should also have the characteristics of simple operation, fast analysis, and low radiation dose to meet the needs of clinical practice.

[0009] The present invention provides a method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs, comprising the following steps:

[0010] Acquire lung slice images of the patient;

[0011] receiving fluorescent tracer injection information;

[0012] Based on the lung slice images, construct a three-dimensional lung model;

[0013] Based on the three-dimensional lung model, establishing a blood vessel model and a blood flow model;

[0014] According to the fluorescent tracer injection information, simulating the fluorescent tracer diffusion process in the three-dimensional lung model and the blood vessel model;

[0015] Based on the diffusion process, the optical flow method is used to calculate the flow rate of all pipes;

[0016] Determine key time points based on preset conditions;

[0017] Based on the key time points, calculating the flow value of each pipeline at the key time points of the tracing process;

[0018] According to the flow value, calculating the maximum flow rate change rate and the maximum flow difference;

[0019] Acquiring physiological parameter information corresponding to the three-dimensional lung model;

[0020] A vascular function assessment model is generated based on the maximum flow rate change rate, the maximum flow difference, the physiological parameter information and the light intensity curve.

[0021] Preferably, the constructing of the three-dimensional lung model specifically includes:

[0022] Dividing the three-dimensional lung model into N×N grids, where N is a preset number of grids;

[0023] Define each mesh as a node;

[0024] Adjacent nodes are defined as pipelines to form a virtual vascular network;

[0025] The pipes in the three-dimensional lung model are consistent with the pipes in the blood vessel model.

[0026] Preferably, the calculating the flow rates of all pipelines specifically includes:

[0027] Based on the optical flow method, the flow velocities of all pipelines are calculated;

[0028] The maximum flow rate Q of each pipe is calculated based on the pipe radius and the flow velocity, where:

[0029] Q=πr 2 v

[0030] Among them, π is pi, r is the inner radius of the pipe, and v is the maximum flow velocity;

[0031] Calculate the maximum flow of each node, where v i,j is the flow rate into pipe j, v i+1,j is the flow rate out of the i-th pipe j.

[0032] Preferably, the method for determining the key time point comprises:

[0033] Obtain the difference between two adjacent tracer concentrations;

[0034] Set the standard deviation value;

[0035] When the difference between any two adjacent tracer concentrations is greater than the standard deviation, the corresponding two time points are set as key time points.

[0036] Preferably, the method for calculating the maximum flow rate change rate and the maximum flow difference comprises:

[0037] Calculate the maximum flow rate change:

[0038]

[0039] Calculate the maximum flow difference:

[0040] Q n -Q n-1

[0041] Among them, Δt n is the time interval, Q n is the flow value at the nth key time point.

[0042] Preferably, the generation of a vascular function assessment model further comprises:

[0043] Set the maximum flow standard value of the node;

[0044] Analyze the relationship between the maximum flow rate change rate, the maximum flow difference, and the node maximum flow value and the standard value;

[0045] Based on the analysis results, the vascular function assessment model is established.

[0046] Preferably, the method further comprises the step of generating the light intensity curve:

[0047] Based on the key time points, obtaining the tracer concentration of the fluorescent tracer at the vascular nodes inside the three-dimensional lung model;

[0048] Plot tracer concentration curves over time;

[0049] The light intensity curve is generated according to the tracer concentration curve.

[0050] Preferably, the fluctuation amplitude of the light intensity curve is proportional to the light penetration.

[0051] Preferably, the method further comprises the step of abnormal pattern recognition:

[0052] Drawing a tracer concentration curve varying with time according to the tracer concentration of the vascular node at a number of the key time points;

[0053] generating a light intensity curve according to the tracer concentration curve;

[0054] performing anomaly detection on a plurality of said light intensity curves;

[0055] Extract abnormal features;

[0056] Build a dataset for identifying unusual patterns.

[0057] As a preference, it also includes:

[0058] Based on the vascular function evaluation model, analyzing the input lung ICG fluorescence image;

[0059] Generate a report on the assessment of pulmonary vascular function;

[0060] The pulmonary vascular function assessment report includes vascular perfusion status, abnormal area identification and oxygen supply capacity assessment.

[0061] By combining advanced image processing technology, fluid dynamics analysis and machine learning algorithms, the present invention proposes a new quantitative analysis method for dynamic diffusion of pulmonary ICG fluorescence. This method can not only accurately construct a three-dimensional pulmonary vascular model, but also accurately capture the dynamic diffusion process of ICG in blood vessels through the optical flow method. By introducing the concepts of virtual vascular networks and key time points, this method achieves simplified representation and efficient analysis of complex vascular structures. At the same time, by comprehensively considering multiple hemodynamic parameters and physiological indicators, this method establishes a comprehensive vascular function evaluation model.

[0062] The core innovation of the method of the present invention lies in its unique multidisciplinary integration method. By organically combining computer vision technology (such as optical flow method), fluid dynamics analysis and machine learning algorithms, this method achieves accurate quantification and in-depth analysis of the dynamic diffusion process of ICG fluorescence. This interdisciplinary approach not only overcomes the limitations of traditional technology in the assessment of microvessels, but also greatly improves the accuracy and comprehensiveness of vascular function assessment.

[0063] In addition, the method of the present invention can achieve high-precision analysis while significantly reducing the radiation dose received by the patient and shortening the examination time. This balance between effect and efficiency makes the method have broad clinical application prospects.

[0064] Specifically, the main beneficial effects brought about by the method of the present invention include:

[0065] 1. The accuracy of blood vessel recognition has been significantly improved, especially the recognition of micro-vessels (with a diameter as small as 0.5 mm). This makes it possible to detect micro-vascular lesions at an early stage.

[0066] 2. By accurately capturing the dynamic diffusion process of ICG, this method achieves high-precision measurement of hemodynamic parameters, with a relative error as low as ±5.2%. This provides a reliable basis for accurately evaluating the functional status of blood vessels.

[0067] 3. This method significantly improves the detection rate of abnormal areas (up to 92.5%), which is of great significance for the early diagnosis and precise positioning of lung diseases.

[0068] 4. While ensuring high-quality diagnostic information, this method reduces the radiation dose received by patients by nearly half and shortens the examination time by 40%. This not only reduces the burden on patients, but also improves the work efficiency of medical institutions.

[0069] 5. By establishing a comprehensive vascular function assessment model, this method provides a reliable basis for the formulation of individualized treatment plans, and is expected to significantly improve the accuracy and effectiveness of lung disease treatment.

[0070] In summary, the pulmonary ICG fluorescence dynamic diffusion quantitative analysis method proposed in the present invention innovatively solves several key problems faced by the existing technology and achieves a high-precision, low-invasive assessment of pulmonary vascular function. This method not only has significant technical advantages, but also has broad clinical application prospects, and is expected to bring major breakthroughs in the early diagnosis, precision treatment and prognosis assessment of lung diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 The present invention is a flow chart of the method.

[0072] Figure 2Flow chart of diffusion simulation and flow rate calculation of the present invention.

[0073] Figure 3 The flowchart of the flow analysis of the present invention.

[0074] Figure 4 A flow chart established for the evaluation model of the present invention. DETAILED DESCRIPTION

[0075] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0076] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0077] See also Figure 1-4 The present invention proposes a quantitative analysis method for dynamic diffusion of ICG fluorescence in the lungs, comprising an acquisition step, a processing step and an output step. In the acquisition step, a slice image of the patient's lungs is first acquired. These images are usually obtained by CT or MRI scanning, with high resolution and clear tissue structure information. Preferably, in one embodiment of the present invention, the layer thickness of the CT scan can be set to 1-2 mm to ensure that sufficiently detailed lung structure information is obtained. At the same time, the method also needs to receive fluorescent tracer injection information, including important parameters such as injection time and dosage. Typically, the injection dose of ICG can be 0.1-0.5 mg / kg body weight, and the injection time is usually 5 to 10 seconds before the start of the scan.

[0078] In the processing step, the present invention constructs a three-dimensional lung model based on the acquired lung slice images. This step involves complex image processing algorithms, such as image segmentation, three-dimensional reconstruction, etc. Preferably, in one embodiment of the present invention, a deep learning algorithm, such as U-Net or MaskR-CNN, can be used to improve the accuracy and efficiency of model construction. For example, when using U-Net for lung tissue segmentation, an input image size of 256x256 and a 5-layer deep network structure can be used to strike a balance between accuracy and computational efficiency.

[0079] Next, this method further builds a vascular model and a blood flow model based on the constructed 3D lung model. The vascular model needs to accurately describe the anatomical structure of the pulmonary blood vessels, while the blood flow model needs to consider the fluid mechanics of the blood. When building the blood flow model, the viscosity (usually between 3.5-5.5 mPa·s) and density (about 1060 kg / m 3 ) and other parameters.

[0080] Subsequently, the present invention simulates the diffusion process of the fluorescent tracer in the three-dimensional lung model and the blood vessel model according to the received fluorescent tracer injection information. This process is usually simulated using a computational fluid dynamics (CFD) method, which can more accurately reflect the movement and diffusion of the tracer in the blood vessel. In the CFD simulation, an appropriate turbulence model (such as the k-ε model) can be selected to describe the blood flow characteristics.

[0081] During the simulation process, the method uses the optical flow method to calculate the flow rate of all pipes. The optical flow method is a commonly used motion estimation algorithm. In the present invention, it is applied to the dynamic diffusion analysis of fluorescent tracers. This innovative application enables the method to accurately capture the motion characteristics of the tracer in the blood vessel. Preferably, the Horn-Schunck algorithm can be used to implement the optical flow method, where the smoothness parameter α can be set between 0.1 and 1.0, and the number of iterations is usually 50-200 times.

[0082] According to the preset conditions, the method determines the key time points. These key time points are essential for accurately evaluating vascular function. Preferably, the key time points can be determined according to the rate of change of the tracer concentration. For example, when the rate of change of the concentration exceeds 10% / s, the moment can be defined as the key time point.

[0083] Based on the determined key time points, this method calculates the flow value of each pipeline at the key time point of the tracing process. The flow value is calculated using the following formula:

[0084] Q=πr 2 v

[0085] Where Q is the flow rate, r is the pipe radius, and v is the flow velocity. Based on the calculated flow value, this method further calculates the maximum flow rate change rate and the maximum flow difference. These two parameters are of great significance for evaluating vascular function. The calculation formula for the maximum flow rate change rate is:

[0086]

[0087] Among them, Q i and Q i+1 are the flow values ​​of two adjacent key time points, t i and t i+1is the corresponding time point.

[0088] The maximum flow difference is calculated as:

[0089] Maximum flow difference = max i (Q i+1 -Q i )

[0090] These calculation results provide important basis for subsequent vascular function assessment.

[0091] During the processing, the method also obtains physiological parameter information corresponding to the three-dimensional lung model. These parameters may include the patient's age, gender, weight, smoking history, etc., which have an important impact on the accurate assessment of vascular function.

[0092] In the output step, the method generates a vascular function evaluation model based on the calculated maximum flow rate change, maximum flow difference, physiological parameter information and light intensity curve. This model takes into account multiple factors and can more comprehensively and accurately evaluate pulmonary vascular function. Preferably, a machine learning algorithm (such as a random forest or a support vector machine) can be used to construct the evaluation model, where the feature importance threshold can be set to 0.05 to screen out the most influential features.

[0093] Finally, this method outputs a vascular function assessment model, which can intuitively display the vascular function status of different lung regions and provide an important reference for clinical diagnosis and treatment.

[0094] Preferably, the specific step of constructing the three-dimensional lung model in the present invention includes dividing the three-dimensional lung model into N×N grids, where N is a preset number of grids. Preferably, the value of N can be determined according to the required accuracy and computing resources, and a power of 2 value such as 32, 64 or 128 can usually be selected. For example, when N=64, a good balance can be achieved between accuracy and computing efficiency.

[0095] Next, this method defines each grid as a node and defines adjacent nodes as pipes to form a virtual vascular network. This gridding method simplifies the complex pulmonary vascular structure and facilitates subsequent numerical calculations and analysis. It should be noted that the pipes in the 3D lung model are consistent with those in the vascular model, which ensures the correspondence between the virtual network and the actual vascular structure.

[0096] Preferably, the specific step of calculating the flow velocity of all pipes in the present invention first calculates the flow velocity of all pipes based on the optical flow method. The application of the optical flow method enables the present method to accurately capture the motion characteristics of the fluorescent tracer in the blood vessel. Preferably, the Lucas-Kanade algorithm or the Horn-Schunck algorithm can be used to implement the optical flow method. For example, when using the Horn-Schunck algorithm, the smoothness parameter α can be set to 0.5 and the number of iterations can be set to 100 to strike a balance between accuracy and computational efficiency.

[0097] Then, based on the pipe radius and flow velocity, this method calculates the maximum flow Q of each pipe. The calculation formula is:

[0098] Q=πr 2 v

[0099] Among them, π is the pi, r is the inner radius of the pipe, and v is the maximum flow velocity; this formula is based on the principles of fluid mechanics and accurately describes the relationship between the flow rate in a circular pipe and the pipe radius and flow velocity.

[0100] Finally, the method calculates the maximum flow at each node. In this step, considering the conservation of flow at the node, the inflow is equal to the outflow. Therefore, for node j, we have:

[0101]

[0102] Among them, v i,j is the flow rate into pipe j, v i+1,j is the flow rate out of the i-th channel j. This equation ensures the flow balance in the virtual vascular network.

[0103] Preferably, the method for determining the key time point in the present invention includes obtaining the difference between two adjacent tracer concentrations. This step is intended to capture significant changes in tracer concentrations, thereby identifying time points that are important for vascular function assessment.

[0104] Next, the method sets a standard deviation value. This standard deviation value is used as a threshold for judging whether the concentration change is significant. Preferably, this threshold can be set according to the actual application scenario and requirements. For example, the standard deviation value can be set to 5% of the average concentration. The selection of this value is based on experience, which can capture meaningful concentration changes and avoid misjudgment due to noise or small fluctuations.

[0105] Finally, when the difference between any two adjacent tracer concentrations is greater than the set standard deviation, the method sets the corresponding two time points as key time points. This method can effectively identify important moments in the tracer diffusion process and provide key information for subsequent vascular function assessment.

[0106] Preferably, the method for calculating the maximum flow rate change rate and the maximum flow difference in the present invention comprises the following steps:

[0107] First, calculate the maximum flow rate change rate. The calculation formula is:

[0108] (Q n -Q n-1 ) / Δt n

[0109] Among them, Q n is the flow value at the nth key time point, Q n-1 is the flow value at the n-1th key time point, Δt n is the time interval between these two time points. This formula reflects the degree of change in flow per unit time and is an important indicator for evaluating vascular function.

[0110] Then, calculate the maximum flow difference. The calculation formula is:

[0111] Q n -Q n-1

[0112] This difference directly reflects the absolute value of the flow change between two adjacent key time points and can capture sudden changes in flow.

[0113] By calculating these two indicators, this method can comprehensively evaluate the functional status of blood vessels at different time points and provide important input for the subsequent vascular function evaluation model. Preferably, the maximum flow rate change rate and flow difference can be selected as feature inputs into the evaluation model to capture the most significant changes in vascular function.

[0114] Through the above steps, it can be seen that the quantitative analysis method of dynamic diffusion of pulmonary ICG fluorescence of the present invention has the following advantages: First, it combines advanced image processing technology and fluid dynamics analysis, and can accurately construct a pulmonary vascular model and simulate the diffusion process of fluorescent tracers. Secondly, by introducing the concept of optical flow method and key time points, this method can accurately quantify the dynamic diffusion process and provide rich vascular function information. Finally, by comprehensively considering multiple parameters and establishing an evaluation model, this method can comprehensively and accurately evaluate pulmonary vascular function and provide strong support for clinical diagnosis and treatment.

[0115] As preferably, the step of generating a vascular function evaluation model in the present invention also includes setting a standard value for the maximum flow rate of a node. The setting of this standard value is crucial for accurately evaluating vascular function. Preferably, in one embodiment of the present invention, the maximum flow rate distribution of nodes in a normal population under different factors such as age, gender, and weight can be obtained based on a large amount of clinical data statistics, thereby determining a reasonable standard value range. For example, for adult males, the standard flow rate of the pulmonary lobe aorta can be set to 4-6L / min, while for adult females, this value may be slightly lower, set to 3.5-5.5L / min.

[0116] Next, the method analyzes the maximum flow rate change, the maximum flow difference, and the relationship between the node maximum flow value and the standard value. This step aims to comprehensively evaluate the functional status of the blood vessels. Specifically, the maximum flow rate change reflects the response speed of the blood vessels to changes in blood flow, the maximum flow difference indicates the amplitude of the change in the blood vessel capacity, and the comparison of the node maximum flow value with the standard value can directly reflect whether the blood vessel function is normal.

[0117] Based on the above analysis results, the present invention establishes a vascular function assessment model. This model takes into account multiple factors and can evaluate pulmonary vascular function more comprehensively and accurately. Preferably, a machine learning algorithm can be used to construct an assessment model, such as a random forest or a support vector machine (SVM). For example, when using a random forest algorithm, the number of trees can be set to 100-500 and the maximum depth to 10-20 to strike a balance between model complexity and generalization ability. The input features of the model may include the maximum flow rate of change, the maximum flow difference, the ratio of the node maximum flow value to the standard value, etc. The output can be a score of 0-100, indicating the health of vascular function.

[0118] Preferably, the present invention also includes the step of generating a light intensity curve. First, based on the determined key time points, the method obtains the tracer concentration of the fluorescent tracer at the vascular node inside the three-dimensional lung model. This step requires precise image processing technology to extract accurate concentration information from the fluorescent image. Preferably, an image segmentation algorithm (such as threshold segmentation or region growing method) can be used to identify the vascular area, and then the fluorescence intensity is quantified by pixel intensity analysis to infer the tracer concentration.

[0119] Next, the method plots a tracer concentration curve that changes with time. This curve intuitively shows the diffusion process of the tracer in the blood vessel. Preferably, a spline interpolation method can be used to smooth the curve to reduce the influence of measurement noise. For example, cubic spline interpolation can be used, and the control point interval can be set to 1 to 2 seconds to strike a balance between smoothness and detail preservation.

[0120] Finally, the method generates a light intensity curve based on the tracer concentration curve. This step takes into account the relationship between fluorescence intensity and tracer concentration, which can usually be assumed to be linear within a certain range. Preferably, this relationship can be determined by a preliminary calibration experiment, for example, a relationship similar to I = kC + b can be obtained, where I is the light intensity, C is the concentration, and k and b are coefficients to be determined.

[0121] Preferably, the fluctuation amplitude of the light intensity curve in the present invention is proportional to the light penetration. This characteristic reflects the absorption and scattering effect of biological tissue on fluorescence. Specifically, the stronger the light penetration, the deeper the fluorescent signal can penetrate into the tissue, resulting in a larger fluctuation amplitude. Preferably, the light penetration coefficient of different tissue types can be determined experimentally. For example, for lung tissue, the penetration coefficient of near-infrared light (such as fluorescence emitted by ICG) may be between 0.5 and 1.0 mm. -1 This information can be used to correct the light intensity curve and improve the accuracy of vascular function assessment.

[0122] Preferably, the present invention further comprises an abnormal pattern recognition step. First, based on the tracer concentration of the vascular nodes at several key time points, the method draws a tracer concentration curve that changes with time. This step is similar to the above description, but the purpose here is to recognize abnormal patterns.

[0123] The method then generates a light intensity curve based on the tracer concentration curve. This step is also similar to the one described previously, but here we focus specifically on the anomalous features of the curve.

[0124] Next, the method performs anomaly detection on several light intensity curves. This step aims to identify vascular function patterns that do not conform to normal physiological states. Preferably, multiple anomaly detection algorithms can be used, such as Isolation Forest or single-class SVM. For example, when using the Isolation Forest algorithm, the contamination parameter can be set to 0.1, which means that it is expected that about 10% of the samples may be abnormal.

[0125] Based on anomaly detection, the method extracts anomaly features. These features may include the peak value, valley value, rising rate, falling rate, etc. of the curve. Preferably, a dimensionality reduction technique such as principal component analysis (PCA) or autoencoder can be used to extract the most representative anomaly features. For example, when using PCA, the number of principal components required to explain 95% of the variance can be selected, which is usually between 3 and 10.

[0126] Finally, the method establishes a data set for identifying abnormal patterns. This data set will become the basis for subsequent abnormal pattern recognition. Preferably, a semi-supervised learning method can be used to enhance the representativeness of the data set. For example, a small number of labeled abnormal samples and a large number of unlabeled samples can be used to expand the number of abnormal samples through pseudo-labeling technology.

[0127] Preferably, the present invention further comprises a step of analyzing the input lung ICG fluorescence image based on the vascular function evaluation model. This step applies the previously established evaluation model to new image data to achieve vascular function evaluation of unknown samples. Preferably, a sliding window technique can be used to process large-sized lung images, the window size can be set to 64x64 or 128x128 pixels, and the step size can be set to half of the window size to ensure sufficient overlap and smooth evaluation results.

[0128] Finally, the method generates a pulmonary vascular function assessment report. This report integrates all the previous analysis results and provides an important reference for clinical diagnosis. Preferably, the report can include the following parts: first, the vascular perfusion situation, which can use a heat map to intuitively display the blood flow conditions in different areas; second, the abnormal area identification, which can use different colors or marks to highlight the vascular areas that may have problems; finally, the oxygen supply capacity assessment, which can calculate the oxygen supply capacity index of each lobe or smaller area based on blood flow and oxygen saturation information.

[0129] Through the above steps, it can be seen that the pulmonary ICG fluorescence dynamic diffusion quantitative analysis method of the present invention can not only comprehensively evaluate vascular function, but also identify abnormal patterns and generate intuitive evaluation reports. This method combines advanced image processing technology, fluid dynamics analysis and machine learning algorithms to provide a powerful tool for early diagnosis and precise treatment of lung diseases. Especially in the diagnosis of diseases such as pulmonary embolism and pulmonary hypertension, this method is expected to significantly improve the accuracy and timeliness of diagnosis, thereby improving the prognosis of patients.

[0130] In order to further verify the superiority of the quantitative analysis method of dynamic diffusion of ICG fluorescence in lungs of the present invention, a group of typical cases were selected for example testing and compared with the traditional CT enhanced scanning method.

[0131] Example 1: The patient is a 55-year-old male, 175cm tall, 70kg in weight, with a 10-year history of smoking, and recently developed symptoms of chest tightness and shortness of breath. We first examined the patient using the method of the present invention. The scanning parameters are set as follows: CT layer thickness 1.5mm, ICG injection dose 0.3mg / kg body weight, and injection time 7 seconds before scanning. U-Net is used for lung tissue segmentation, the network input image size is 256x256, and the network depth is 5 layers. When constructing the virtual vascular network, we choose N=64 for grid division. The optical flow method uses the Horn-Schunck algorithm, sets the smoothness parameter α=0.5, and the number of iterations is 100.

[0132] Comparative Example 1: The same patient received a conventional enhanced CT scan the day after the examination by the method of the present invention. The scanning parameters were the same as those in Example 1, but conventional iodine contrast agents were used for enhanced scanning.

[0133] Testing indicators, methods and results:

[0134] The present invention selects the following key indicators for comparison to verify the superiority of the method of the present invention:

[0135] 1. Blood vessel recognition accuracy: The ratio of correctly identified blood vessels to the total number of blood vessels is calculated through manual labeling.

[0136] 2. Minimum detectable vessel diameter: the minimum vessel diameter that can be accurately identified and evaluated.

[0137] 3. Hemodynamic parameter measurement accuracy: The relative error of blood flow velocity measurement is calculated by comparing with the ultrasound Doppler measurement results.

[0138] 4. Abnormal area detection rate: By comparing with the clinical diagnosis results, calculate the ratio of the number of correctly detected abnormal areas to the total number of abnormal areas.

[0139] 5. Radiation dose: The radiation dose received by the patient during the examination.

[0140] 6. Inspection time: the total time from the start of the scan to obtaining the final assessment report.

[0141] The test results are shown in the following table:

[0142]

[0143]

[0144] It can be seen from the above results that the method of the present invention is significantly superior to the traditional CT enhanced scanning method in all key indicators. It is particularly noteworthy that the method of the present invention can identify and evaluate blood vessels as small as 0.5mm in diameter, which greatly improves the detection ability of microvascular lesions. At the same time, the significant improvement in the measurement accuracy of hemodynamic parameters makes the evaluation of pulmonary vascular function more accurate and reliable.

[0145] In addition, the method of the present invention has a particularly outstanding advantage in the detection rate of abnormal areas. In the examination of this patient, the method of the present invention successfully detected a tiny pulmonary embolism lesion in the right upper lobe of the lung, which was not clearly identified in the traditional CT enhanced scan. This fully demonstrates the advantages of the method of the present invention in early diagnosis and precise positioning.

[0146] It is worth mentioning that the method of the present invention not only ensures high-quality diagnostic information, but also significantly reduces the radiation dose received by the patient, which is particularly important for patients who need long-term follow-up. The shortened examination time helps to improve the work efficiency of medical institutions and also reduces the burden on patients.

[0147] Based on the above embodiments and a large number of clinical practices, the optimal implementation parameters of the method of the present invention are summarized as follows:

[0148] 1. CT scanning parameters: slice thickness 1.0-1.5 mm, tube voltage 120 kV, tube current automatic modulation.

[0149] 2. ICG injection: dosage 0.3-0.4 mg / kg body weight, injection time 5-8 seconds before scanning.

[0150] 3. Image segmentation: Use U-Net or MaskR-CNN, input image size 256x256 or 512x512, network depth 5-7 layers.

[0151] 4. Virtual vascular network: N = 64 or 128, depending on the specific situation.

[0152] 5. Optical flow method: using Horn-Schunck algorithm, smoothness parameter α = 0.4-0.6, iteration times 100-200.

[0153] 6. Anomaly detection: The isolation forest algorithm is used, and the contamination parameter is set to 0.05-0.15.

[0154] This set of parameters not only ensures high-precision analysis results, but also takes into account computational efficiency and clinical practicality. It should be noted that for different patient groups (such as children, pregnant women or the elderly), some parameters may need to be appropriately adjusted to ensure safety and diagnostic effectiveness.

[0155] In general, the pulmonary ICG fluorescence dynamic diffusion quantitative analysis method of the present invention realizes a high-precision, low-invasive assessment of pulmonary vascular function by innovatively combining optical flow method, virtual vascular network construction and machine learning technology. This method not only improves the accuracy and timeliness of lung disease diagnosis, but also provides a reliable basis for the formulation of individualized treatment plans.

[0156] 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 principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for quantitative analysis of dynamic diffusion of ICG fluorescence in the lungs, characterized in that: The following steps are involved: Acquire lung slice images of the patient; receiving fluorescent tracer injection information; Based on the lung slice images, construct a three-dimensional lung model; Based on the three-dimensional lung model, establishing a blood vessel model and a blood flow model; According to the fluorescent tracer injection information, simulating the fluorescent tracer diffusion process in the three-dimensional lung model and the blood vessel model; Based on the diffusion process, the optical flow method is used to calculate the flow rate of all pipes; Determine key time points based on preset conditions; Based on the key time points, calculating the flow value of each pipeline at the key time points of the tracing process; According to the flow value, calculating the maximum flow rate change rate and the maximum flow difference; Acquiring physiological parameter information corresponding to the three-dimensional lung model; A vascular function assessment model is generated based on the maximum flow rate change rate, the maximum flow difference, the physiological parameter information and the light intensity curve.

2. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The constructing of the three-dimensional lung model specifically includes: Dividing the three-dimensional lung model into N×N grids, where N is a preset number of grids; Define each mesh as a node; Adjacent nodes are defined as pipelines to form a virtual vascular network; The pipes in the three-dimensional lung model are consistent with the pipes in the blood vessel model.

3. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The calculation of the flow rates of all pipelines specifically includes: Based on the optical flow method, the flow velocities of all pipelines are calculated; The maximum flow rate Q of each pipe is calculated based on the pipe radius and the flow velocity, where: Q=πr 2 v Among them, π is pi, r is the inner radius of the pipe, and v is the maximum flow velocity; Calculate the maximum flow of each node, where v i,j is the flow rate into pipe j, v i+1,j is the flow rate out of the i-th pipe j.

4. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The method for determining the key time point comprises: Obtain the difference between two adjacent tracer concentrations; Set the standard deviation value; When the difference between any two adjacent tracer concentrations is greater than the standard deviation, the corresponding two time points are set as key time points.

5. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The method for calculating the maximum flow rate change rate and the maximum flow difference comprises: Calculate the maximum flow rate change: Calculate the maximum flow difference: Q n -Q n-1 Among them, Δt n is the time interval, Q n is the flow value at the nth key time point.

6. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The generated vascular function evaluation model further comprises: Set the maximum flow standard value of the node; Analyze the relationship between the maximum flow rate change rate, the maximum flow difference, and the node maximum flow value and the standard value; Based on the analysis results, the vascular function assessment model is established.

7. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: The step of generating the light intensity curve is also included: Based on the key time points, obtaining the tracer concentration of the fluorescent tracer at the vascular nodes inside the three-dimensional lung model; Plot tracer concentration curves over time; The light intensity curve is generated according to the tracer concentration curve.

8. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 7, characterized in that: The fluctuation amplitude of the light intensity curve is proportional to the light penetration.

9. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: It also includes the abnormal pattern recognition step: Drawing a tracer concentration curve varying with time according to the tracer concentration of the vascular node at a number of the key time points; generating a light intensity curve according to the tracer concentration curve; performing anomaly detection on a plurality of said light intensity curves; Extract abnormal features; Build a dataset for identifying unusual patterns.

10. The method for quantitative analysis of dynamic diffusion of ICG fluorescence in lungs according to claim 1, characterized in that: Also includes: Based on the vascular function evaluation model, analyzing the input lung ICG fluorescence image; Generate a report on pulmonary vascular function assessment; The pulmonary vascular function assessment report includes vascular perfusion status, abnormal area identification and oxygen supply capacity assessment.

Images