153 results about "Blood vessel structure" patented technology

The invention relates to the technical field of medical image analysis, and discloses a cardiovascular disease detection method based on image analysis. The method comprises the steps of obtaining a medical image sequence of a target patient, and extracting a blood vessel region contour to generate an initial blood vessel topological graph; the initial topological graph is registered with a standard cardiovascular model, the curvature deviation degree and the pipe diameter variation coefficient of each blood vessel branch are calculated, and a blood vessel elastic characteristic matrix is generated in combination with abnormal displacement nodes; and performing multi-scale fusion on the curvature deviation degree, the pipe diameter variation coefficient and the elastic characteristic matrix, outputting a vascular structure anomaly index, and generating a hemodynamic parameter set. Dividing risk areas according to gradient distribution of the parameter set on the three-dimensional model, performing texture co-occurrence matrix analysis on pixel clusters in the high-risk areas, and extracting texture fingerprints of calcified plaques and lipid deposition. And determining the type and severity level of the cardiovascular disease according to the peak value distribution. According to the invention, accurate and objective detection and risk assessment of cardiovascular diseases are realized.

The invention discloses a human body peripheral blood vessel photoacoustic three-dimensional imaging device and an imaging method thereof, and belongs to the field of blood vessel imaging. Through high resolution and functional monitoring capability of photoacoustic imaging, detailed blood vessel structure and blood flow dynamic information are provided, and early recognition and treatment effect evaluation of common lesions such as diabetic foot and peripheral artery diseases are promoted; meanwhile, a mechanical system is adopted, so that the detection front end with relatively high quality can be borne, and stable operation is realized, and a relatively good photoacoustic imaging result is obtained; the system has a highly stable accurate scanning imaging mode, and is convenient for multiple imaging modes of clinical application; through the arrangement of the imaging water tank and the lifting structure of the imaging water tank, switching of imaging modes in a limited space can be ensured, so that the device can be suitable for the layout of a portal type three-dimensional moving device. The system is expected to promote the portability of equipment, reduce the research and development cost of the equipment, be widely applied to clinical practice and community medical treatment, and improve the disease management level.

The invention relates to the technical field of image modeling, in particular to a three-dimensional modeling method and system for medical image data, and the method comprises the steps: carrying out the preprocessing of the obtained two-dimensional or three-dimensional medical image data, extracting a blood vessel region based on the image texture and gray features, and generating a preliminary blood vessel structure segmentation image; in combination with a blood flow simulation or real acquisition means, obtaining dynamic feature data from the vascular structure segmentation map, and mapping the dynamic feature data to a corresponding vascular structure position to form a vascular distribution map with blood flow constraint; and inputting the blood vessel distribution diagram into a three-dimensional modeling module, expanding and complementing small blood vessel branches by utilizing blood flow constraint, dynamically correcting the form of a main blood vessel, and generating a three-dimensional blood vessel structure model with real fluid physiological characteristics. The method can be widely applied to medical research and clinical aid decision-making scenes such as preoperative simulation, individualized intervention path planning and blood flow reconstruction analysis, and has good popularization and application prospects and engineering practical value.

The invention discloses a medical image optimization method and system based on vascular branch selective blurring. The method comprises the following steps: carrying out preprocessing and blood vessel enhancement on a three-dimensional angiography DICOM image; performing precise blood vessel segmentation by using an encoder-decoder network comprising a wide activation and residual cavity space pyramid module; identifying an interference branch which shields the target blood vessel structure at a specific working angle through blood vessel topology analysis; selective blurring repair is performed on the interference branches based on a Poisson equation and a bidirectional convolution LSTM to generate an optimized image which is visually unobstructed and keeps topological continuity. And a safety mechanism of virtual-real combined display and multi-angle plan planning is introduced, so that the reliability of surgical navigation is ensured. And finally, integrating the optimized 3D blood vessel model to a radiography system supporting real-time synchronization, and using the 3D blood vessel model as a road map. The clinical problem that the optimal working angle cannot be used due to blood vessel shielding can be effectively solved, and the precision and safety of an endovascular interventional operation are remarkably improved.

The invention relates to the technical field of blood vessel structure recognition, and discloses an automatic recognition method and system for an OCT image blood vessel structure, and the method comprises the steps: 1, blood vessel layer boundary feature extraction: extracting blood vessel layer boundary features, including geometric features and physical features, from an OCT image based on multi-scale Hessian matrix response, and forming a feature response graph; step 2, three-layer topological structure identification: constructing a topological relation of a blood vessel structure according to the feature response diagram, and extracting a nested annular structure; identifying and outputting a three-layer initial boundary coordinate point set and a structure score of an inner membrane, a middle membrane and an outer membrane of a blood vessel; and step 3, optimizing a blood vessel structure: accurately optimizing the initial boundary coordinate point set, detecting special cells in an optimized region, and generating a result report. According to the method, full-automatic recognition of the OCT blood vessel structure is achieved, reliability, accuracy and interpretability of a recognition result are ensured, and standardization and unification of vascular disease diagnosis are ensured.

The invention relates to a neural puncture dynamic obstacle avoidance navigation method and system based on multi-modal image fusion, and belongs to the technical field of medical image navigation and surgical operations. A unified space-time reference model is constructed by fusing preoperative high-resolution static images and intraoperative real-time dynamic ultrasonic data; a three-dimensional model of a key nerve and blood vessel structure is automatically segmented and reconstructed by using an artificial intelligence technology, an optimal puncture path can be dynamically calculated based on real-time image data, a collision risk in a needle inserting process can be monitored in real time, and visual navigation guidance of multi-sensory fusion is provided for an operator through an augmented reality technology; by recording and analyzing data of the whole operation process, a big data platform is used for carrying out continuous iterative optimization on a segmentation and obstacle avoidance algorithm, and a self-perfect intelligent closed loop is formed. The problems that in a traditional nerve puncture operation, experience of an operator is relied on, real-time dynamic obstacle avoidance cannot be achieved, and the system lacks learning ability are effectively solved, and the accuracy, safety and intelligent level of the operation are remarkably improved.

The invention relates to a method for operating an X-ray device, comprising a carrier, in particular a C-arm, and an adjusting device at least for moving the carrier for setting a projection geometry. The method comprises: providing a vascular model of a vascular structure of an examination subject, in which a target position of an instrument movable in the vascular structure is marked in the vascular model, providing an initial position of the instrument in the vascular structure of the vascular model, and providing a path from the initial position to the target position; specifying a support point along the path; in the optimization process of the target function by using the blood vessel model, the optimization trend of the projection geometric structure is determined aiming at the supporting points; wherein the objective function comprises at least one first term relating to optimization of the image content of the observer at the respective support point and at least one second term that minimizes the number of changes in the projection geometry along the path by movement of the carrier; and determining, for each projection geometry, an adjustment parameter for controlling the adjustment device.

The utility model discloses a high-simulation individualized liver intervention model. The high-simulation individualized liver intervention model comprises a model body, a puncture needle and a computer system, a liver model and a tumor model are arranged on the model body, a first simulation pipeline for simulating an intrahepatic bile duct and a plurality of second simulation pipelines for simulating intrahepatic blood vessels are arranged in the liver model, the first simulation pipeline is connected with a first water bag through a first external pipeline, and the first external pipeline is connected with a first water pump; the second simulation pipeline is connected with a second water bag through a second external pipeline, the second water bag is connected with a second water pump, and the color of liquid carried in the first water bag is different from that of liquid carried in the second water bag; a first sensing element is arranged in the tumor model, and second sensing elements are arranged on the surface of the first simulation pipeline and the surface of the second simulation pipeline. The device can simulate structures such as intrahepatic tumors, intrahepatic bile ducts, intrahepatic blood vessels and the like, can simulate puncture targets with different puncture difficulties, and can pre-warn puncture risks and record the puncture success rate.

The invention relates to the technical field of blood vessel image segmentation, and provides a fundus blood vessel image accurate segmentation method based on a U-Net model, aiming at solving the problem that the existing blood vessel segmentation method based on deep learning is difficult to effectively balance global context and local details. In order to solve the problem of inaccurate blood vessel structure segmentation caused by factors such as poor image quality and low contrast ratio, a method of combining a wavelet and a visual state space model and an adaptive dynamic fusion module is provided to improve the integrity of a slender blood vessel and the structural segmentation accuracy of a complex blood vessel network in a fundus blood vessel image. By integrating multiple efficient feature extraction mechanisms, accurate recognition of small blood vessels and the whole blood vessel structure is achieved, the accuracy and robustness of blood vessel segmentation in complex medical images are remarkably improved, and the method can be widely applied to analysis of eye blood vessels such as retinopathy and hypertensive fundus lesion and research of biomarkers related to the blood vessels.

The invention provides a hierarchical decoupling autonomous catheter navigation control method and system, and the method comprises the steps: extracting the position of a catheter and the structure information of a blood vessel based on an obtained intraoperative real-time DSA image, and constructing a topological structure diagram for describing the connectivity of a blood vessel network; performing global path planning and navigation sub-target selection according to the vascular topological graph and the final target; receiving a current navigation sub-target specified by the high-level semantic planning step, and generating a continuous motion control instruction for driving a catheter to the sub-target according to real-time local visual feedback and a pre-trained reinforcement learning strategy; and converting the control instruction into a robot driving signal to drive the surgical robot to move. According to the method, the interpretability and transparency of decisions are improved, doctors can understand and trust the decisions conveniently, the modularization degree and robustness of the system are enhanced, and development, maintenance and clinical popularization of the system are facilitated.

The application discloses a percutaneous lung puncture path planning method and device based on lung segment level blood vessel structure constraint, and the method comprises the following steps: performing isotropic resampling, lung lobe segmentation and bronchial tree extraction on CT image data to generate a three-dimensional lung dissection model; inputting the three-dimensional lung dissection model into a lung segment semantic segmentation module to output a lung segment boundary; reconstructing the topological connection relationship between lung segment level blood supply arteries and drainage veins from the lung segment boundary to obtain a structured lung segment level blood vessel atlas; constructing a dynamic puncture risk field voxel map according to the structured lung segment level blood vessel atlas; sampling a plurality of candidate needle entry points in a surface feasible needle entry region to generate an initial path set in the three-dimensional lung dissection model; and weighting the path length, cumulative risk value, puncture angle penalty and high-density lung tissue volume according to a multi-objective optimization function to output a final puncture path. The application significantly improves the safety boundary and clinical applicability of percutaneous lung puncture.

The invention discloses a coronary artery segmentation method and system based on boundary perception and a dynamic multi-scale attention network. An image segmentation model is designed based on a U-shaped network, a dual attention enhancement module and a dynamic multi-scale feature fusion module are introduced into a network bottleneck layer, and a boundary optimization module is introduced into a decoder stage. The dual attention enhancement module focuses the network on a key vascular structure in combination with channel and space attention so as to suppress background interference; the dynamic multi-scale feature fusion module fuses different receptive field features through a self-adaptive gating mechanism to adapt to the blood vessel scale change; and the boundary optimization module explicitly enhances blood vessel boundary information by using superficial layer characteristics. In addition, a composite loss function containing boundary weighted loss is adopted for training, and the sensitivity of the model to the blood vessel fuzzy boundary is enhanced. Experiments show that compared with an existing method, the method can effectively improve the segmentation performance of the blood vessel in the coronary artery angiography image.

The present application relates to the technical field of medical navigation, in particular to an augmented reality blood vessel fusion navigation system for transcatheter occlusion, comprising a real-time image acquisition module, a blood vessel three-dimensional model construction module, an image enhancement module, a navigation module and a fusion module, the image acquisition module is connected with an ultrasound catheter to acquire blood vessel three-dimensional images, the model construction module creates a blood vessel three-dimensional model based on the images, the image enhancement module enhances the blood vessel images through spatial adaptive transformation, key point recognition, region adaptive segmentation and edge feature extraction, the navigation module determines the catheter tip position by using the blood vessel three-dimensional model and the enhanced images, the fusion module fuses the enhanced images with the blood vessel model to generate fusion display data, and an AR device overlays and displays the blood vessel three-dimensional model at the catheter tip position based on the catheter tip position and the fusion data, which significantly improves the visualization effect of complex blood vessel structures and provides an accurate spatial reference for catheter navigation.

The invention provides an angiography image segmentation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining an initial angiography image which comprises a to-be-segmented blood vessel structure region; based on a preset blood vessel response function, performing enhancement processing on a blood vessel structure region contained in the initial angiography image to obtain an enhanced angiography image; segmenting the enhanced angiography image through the target segmentation model to obtain an initial blood vessel structure region corresponding to the enhanced angiography image, the initial blood vessel structure region including a blood vessel center line; and performing expansion processing on the blood vessel center line based on the initial blood vessel structure region to obtain a blood vessel segmentation result. According to the embodiment provided by the scheme, the accuracy and reliability of angiography image segmentation can be improved.

A method for processing a three-dimensional model of a vessel structure comprises: providing the model, the model including a topological graph representation of the vessel structure and first courses of centerlines of vessels; receiving user input data describing at least one corrected course of a centerline; determining, from the user input data and the model, an affected section of the chosen centerline, at least one non-affected section of the chosen centerline, and, for each branch-off along the affected section, a modifiable section connected to the affected section, and a non-modifiable section outside the affected section; determining a deformation field from the affected section, the non-affected section, the non-modifiable section, and the corrected course, such that the deformation field maps the first course of the chosen centerline to the corrected course without changing non-modifiable sections; and applying the deformation field to the affected section and all modifiable sections.

The application discloses a kind of free skin flap postoperative nursing scheme evaluation and early warning method based on multi-modal monitoring data fusion, it is related to intelligent medical care technical field, for solving the problem that existing monitoring cannot distinguish vascular pathological crisis and nursing environment interference.First, a multi-dimensional time-varying state matrix containing physiological drive and nursing environment characteristics is constructed, realizing the time sequence alignment of heterogeneous data;Subsequently, the pressure-flow conduction characteristics are analyzed, and the blood flow conduction efficiency index reflecting the inherent patency of blood vessels is extracted;At the same time, a physical gain mapping is established based on nursing parameters, and the theoretical target perfusion trajectory under the current nursing state is deduced;Finally, the dynamic residual vector is calculated and combined with the conduction efficiency index for joint determination, accurately identifying vascular structural occlusion and functional inhibition of nursing program, achieving accurate tracing of postoperative crisis and dynamic evaluation of nursing program.

The invention relates to the field of retinal arteriovenous blood vessel image segmentation, and provides a retinal arteriovenous blood vessel image segmentation method and device, and the method comprises the steps: obtaining a to-be-segmented fundus color photo; inputting the fundus color photo into the trained retina artery and vein segmentation model to obtain corresponding artery and vein segmentation maps; wherein the retina arteriovenous blood vessel segmentation model comprises a feature extraction module, a feature fusion module and a result generation module; the feature extraction module is used for extracting global features, local features and frequency features of the input image; the feature fusion module is used for fusing the global features and the local features into multi-scale features and fusing the local features and the frequency features into multi-dimensional features; and the result generation module is used for generating artery and vein segmentation maps according to the multi-scale features and the multi-dimensional features. According to the invention, the detail part of the segmentation result is more complete and real, the segmentation accuracy of the vascular structure can be improved, and the segmentation result is more reasonable in topology.

The invention discloses a medical image key structure extraction method for vascular surgery preoperative evaluation, which comprises the following steps: an image acquisition and multi-modal data acquisition stage: acquiring a multi-source medical image and clinical hemodynamic parameters, and generating a blood vessel initial segmentation map after processing; a structural feature reconstruction stage: extracting morphological features of a vascular wall and geometric features of a lumen, and generating a vascular tree three-dimensional structure map; in the pathological feature fusion stage, the vascular tree three-dimensional structure atlas and clinical hemodynamic parameters are fused, and vasculopathy risk confidence is generated; in the operation planning decision-making stage, the vasculopathy risk confidence coefficient, the vascular tree three-dimensional structure atlas and the clinical hemodynamic parameters are converted into an executable operation planning suggestion set, and the executable operation planning suggestion set is issued to an operation navigation system. According to the method, multi-modal images and hemodynamic parameters can be seamlessly integrated, accurate vascular structure extraction and lesion risk assessment are realized, and a more comprehensive and objective decision basis is provided for vascular surgical operation planning.

Embodiments of the invention relate generally to systems and methods for facilitating determining a risk of a patient for an adverse health outcome. In one embodiment, an inventive method includes using an AI-enabled volume calculator to estimate cardiac structure volume from CT scan images; and using a computer enabled risk calculator to, based at least in part on the estimated volume, determine a risk of a patient for an adverse health condition. In some embodiments, a system for detecting patients at risk of an adverse health condition includes a computer enabled volume calculator configured to facilitate assessing the volume of a cardiovascular structure based at least in part on a set of CT scan images; the system can include a computer enabled risk calculator configured to determine, based at least in part on the assessed volume, whether the patient is at risk for an adverse health condition.

According to the blood vessel blood oxygen saturation degree calculation method based on the multispectral eye fundus image, blood vessel skeleton extraction and blood vessel topological graph construction are introduced in the method, and refined description of a blood vessel structure is achieved. The blood vessel radius self-adaptive calculation is carried out based on the blood vessel topological graph, the blood vessel positioning precision is improved, and a more accurate pixel value basis is provided for subsequent blood vessel internal and external pixel value calculation. According to the method, a light intensity ratio compensation mechanism is introduced, in 548nm light density calculation, a 605nm extravascular pixel value is converted into equivalent 548nm incident light intensity through a pre-calibrated light intensity ratio, the difference of reflection characteristics of tissues to different wavelengths is eliminated, the accuracy of a light density calculation result is improved, and then the accuracy of a blood oxygen saturation degree calculation result is improved.

The invention provides a multi-modal imaging system fusing physiological parameter monitoring, and the system comprises an ultrasonic collection module which collects ultrasonic detection signals according to a preset time sequence; the physiological acquisition module is used for acquiring time sequence reflecting physiological signals according to a preset time sequence; the feature fusion module is used for carrying out time registration on the ultrasonic detection signals and the time sequence physiological signals, extracting multi-modal features and fusing the multi-modal features to form a joint feature vector; the multi-modal estimation module is used for analyzing and obtaining an imaging result and multi-modal physiological parameters of the target tissue or blood vessel based on the joint feature vector; and the output module is used for outputting the imaging result and the multi-modal physiological parameters. According to the method, ultrasonic detection signals and various time sequence physiological signals are deeply fused, so that global accurate capture of vascular structures, dynamic characteristics and physiological time sequence information is realized, and a full-link closed loop of structural imaging, functional imaging and multi-parameter continuous monitoring is successfully cut through; and a high-precision, multi-dimensional and explainable solution is provided for cardiovascular risk assessment and individualized health management.

The application provides a 4D printing method for complex branched blood vessel structure based on illumination. The printing method is to obtain a precursor solution by dissolving and mixing a compound having a ortho-nitrobenzyl four-arm polyethylene glycol and a methacrylated hyaluronic acid with PBS; a modeling software is used to create a tube model of the complex branched blood vessel, wherein the tube model comprises a Y-shaped tube model, a T-shaped tube model, a first-stage branched tube model and a second-stage branched tube model; the prepared precursor solution is put into a DLP printer tank, material printing is performed according to a design scheme based on the tube model to be printed to form a tube piece structure, the tube piece structure is completely soaked in a PBS solution to self-curl to form a complex branched blood vessel structure, and the complex branched blood vessel structure is subjected to secondary crosslinking under ultraviolet light to form a closed complex branched blood vessel structure. The printing model of the complex pipeline is designed, integrated printing is performed, the model design of the connection is decoupled by changing the radial expansion characteristics of the 4D material to realize the self-curling curvature of the complex closed pipeline, and the 4D printing of the complex closed pipeline is realized.

An ultrasound system (110) can acquire (510) ultrasound data of a vascular structure of a region of interest of a subject from an ultrasound probe (202). The ultrasound system (110) can acquire tracking data of the ultrasound probe (202) from a tracking system (120). The ultrasound system (110) can generate (530) a three-dimensional (3D) vascular model of the vascular structure using the ultrasound data and the tracking data. The ultrasound system (110) can acquire (540) angle-compensated flow information of the vascular structure using a flow imaging technique and the 3D vascular model of the vascular structure. The ultrasound system (110) can display (550) the angle-compensated flow information of the vascular structure.

The invention discloses a retina artery and vein segmentation and blood vessel marker combined prediction method based on blood vessel structure analysis, and relates to the problem that in the retina blood vessel structure analysis, an existing method is difficult to accurately extract the structural features of retina blood vessels and predict key blood vessel biomarkers at the same time. Feature changes such as retinal blood vessel width and sparse degree are important signals in the early stage of diseases. A current arteriovenous segmentation method based on deep learning is difficult to accurately extract retina arteriovenous, and retina vessel form quantization cannot be realized. In order to solve the problem, the invention provides a retina artery and vein segmentation and blood vessel marker combined prediction method based on blood vessel structure analysis. Experiments show that the method has the following advantages: (1) the accuracy of retinal artery and vein segmentation is remarkably improved by extracting blood vessel features; and (2) realizing prediction of a plurality of blood vessel biomarkers through a learnable blood vessel marker Token and a semantic alignment mechanism. The method can be used for fundus blood vessel morphology evaluation.

This invention provides a deep learning-based intelligent analysis method for cardiovascular medical images, comprising the following steps: S1, heterogeneous data fusion and acquisition of multimodal medical images and physiological signals; S2, intelligent extraction and quantitative analysis of vascular structural features based on deep learning; S3, hemodynamic parameter modeling and risk region segmentation based on fused structural features; S4, texture feature mining and intelligent identification of lesion types in high-risk areas; S5, cardiovascular disease risk stratification prediction based on multi-dimensional feature fusion; S6, federated learning optimization and closed-loop update of the model. This invention's deep learning-based intelligent analysis method for cardiovascular medical images achieves, for the first time, heterogeneous fusion of medical images, millimeter-wave radar-PPG physiological signals, and clinical molecular data, and achieves precise spatiotemporal alignment through GPS-controlled crystal oscillators, overcoming the limitations of single image analysis and providing a more comprehensive feature foundation for subsequent feature extraction and risk prediction.

The invention relates to characterization of perfusion defects. In order to characterize a perfusion defect in a vascular structure of a blood supply of an organ (5, 8) of a patient on the basis of imaging, medical imaging data (11, 12, 17) are received, the medical imaging data (11, 12, 17) containing energy-resolved CT imaging data (11, 12, 17). A blood flow occlusion object (7) in the vascular structure is detected based on the medical imaging data (11, 12, 17). A perfusion defect score (16) of a target region (6a, 6b) of the at least one organ (5, 8) is determined from the energy-resolved CT imaging data (11, 12, 17), the blood perfusion of which is potentially influenced by the occluded object (7).

The invention provides a capillary network hemodynamic modeling method with a multilayer structure, and aims to solve the problem that a traditional model is difficult to truly reflect a capillary multilayer structure and blood flow topological characteristics, and combines the hierarchical structure characteristics of a capillary network disclosed by a high-resolution imaging image. And constructing a multi-layer blood vessel network model consisting of a microarteriovenous layer, a connecting layer and a capillary network layer. Wherein the micro artery and vein layer is a tree-shaped blood vessel network formed by micro arteries and micro veins; the capillary network layer is of a regular network blood vessel structure constructed based on a hexagonal grid; and the connecting layer is a connecting blood vessel for connecting the microarteriovenous layer and the capillary network layer. Through standardized numbering rules, node insertion and blood vessel splitting operation, embedding, reconstruction and overall structure integration of blood vessel segments are completed, and therefore a capillary network system which is complete in structure and is in topological communication is constructed. On the basis of structural modeling, a linear equation solving system is established for hemodynamic simulation. The method has the advantages in the aspects of structural definition, simulation precision and expansibility, and has important scientific research and engineering application value.

The application belongs to the technical field of frozen section, and particularly relates to an animal blood vessel tissue frozen embedding and sectioning and spatial transcriptome sequencing method. In the first aspect, the application provides an animal blood vessel tissue frozen embedding method, wherein a first buffer is used to replace blood in an animal body; a blood vessel is peeled off and surrounding tissues are removed; a perfusion liquid is used to perfuse the blood vessel to obtain a perfused blood vessel, the perfusion liquid comprising OCT and a second buffer; and a plurality of perfused blood vessels are embedded in the same OCT embedding block and then rapidly frozen. The blood vessel is peeled off from the tissue and embedded alone, and a plurality of blood vessels in each embedding block are laid flat after sectioning to ensure that the chip surface is covered with sufficient blood vessel structures, thereby solving the problems of small volume of blood vessel tissue, small number of cells, small amount of RNA and difficulty in spatial transcriptome sequencing and the like.

In order to solve the problems of limited labeled data, differences between blood vessels and interference of lesion area, the present application discloses a retinal blood vessel image segmentation method based on a multi-scale dilated convolution residual network, which is still a challenge problem for accurately segmenting retinal blood vessels, especially fine blood vessels, on a retinal fundus image. The present application designs a multi-scale residual input and output module to make up for the loss of part of the blood vessel structure information due to down-sampling, combines the advantages of dilated convolution and DropBlock to relieve network overfitting and reduce the influence of the lesion area on blood vessel feature extraction, and further introduces a multi-scale mean pooling module to obtain high-level features and retain context information. Finally, by improving the way of jump connection, the dilated convolution is effectively used to improve the information transmission capacity of the jump connection. Compared with other algorithms, the present application can more accurately segment the fine blood vessels in the retinal image under complex conditions and has better robustness.

Embodiments of the present disclosure provide systems and methods for analyzing medical images containing vascular structures using a sequential model. An example system includes a communication interface configured to receive the medical image and the sequential model. The sequential model includes a vessel extraction sub-model and a lesion analysis sub-model. The vessel extraction sub-model and the lesion analysis sub-model are trained independently or jointly. The example system further includes at least one processor configured to apply the vessel extraction sub-model to the received medical image to extract location information of the vascular structure. The at least one processor also applies the lesion analysis sub-model to the received medical image and the location information extracted by the vessel extraction sub-model to obtain a lesion analysis result of the vascular structure. The at least one processor further outputs the lesion analysis result of the vascular structure.