Automatic identification method and system for OCT image blood vessel structure

Abstract

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.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of blood vessel structure recognition, in particular to an automatic recognition method and system for OCT image blood vessel structure. BACKGROUND

[0002] In the field of medical image analysis of blood vessel structure, optical coherence tomography (OCT) technology has realized the visualization of the three-layer structure (intima, media, and adventitia) of blood vessels due to its high-resolution imaging capability, providing an important image basis for the diagnosis and research of blood vessel diseases.

[0003] However, the existing OCT technology still highly depends on the naked eye observation and subjective judgment of professionals in the process of identifying the three-layer structure of blood vessels. Due to the differences in experience level, observation angle, and judgment standard of different operators, it is easy to cause significant individual differences in the structure determination results of the same blood vessel image, making it difficult to form a unified recognition standard. The errors and inconsistencies caused by human factors directly result in a significant reduction in the accuracy of blood vessel structure analysis results, which may lead to blurred lesion positioning, biased lesion severity assessment, or even delayed treatment opportunities or inappropriate treatment decisions due to misjudgment.

[0004] Even though there are some systems or algorithms that assist in recognition, they still face core bottlenecks in actual application. On the one hand, these systems or algorithms lack sufficient accuracy in extracting blood vessel edge features, making it difficult to accurately distinguish the subtle boundaries of the three-layer structure; on the other hand, they lack standardized recognition logic and still rely on personnel intervention for result correction, making it difficult to achieve truly precise and unified recognition and unable to meet the high-precision and high-consistency requirements of clinical blood vessel structure analysis. SUMMARY

[0005] The present application aims to provide an automatic recognition method and system for OCT image blood vessel structure to achieve automatic recognition of OCT image blood vessel structure and solve the problems of inaccurate and non-uniform recognition and easy errors in the prior art.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical solution, an automatic recognition method for OCT image blood vessel structure, comprising the following steps: Step 1: Blood vessel layer boundary feature extraction: extracting blood vessel layer boundary features including geometric features and physical features from OCT images based on multi-scale Hessian matrix response to form a feature response map; Step 2: Three-layer topological structure recognition: constructing the topological relationship of blood vessel structure according to the feature response map and extracting the nested ring structure; identifying and outputting the three-layer initial boundary coordinate point set and structure score of the intima, media, and adventitia of blood vessels; Step 3, blood vessel structure optimization: accurate optimization is performed on the initial boundary coordinate point set, and special cells are detected in the optimized region to generate a result report.

[0007] Meanwhile, the present scheme also provides an automatic identification system for OCT image blood vessel structure, which is applied to the automatic identification method for OCT image blood vessel structure and comprises: An image input module is configured to access an OCT device and acquire OCT image information; A preprocessing module is configured to pre-process the image information, suppress noise information, and enhance structure boundaries; A feature recognition module is configured to recognize features from the OCT image and extract blood vessel layer boundary features to form a feature response map; A boundary determination module is configured to determine blood vessel three-layer structure boundaries based on the extracted features, calibrate the boundaries, and identify pathological parameters; An output module is configured to output the identification results in a visual form.

[0008] The principle and advantages of the present scheme are as follows: Traditional OCT blood vessel analysis techniques mainly adopt image segmentation algorithms, and such a method is heavily dependent on the continuity of image gray scale or the saliency of gradient. When there are calcification shadows, blood artifacts, noise or local boundary blurring in the blood vessel wall, these methods will completely fail due to the interruption or distortion of pixel information, and a complete and coherent structure cannot be obtained. Moreover, the existing methods are only pure image processing tools, and the output results may be correct in geometry but completely unreasonable in anatomy, and it is difficult to integrate the medical prior knowledge that the blood vessel is one layer wrapped in one layer with a reasonable thickness range into the analysis process.

[0009] Although the method based on the deep learning black box model may be effective in some cases, its decision-making process is not traceable. Doctors cannot trust a system that cannot explain why a certain area is determined as a boundary, which greatly hinders the application of the system in clinical practice and becomes a technical prejudice that is difficult to break through.

[0010] The present scheme breaks out of the traditional "image segmentation" mindset and no longer regards the problem as "distinguishing foreground and background pixels", but upgrades it to "identifying and verifying a complex topological structure that meets specific anatomical rules in a noisy environment". The present scheme converts the visual diagnostic experience and anatomical knowledge of clinicians into a calculable mathematical model, and realizes the automatic, high-precision and standardized identification of the blood vessel three-layer structure in the OCT image through a three-order progressive analysis framework of "feature enhancement-topological identification-constraint optimization".

[0011] The present scheme is based on the tubular geometry of blood vessels to extract features, rather than relying solely on gray scale gradients. Through multi-scale analysis, different sizes of blood vessels are adapted, and through algorithm, the noise response of non-blood vessel edges is suppressed, the recognition accuracy is improved, and the subsequent accurate quantitative calculation is ensured. Secondly, by introducing topological analysis, it is no longer dependent on pixel-by-pixel continuity, but identifies the existence and persistence of "ring-shaped cavities" in the image. Even if the boundary is locally missing, as long as its topological structure exists stably, the system can automatically infer and complete the complete contour, ensuring that reliable structures can be output in complex lesion blood vessels. Thirdly, the doctor's experience (such as interlayer spacing, angle, area ratio) is encoded as a mathematical constraint condition. The optimization process not only pursues image feature matching, but also requires the output result to fall within the anatomically reasonable parameter space, ensuring that the output result is credible in biomechanics and anatomy, solving the problem that traditional methods are prone to output absurd results. And each decision has a clear mathematical and physical basis, and the decision-making process of any boundary point can be traced back, which is easy for clinicians to accept and adopt, improving the efficiency and credibility of the decision, and realizing the automatic recognition of OCT blood vessel structure, making the diagnosis process more standardized and unified. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 A flowchart of the automatic recognition method for OCT image blood vessel structure of the present application; Figure 2 An existing OCT blood vessel image in the automatic recognition method for OCT image blood vessel structure of the present application; Figure 3 A flowchart of step 1 in the automatic recognition method for OCT image blood vessel structure of the present application; Figure 4 A blood vessel structure cross-sectional view in the automatic recognition method for OCT image blood vessel structure of the present application; Figure 5 A schematic diagram of constructing a blood vessel structure distance field in step 2 of the automatic recognition method for OCT image blood vessel structure of the present application; Figure 6 A three-layer blood vessel schematic diagram for constructing a distance field in the automatic recognition method for OCT image blood vessel structure of the present application; Figure 7 A structural schematic diagram of the automatic recognition system for OCT image blood vessel structure of the present application. DETAILED DESCRIPTION

[0013] The following will be further described in detail through specific embodiments: Example 1 The automatic recognition method for OCT image blood vessel structure in the embodiment converts anatomical prior knowledge (layer thickness range) into mathematical constraint conditions, and defines pathological quantification parameters, so as to realize automatic recognition of blood vessel structure, improve recognition accuracy and uniformity, and reduce diagnostic risk.

[0014] In the embodiment, as shown in the accompanying drawings, Figure 1 the following steps are included: S1, blood vessel layer boundary feature extraction: based on multi-scale Hessian matrix response, blood vessel layer boundary features are extracted from the OCT image, including geometric features and physical features, to form a feature response map.

[0015] In the embodiment, since the blood vessel cross section presents an approximately elliptical tubular structure in the OCT image, its features mainly include continuity along the blood vessel direction, symmetric gradient change perpendicular to the blood vessel direction, and local curvature characteristics, which are extracted as geometric features.

[0016] As shown in the accompanying drawings, Figure 2 in the OCT image, the blood vessel layer presents a ring with alternating light and dark. The scattering characteristic difference between the blood vessel wall and the blood / outer membrane tissue causes the blood vessel wall area to have a higher gray value, that is, strong scattering, the lumen has a lower gray value, that is, weak scattering, and there is an obvious gray transition at the layer boundary, so the physical features are extracted.

[0017] Based on the original OCT image, the extracted tubular structure features are quantified. In the embodiment, as shown in the accompanying drawings, Figure 3 the tubular geometric characteristics and gray gradient characteristics of the blood vessel layer boundary are used to perform Gaussian filtering on the OCT image, to construct a multi-scale space and enhance the blood vessel boundary features. Traditional OCT blood vessel recognition mainly relies on gradient information, and has limited effect on fuzzy boundaries between layers. Therefore, in the embodiment, the Hessian matrix is used to quantify the local structure curvature, and the minimum eigenvalue is extracted to respond to the tubular structure, combined with the gradient attenuation term to suppress non-blood vessel edge noise, so that the brightness of the blood vessel layer boundary is significantly enhanced, the interlayer boundary of blood vessels of different diameters is enhanced, multi-scale fusion is realized, the blood vessel structure recognition accuracy is improved, and different diameter blood vessels can be adapted. The extracted feature response map fusion can be expressed as: ; In the formula, is the two-dimensional space coordinate of the pixel point in the image; is a scale parameter (blood vessel size); is the Hessian matrix under the scale to detect the local curvature; is the tubular enhancement calculation, represents the minimum eigenvalue of the matrix, and the greater the negative value, the more obvious the tubular structure is; is the gradient suppression calculation; is the OCT image after Gaussian filtering, ; is the OCT image original image information, denoted as ; is the attenuation coefficient.

[0018] The tubular structure of the blood vessel is reflected by the geometric feature, and the direction feature of the blood vessel is determined to indicate the normal direction of the blood vessel wall. The boundary strength of the blood vessel is reflected by the physical feature, and it is ensured that the found tubular structure is a real boundary. At the same time, through multi-scale analysis, the scale feature can also be obtained to determine the scale data with the strongest response of the tubular feature at each position, so as to obtain the approximate thickness information of the blood vessel structure at the position.

[0019] The above feature information is fused into a feature response map to clearly show the boundary position of the potential blood vessel layer.

[0020] S2, three-layer topological structure identification: according to the feature response map, the topological relationship of the blood vessel structure is constructed, and the nested ring structure is extracted; the three-layer initial boundary coordinate point set and the structure score of the intima, media and adventitia of the blood vessel are identified and output.

[0021] In the traditional method, when the local fracture of the blood vessel wall is encountered, the integrity of the construction of the blood vessel ring structure will be invalid, which leads to the fact that the existing method is difficult to guarantee the integrity and accuracy when the blood vessel structure is established. In order to break through this problem, we get out of the fixed thinking and return to the initial three-layer structure of the normal blood vessel as three nested rings, and find that the real blood vessel layer will form a "hollow" in the image, as shown in the shadow part of the attached Figure 4 , and the artifacts or noises cannot form a stable hollow. Therefore, the present scheme creatively solves the problem of the integrity of the blood vessel ring structure by using the topological theory, penetrates the invalid interference in a homotopy manner, solves the misjudgment problem of the traditional segmentation method in the blood vessel fracture and noise interference, and thus constructs a complete and accurate blood vessel topological structure.

[0022] In this embodiment, the topological relationship of the blood vessel structure is constructed by distance transformation, the nested ring structure with significant persistence is extracted, and the initial boundary coordinate point set and the structure integrity score of the three-layer structure are preliminarily confirmed.

[0023] Specifically, the following sub-steps are included: S2.1, constructing a blood vessel structure distance field model.

[0024] In this embodiment, combined with the attached Figure 2As shown, instead of directly processing grayscale images, it transforms them into a topographic map through topological structure. In this map, vessel boundaries (bright rings) are transformed into "valleys," and vessel lumens and interlayer regions are transformed into "peaks." The "cavities" described in this scheme are "peaks" surrounded by "valleys." In this embodiment, the feature response map generated by S1 has significantly enhanced the vessel boundary information (boundary response map), as shown in the attached figure. Figure 5 As shown, the high-brightness (white) areas in the image correspond to the vessel boundaries (such as the internal limiting membrane, media, and external membrane), while the low-brightness (black) areas correspond to the vessel lumen and background tissue.

[0025] Subsequently, the characteristic response Figure Two Value-based mapping is used, setting the blood vessel boundary region as the target. Based on this, the Euclidean distance from each pixel in the image to the nearest blood vessel boundary point is calculated, generating a distance field topographic map similar to "elevation". Combined with the attached... Figure 6 As shown (boundaries unclear), in this topographic map, distance values ​​are mapped to altitude. The distance values ​​are smallest at the vessel boundary itself and in its adjacent areas, even reaching 0, and are represented as "valleys." Conversely, areas far from the vessel boundary (such as the center of the vessel lumen or the center of the adventitia) have the largest distance values, and are represented as "peaks." This transforms geometric features into structural relationships. The distance field can then be calculated as follows: ; In the formula, For a binary boundary point set, when the distance field The larger the value, the closer it is to the center of the blood vessel. =0 indicates the location of the blood vessel boundary.

[0026] S2.2 Perform continuous homology analysis on the model to screen out the void feature group and calculate the void persistence.

[0027] In this embodiment, assuming there is a complete vascular layer (bright ring) in the original OCT image, its range field should resemble a "mountain peak surrounded by a ring valley". A water level threshold is set. The algorithm increments from 0 to the maximum value, performing a "water level" scan. When the water level is low (e.g., at the threshold), the algorithm detects the water level. =2), at this point, most of the mountain peak is above water and may be connected to other surrounding areas, not forming an isolated island. When the water level rises (e.g., to the threshold), If the water level is ≤5, then the area ≤5 is submerged, and the originally connected peaks are completely separated by a ring-shaped valley (transformed from a bright ring), forming an independent "island." This is the birth of a "cavity." As the water level continues to rise (e.g., the threshold),... =8), meaning that the area ≤8 is submerged, the original island is reduced to just a tiny top, and eventually sinks completely underwater, at which point the "hole" disappears.

[0028] Then the binarized region can be represented as: ; For each The 1-dimensional homology group is calculated to detect the "hole" structure in the image and record the generation and disappearance of the hole, i.e. when increases, if a new hole is generated, then is recorded, indicating the "water level" at which the vascular layer begins to appear; when the hole disappears, then is recorded, indicating the water level at which the vascular layer is completely submerged, thereby generating the hole feature group, represented as: .

[0029] And the hole persistence is calculated as: By determining the duration of the persistent hole, the structural stability is quantified as an index of the structural stability of the vascular layer, and finally the hole feature set is generated.

[0030] S2.3, based on the persistence, the three-layer structure is screened, and the initial boundary coordinate point set is generated.

[0031] Since the islands formed by the real vascular layer can remain independent in larger water level changes, i.e. is larger, while the small fluctuations formed by noise will be quickly submerged. According to the calculated persistence, the holes that satisfy the anatomical constraints are screened to form the nested relationship of the three-layer structure. In this embodiment, when > the persistence threshold T, it represents a stable hole that exists for a long time, which is the true vascular layer; when < the persistence threshold T, it represents a short-lived hole that is quickly submerged, which is a false noise. The hole set formed by the true vascular layer is selected, and the nested relationship is constructed, so that the hole set can be represented as: ; In the formula, is the topological persistence threshold, which can be set in combination with actual experience, such as within 0-10, = 5. That is, through the persistence of the "i-th hole", it is determined whether the overall structure of this hole is stable and significant enough to determine the structural stability. Even if the original image has local blur, complete, continuous and anatomically correct boundaries can be obtained, solving the problem of recognition failure caused by local breakage.

[0032] The holes screened out in the hole set are arranged in descending order of hole area, and the sorted holes are checked to verify the extracted nested relationship. For each valid hole, the average radius position is taken as the initial coordinate value of the boundary, i.e. ; then the boundary coordinate point set of the ith layer can be expressed as .

[0033] The boundary coordinate point set of each layer of film is an ordered, discrete two-dimensional coordinate point sequence, and the connection of each point sequence represents a closed contour line. The boundary of each layer is a list composed of N points, and each point is a (x, y) coordinate pair. For example, the inner membrane boundary point set can be expressed as: = { (x 11 , y 11 ), (x 12 , y 12 ),..., (x 1n , y 1n )} ; the middle membrane boundary point set can be expressed as: = { (x 21 , y 21 ), (x 22 , y 22 ),..., (x 2n , y 2n )} ; and the outer membrane boundary point set can be expressed as: = { (x 31 , y 31 ), (x 32 , y 32 ),..., (x 3n , y 3n )}.

[0034] The initial boundary coordinate point set of the three-layer structure can be expressed as: ; in which, is the inner membrane boundary; is the middle membrane boundary; is the outer membrane boundary.

[0035] The confirmed structure is also scored according to the calculated persistence, and the score is calculated as: ; in which, is the inner membrane boundary structure score; is the middle membrane boundary structure score; is the outer membrane boundary structure score.

[0036] In this embodiment, the blurred image that is difficult to judge by the naked eye is converted into a quantitative graph with clear boundaries and clear labels through cavity persistence judgment, which can solve the problems of segmentation interruption, layer structure loss, misjudgment as an independent structure, false boundary and layer number statistical error caused by local fracture of blood vessels, calcified plaque shielding, blood turbulence artifacts and branch vessel interference in traditional methods, and ensure that the complete ring can be identified, short-lived cavities are automatically filtered, and the identification and accurate establishment of nested topological relationship are improved.

[0037] S3, vascular structure optimization: accurate optimization of the initial boundary coordinate point set, and detection of special cells in the optimized area to generate a result report.

[0038] In this embodiment, after smoothing according to the calculated initial boundary coordinate point set, three-layer boundary initial contours are established, represented as , which represent the boundary curves of the intima, media and adventitia, respectively. At the same time, the initial boundary coordinates are used as the starting point for optimization, the structure score is used to set the optimization weight, and the energy function parameters are initialized, with weight coefficients of ; and constraint condition parameters are set.

[0039] An energy function is constructed as follows: ; In the formula, is the area uniformity term (consistency); is the interlayer distance constraint term; is the boundary smoothing term.

[0040] wherein, is the area uniformity weight, which can be adjusted according to the persistence; is the three-layer boundary to be optimized; is the gray variance in the kth layer area; is the penalty function, which increases when the interlayer distance deviates from the normal range; is the average distance between the ith layer and the jth layer; is the total boundary length, which is used to control the boundary smoothing.

[0041] In this embodiment, the constraint condition parameters are set according to the anatomical constraint conditions, so the constraint conditions include the allowed range of adjacent membrane distance, the maximum allowed value of the adjacent layer boundary normal vector angle, and the adjacent layer area ratio range. Specifically, it can be represented as: ; In the formula, is the allowed range of intima to media distance; is the allowed range of media to adventitia distance, such as 1.0㎛~15.0㎛; is the angle between the normal vectors of adjacent layer boundaries; is the maximum allowed value of the angle between the normal vectors, such as 15°, to ensure that the adjacent layer boundaries have similar orientations; is the area ratio of adjacent layers; is the area ratio range, such as 0.7-1.3, to prevent abnormal expansion or contraction.

[0042] The gradient descent method or the constraint optimization algorithm is used to minimize the energy function , while satisfying the constraint conditions, and iterated to convergence for optimization.

[0043] In the optimized area of each layer, special cells (i.e. abnormal scattering cells: an abnormal cell population different from normal vascular smooth muscle cells inferred based on OCT image features, which are usually closely related to the occurrence, development and vulnerability of vascular lesions) are detected. In this embodiment, the optimized boundary is used to divide the blood vessel into three layers, and then the abnormality of each pixel in each region is calculated, which can be represented as: In the formula, , are the mean and standard deviation of the gray value in the k-th layer region. At the same time, the threshold value is set according to the fluctuation range of the gray value of normal cells, which can be taken as 2.5-3.0, and when > 0 , the position is marked as a special cell.

[0044] Finally, the optimized three-layer boundary coordinates are output, including the thickness, area and other parameters of each layer, and the marked special cell position set is output, which can be represented as .

[0045] By optimizing the boundary coordinates, the boundary can be more clear and smooth, and by constraint optimization, the structure can meet the anatomical rules, and the interlayer distance is stable within the normal range. It can also quickly detect special cells, improve detection accuracy and effectiveness, and improve overall automation efficiency.

[0046] Example Two In this embodiment, an automatic recognition system for OCT image blood vessel structure is provided, which is applied to the above-mentioned automatic recognition method for OCT image blood vessel structure, as shown in the accompanying drawings, which comprises: Figure 7 An image input module is used to access the OCT device and obtain OCT image information.

[0047] ​​In this embodiment, the image input module is connected to the OCT imaging device, which can analyze and compatible with different manufacturers' device data protocol, realize the lossless acquisition and high-speed transmission of the original image data. At the same time, the module will automatically extract and bind the metadata information of each frame of image, including but not limited to acquisition time, spatial resolution, device parameters and patient information, to provide accurate scale calibration and traceability basis for all subsequent processing steps, and ensure the metrological accuracy of the analysis results.

[0048] The preprocessing module is used for preprocessing the image information, suppressing noise information, and enhancing the structure boundary.

[0049] In the original OCT data, interference factors such as tissue scattering noise, optical artifacts and blood residue signals can seriously mask valuable microstructures. In this embodiment, the preprocessing module can realize anisotropic filtering, which can adaptively adjust the filtering strength according to the local content of the image. Strong smoothing is applied in the uniform area to effectively suppress noise, while weak smoothing is maintained in the key structure area such as the blood vessel layer boundary to protect its sharpness and details. After preprocessing, the signal-to-noise ratio of the image can be obviously improved to provide more accurate image architecture for subsequent feature recognition and extraction, rather than simply performing blur processing.

[0050] The feature recognition module is used for identifying features from the OCT image and extracting the blood vessel layer boundary feature to form a feature response map.

[0051] In this embodiment, the feature recognition module converts the preprocessed gray-scale image into a feature map highlighting the geometric characteristics of the blood vessel wall. Based on the fact that the cross-section of the blood vessel presents a unique tubular or layered structure in the OCT image, these structures can be accurately quantified by calculating their local curvature in mathematics. At the same time, the module uses multi-scale analysis to deal with blood vessels of different sizes, i.e. using small analysis scale to capture the fine intima boundary and using large analysis scale to perceive the thick adventitia structure. Finally, a feature response map is identified and extracted, in which the brightness value of each pixel no longer represents the gray scale, but the possibility or significance of the position belonging to the blood vessel layer boundary, so that the blood vessel structure is clearly highlighted from the complex background.

[0052] The boundary determination module is used for determining the blood vessel three-layer structure boundary based on the extracted features and performing calibration; and identifying pathological parameters.

[0053] The boundary determination module receives the extracted feature response map and completes the cross-decision from pixel features to anatomical structures to clinical parameters.

[0054] In this embodiment, the boundary determination module includes a structure layer determination unit and a parameter extraction unit. The structure layer determination unit is used to extract structural features and verify the formation of clear three-layer structure boundaries. The parameter extraction unit is used to identify special cells in each formed structure region and mark them after pathological parameter identification.

[0055] The structure layer determination unit uses topological analysis principles, no longer considering the vascular layer as an isolated edge pixel set, but identifying it as a nested, complete ring structure (including intima, media, and adventitia). This approach has strong anti-interference ability, and even if the blood vessel boundary is partially broken due to calcification or plaque obstruction, the system can infer the complete and anatomically correct contour based on the overall structure, ensuring the accuracy and integrity of the structure.

[0056] After obtaining the initial contour, the initial contour is also optimized with constraints. Clinical knowledge such as typical thickness range, relative position relationship, and smoothness between layers is converted into mathematical constraint conditions to fine-tune the initial contour and obtain a sub-pixel accuracy final boundary that best meets the biomechanical characteristics. On this basis, the parameter extraction unit automatically performs quantitative analysis to calculate a series of key pathological parameters such as layer thickness, thickness variability, lumen area, and plaque burden, and based on image features such as scattering characteristics, automatically identifies and labels suspected lipid-rich necrotic cores, microcalcification, and other vulnerable plaque features.

[0057] The output module is used to output the recognition results in a visual form.

[0058] In this embodiment, the output module converts complex analysis results into an intuitive, reliable, and directly usable form for clinical decision-making. By generating a structured comprehensive report, it not only includes visual images with three-layer boundary and special cell annotations (highlighted with different colors), but also integrates all quantitative parameters and structure integrity scores.

[0059] In this embodiment, the report design conforms to the clinical reading habits, allowing doctors to perform interactive review, such as clicking on any position on the image to display the exact thickness value at that position, or focusing on viewing the marked special cells. Finally, all images, data, and reports can be archived or transmitted to the hospital information system (HIS / PACS) in accordance with medical data standards, seamlessly integrating into the clinical workflow.

[0060] In this embodiment, an automated and standardized analysis pipeline from raw data to clinical insights is constructed. Through the core algorithm innovation of the feature recognition module and the boundary determination module, the doctor's visual diagnostic experience (such as recognizing tubular structures, inferring complete contours, and judging according to anatomical knowledge) is encoded into a computable mathematical model, and finally precise and quantitative objective evidence is provided through the output module. It completely changes the traditional mode of OCT blood vessel evaluation relying on "naked eye observation" and "manual measurement" of doctors, greatly improves the analysis efficiency, repeatability and accuracy, and provides strong data support for the precise diagnosis, treatment planning (such as stent size selection) and postoperative evaluation of vascular diseases.

[0061] The above is only an embodiment of the present application, and the specific technical solutions and / or common knowledge of the scheme are not described in detail. It should be pointed out that for those skilled in the art, without departing from the technical solutions of the present application, a number of modifications and improvements can be made, which should also be considered as the protection scope of the present application, and these will not affect the effect and practicality of the present application. The protection scope claimed in this application should be subject to the content of its claims, and the specific implementation mode and the like recorded in the specification can be used to explain the content of the claims.

Claims

1. An automated method for identifying vascular structures in OCT imaging, characterized in that, Includes the following steps, Step 1, Extraction of vascular layer boundary features: Extract vascular layer boundary features from OCT images based on multi-scale Hessian matrix response, including geometric and physical features, to form a feature response map; Step 2, Three-layer topology recognition: Construct the topological relationship of the vascular structure based on the feature response map, and extract the nested ring structure; identify and output the initial boundary coordinate point set and structure score of the three layers of the vascular intima, media and adventitia; Step 3, Vascular Structure Optimization: The initial boundary coordinate point set is precisely optimized, and special cells are detected in the optimized region to generate a result report.

2. The automated identification method for vascular structures in OCT imaging according to claim 1, characterized in that, The feature response map is represented as follows: ； In the formula, These are the two-dimensional spatial coordinates of pixels in the image; For scale parameters; For tubular reinforcement calculations, For scale The Hessian matrix below, Represents the smallest eigenvalue of a matrix; For gradient suppression calculation; This is the OCT image after Gaussian filtering; This is the attenuation coefficient.

3. The automated identification method for vascular structures in OCT imaging according to claim 1, characterized in that, Step 2 includes the following sub-steps: Step 2.1: Construct a distance field model of the vascular structure, transforming geometric features into structural relationships; Step 2.2: Perform continuous cohomology analysis on the model, screen out the void feature group, and calculate the void persistence; Step 2.3: Perform three-layer structure filtering based on persistence and generate an initial set of boundary coordinate points.

4. The automated identification method for vascular structures in OCT imaging according to claim 1, characterized in that: In step 3, an energy function is constructed based on the structural score. Combined with the set constraint parameters, the initial boundary coordinate point set is optimized to obtain the precise boundary profile. The energy function is expressed as follows: ； In the formula, For the region uniform term; This refers to the interlayer spacing constraint term; This is the boundary smoothing term.

5. The automated identification method for vascular structures in OCT imaging according to claim 1, characterized in that: In step 3, detecting special cells involves calculating the anomaly score for each pixel within each optimized region. A pixel with an anomaly score greater than a threshold is marked as a special cell. The anomaly score is expressed as... ； In the formula, This refers to the original image information of the OCT image; , denoted as the mean and standard deviation of grayscale values ​​within the k-th layer region, respectively.

6. The automated identification method for vascular structures in OCT imaging according to claim 3, characterized in that: In step 2.2, a threshold is set. Increasing from 0 to the maximum value, the void feature group is filtered out by recording the generation and disappearance states of voids, and is represented as... ； The persistence of voids is then represented as In the formula, Threshold for generating voids; The threshold for void disappearance.

7. The automated identification method for vascular structures in OCT imaging according to claim 6, characterized in that: In step 2.3, based on the persistence combined with anatomical constraints, voids that meet the conditions are selected, forming a nested relationship of three layers. The void set can then be represented as... ； In the formula, This is the topology persistence threshold.

8. The automated identification method for vascular structures in OCT imaging according to claim 7, characterized in that: It also includes sorting the voids in the void set in descending order of void area, and checking the sorted voids; for each valid void, the average radius position is taken as the initial boundary coordinate value to generate an initial boundary coordinate point set.

9. The automated identification method for vascular structures in OCT imaging according to claim 8, characterized in that: In step 2, the confirmed structure is scored based on the calculated durability, denoted as follows: ； In the formula, Score the intima boundary structure; Score the boundary structure of the middle membrane; Scoring of the outer membrane boundary structure.

10. An automated system for recognizing vascular structures in OCT imaging, characterized in that, The automated identification method for vascular structures in OCT imaging, applied to any one of claims 1-9, includes: The image input module is used to connect to OCT devices and acquire OCT image information. The preprocessing module is used to preprocess image information, suppress noise, and enhance structural boundaries; The feature recognition module is used to identify features from OCT images and extract vascular layer boundary features to form a feature response map; The boundary determination module is used to determine and calibrate the boundaries of the three-layered vascular structure based on the extracted features; and to identify pathological parameters. The output module is used to output the recognition results in a visual form.

Images