Photon computed tomography non-electrocardiogram gating lung CT plain scan for coronary artery calcification score evaluation system

Abstract

The application discloses a photon CT non-electrocardiogram gating lung CT plain scanning coronary artery calcification score evaluation system and relates to the technical field of coronary artery calcification score evaluation.The CT data acquisition module, the coronary artery calcification identification module, the score calculation and evaluation module and the risk evaluation generation module directly count X-ray photons, eliminate electronic noise, improve the signal-to-noise ratio and the contrast-to-noise ratio, in the conventional chest low-dose CT, without electrocardiogram gating, realize the automatic processing from the low-dose lung CT plain scanning data acquisition to the structured report generation, without manual data transmission and intervention among the links, integrate the coronary artery calcification evaluation into the conventional lung CT examination process, avoid the additional burden of the patient needing to separately perform coronary artery CT scanning, improve the examination efficiency, reduce the error possibly introduced by manual operation, ensure the consistency and reliability of the evaluation result and optimize the clinical examination and diagnosis process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coronary artery calcification score evaluation, and particularly relates to a photon CT non-electrocardiogram gating lung CT plain scan coronary artery calcification score evaluation system. BACKGROUND

[0002] Coronary artery calcification is an important marker of atherosclerosis, and its score evaluation has important clinical value for coronary heart disease risk prediction. A Chinese patent application with publication number CN118864622A discloses a coronary artery calcification score evaluation model training method, evaluation method and system. The model training method comprises: obtaining a 3D chest plain scan CT image to be trained and a 3D electrocardiogram gating CT image to be trained; training a to-be-trained model in a manner of using semantic learning of the 3D electrocardiogram gating CT image to be trained to prompt regression learning of the 3D chest plain scan CT image to be trained, so as to realize coronary artery calcification score evaluation and obtain the coronary artery calcification score evaluation model. The application uses semantic learning of the electrocardiogram gating CT image to prompt the regression learning of the chest plain scan CT image, so that the network can learn the calcification semantic information of the chest plain scan CT, and the accuracy of the coronary artery calcification score is improved.

[0003] Although the above-mentioned patent improves the accuracy of the coronary artery calcification score through joint training of dual-mode data, there are still the following problems:

[0004] 1. The prior art simultaneously obtains chest plain scan CT and electrocardiogram gating CT images for model training and evaluation. In actual application, the patient needs to be scanned twice, which not only increases the examination time and equipment resource consumption, but also may cause the patient to receive a higher radiation dose;

[0005] 2. In clinical practice, low-dose lung CT plain scanning has been widely used for lung disease screening, but due to the lack of electrocardiogram gating, it is difficult to directly use it for coronary artery calcification evaluation due to coronary artery motion artifacts, image resolution limitations, and low contrast between calcification and surrounding tissues.

[0006] 3. The existing calcification recognition method has insufficient artifact filtering capability for non-gating images, and the score calculation accuracy is limited, which cannot realize one-stop examination of lung CT and coronary artery calcification evaluation, increasing the complexity of the clinical process and the burden on patients. SUMMARY

[0007] The purpose of the present application is to provide a photon CT non-electrocardiogram gating lung CT plain scan coronary artery calcification score evaluation system, which realizes automatic processing from low-dose lung CT plain scan data acquisition to structured report generation, eliminates the need for manual data transmission and intervention between each link, integrates coronary artery calcification evaluation into the conventional lung CT examination process, improves examination efficiency, ensures the consistency and reliability of the evaluation results, and optimizes the clinical examination and diagnosis process to solve the problems raised in the above background art.

[0008] To achieve the above purpose, the present application provides the following technical solutions:

[0009] The photon CT non-electrocardiogram gating lung CT plain scan coronary artery calcification score evaluation system comprises:

[0010] The CT data acquisition module is used to acquire low-dose lung CT plain scan data under the condition of no heart cycle synchronization, and process the CT plain scan data to convert it into standard medical image data;

[0011] The coronary artery calcification recognition module is used to perform image processing on the received standard medical image data, locate and segment the region where the main stem and main branches of the coronary artery are located, identify the high-density region in the segmented region, and screen effective calcification foci;

[0012] The score calculation and evaluation module is used to obtain the position information of the effective calcification foci, assign corresponding weights to the effective calcification foci according to the preset scoring standard, and obtain the coronary artery branch calcification score and the total coronary artery calcification score in combination with the actual area of each effective calcification focus;

[0013] The risk evaluation generation module is used to determine the coronary artery calcification risk level based on the coronary artery branch calcification score and the total coronary artery calcification score according to the preset risk level division rule, and generate a structured report.

[0014] Further, the process of the CT data acquisition module acquiring low-dose lung CT plain scan data comprises:

[0015] Based on the condition of no heart cycle synchronization, in combination with the preset low-dose scan parameters, the photon counting CT device is controlled to perform scanning;

[0016] The scanning range covering the main body of the whole lung and the coronary artery running area is set, and the corresponding scanning field boundary is set based on the scanning range, wherein the scanning field boundary covers the running area of the main stem and main branches of the coronary artery;

[0017] According to the preset low-dose scan parameters and the scanning field boundary, the photon counting CT device is continuously scanned to obtain original X-ray projection data;

[0018] The original X-ray projection data is processed for image reconstruction to generate low-dose lung CT tomographic image data, and then the low-dose lung CT tomographic image data is converted into standard medical image data.

[0019] Furthermore, the process of converting low-dose lung CT tomographic image data into standard medical image data includes:

[0020] The original X-ray projection data is preprocessed, and based on the preprocessed original X-ray projection data, a preset image reconstruction algorithm is used to reconstruct the tomographic image, generating low-dose lung CT plain scan tomographic image data containing information on the coronary artery course region.

[0021] The pixel grayscale information and spatial location information of the low-dose lung CT tomographic image data are extracted, and combined with patient identification information and scanning parameters, they are converted into format data that conforms to medical digital imaging and communication standards to form standard medical image format data.

[0022] Furthermore, the coronary artery calcification recognition module includes:

[0023] The image processing unit is used to process the received standard medical image format data, including reducing image noise interference through noise suppression algorithms, while enhancing the grayscale difference between high-density areas and surrounding tissues, and outputting optimized image data.

[0024] The coronary artery region segmentation unit is used to extract features from the optimized image data, locate and segment the target regions of the coronary artery trunk and main branches based on the feature extraction results, and output the image data of the target regions.

[0025] The calcification candidate identification unit is used to filter the grayscale values ​​of the image data of the target area based on the preset calcification CT value threshold, extract the high-density areas that meet the preset CT value threshold as calcification candidate areas, and associate the corresponding candidate area location and grayscale information.

[0026] The effective calcification screening unit is used to perform connected component feature analysis on candidate calcification areas, identify and exclude areas that meet the preset artifact features, screen effective calcifications, and associate the corresponding effective calcification location, actual area, and CT value.

[0027] Furthermore, the coronary artery region segmentation unit adopts a preset deep learning model, using a clinically labeled image dataset of the coronary artery trunk and major branches as training samples. It performs target region segmentation by learning the anatomical morphological features of the coronary artery. The target region image data output by the preset deep learning model is associated with the spatial coordinate information of the coronary artery trunk and major branches.

[0028] Furthermore, the integral calculation and evaluation module includes:

[0029] a weight assignment unit configured to obtain CT values associated with the effective calcification foci, compare the CT values with CT value intervals of a preset scoring standard, and assign corresponding weights to each effective calcification focus based on a comparison result;

[0030] a single-focus score calculation unit configured to obtain an actual area of the effective calcification focus, and obtain a single calcification focus score in combination with the corresponding weight assigned to the effective calcification focus;

[0031] a branch score summary unit configured to obtain positions of the effective calcification foci, match the positions of the effective calcification foci with spatial coordinates of the coronary main stem and main branches based on image data of the target region, determine the coronary branches to which each effective calcification focus belongs, and obtain calcification scores of the coronary branches based on the single calcification focus scores of the single calcification foci under the same coronary branch.

[0032] a total score calculation unit configured to obtain the calcification scores of the coronary branches, cumulatively calculate all the calcification scores of the coronary branches, and generate a total coronary calcification score.

[0033] Further, the process of determining the coronary calcification risk level according to the preset risk level division rule specifically includes:

[0034] obtaining risk assessment elements, wherein the risk assessment elements include the total coronary calcification score and distribution position information of the effective calcification foci, and the distribution position information is associated with coronary branch identifiers to which the calcification foci belong;

[0035] performing a preliminary risk level determination based on the total coronary calcification score, wherein if the total calcification score matches a first preset score interval, the preliminary determination is low risk; if the total calcification score matches a second preset score interval, the preliminary determination is low-to-medium risk; if the total calcification score matches a third preset score interval, the preliminary determination is medium-to-high risk; and if the total calcification score matches a fourth preset score interval, the preliminary determination is high risk;

[0036] dynamically adjusting the preliminary risk level in combination with the distribution position information of the effective calcification foci and a functional priority of the coronary branches, wherein if the preliminary determination is low-to-medium risk or medium-to-high risk, and the calcification foci are distributed in a preset coronary key functional branch, the preliminary risk level is raised by one risk level;

[0037] if the calcification foci are distributed in a preset coronary non-key functional branch, and the number of calcification branches in the preset coronary non-key functional branch set does not exceed a preset calcification value, the preliminary risk level is maintained, and a final coronary calcification risk level is generated by integrating the preliminary determination result and the dynamic adjustment result.

[0038] Further, the structured report includes:

[0039] Obtaining patient basic information and scanning parameters, integrating the obtained patient basic information and scanning parameters into the basic information field of the structured report;

[0040] Filling in the coronary branch score, total calcification score and determined risk level into the calcification score result and risk assessment field of the structured report;

[0041] Generating a calcification distribution diagram labeled with calcification location, area and corresponding coronary branch based on the effective calcification location information, and embedding it in the structured report;

[0042] According to the determined risk level, extracting the corresponding preset clinical suggestion template, filling in the corresponding suggestion into the clinical suggestion field of the structured report, and forming a complete structured report.

[0043] Further, the integral calculation and evaluation module further comprises a dynamic quality correction unit, which is used to correct the integral value (hereinafter referred to as the original integral value ) of the single calcification obtained by the single calcification integral calculation unit to generate a motion and noise corrected integral value , and the branch integral summary unit and the total integral calculation unit are further configured to perform subsequent integral summary and accumulation calculation based on the ;

[0044] The dynamic quality correction unit is specifically configured as:

[0045] For the first effective calcification filtered out by the coronary calcification identification module, the corresponding original integral value , original measurement area and average CT value are first obtained;

[0046] Further, the dynamic quality correction unit communicates with the image processing unit in the coronary calcification identification module to obtain the local image quality measurement parameter for the first effective calcification, and the local image quality measurement parameter at least includes:

[0047] (a) Local motion blur measurement ): By analyzing the edge profile of the first effective calcification, the full width at half maximum (Full Width at Half Maximum, FWHM) of the edge spread function (Edge Spread Function, ESF) or line spread function (Line Spread Function, LSF) is calculated, which is defined as the , for quantitatively characterizing the image blurring degree caused by cardiac motion (without cardiac cycle synchronization condition); and

[0048] (b) local background noise standard deviation (σ ): within a pre-defined neighborhood (Region of Interest, ROI) around the i -th valid calcified lesion, which excludes other high-density regions and only contains background tissues (e.g. myocardium or blood), the statistical standard deviation of CT pixel values in the neighborhood is calculated, which is defined as the σ , for quantitatively characterizing the local signal-to-noise ratio level caused by low-dose scanning;

[0049] Finally, the dynamic quality correction unit calculates the motion-and-noise-corrected integral value C :

[0050]

[0051]

[0052] wherein:

[0053] is the index of the valid calcified lesion;

[0054] is the motion-and-noise-corrected integral value (unitless) of the i -th calcified lesion;

[0055] is the original integral value (unitless) of the i -th calcified lesion, which is calculated by the single-lesion integral calculation unit based on and (and their corresponding weights);

[0056] is the motion confidence factor (unitless) of the i -th calcified lesion, which has a value range between 0 and 1;

[0057] is the noise confidence factor (unitless) of the i -th calcified lesion, which has a value range between 0 and 1;

[0058] is the pre-defined motion correction tuning coefficient (unitless), which is obtained through phantom calibration experiments or based on training of clinical data sets;

[0059] is the local motion blurring metric (unit: meter ), a characteristic scale representing the motion artifact;

[0060] is the original measured area (unit: square meter ) of the th calcified plaque, is the characteristic size (unit: meter ) of the calcified plaque;

[0061] is a very small stabilization length constant (unit: meter , for example ) to prevent the denominator from being zero;

[0062] is a preset noise correction tuning coefficient (unitless) obtained by phantom calibration experiments or training based on clinical data sets;

[0063] is the local background noise standard deviation (unit: Hounsfield unit );

[0064] is the average CT value (unit: Hounsfield unit ) of the th calcified plaque;

[0065] is a preset calcification determination CT value threshold (unit: Hounsfield unit , for example 130 ) used by the coronary calcification identification module, represents the effective intensity of the calcified plaque signal being higher than the threshold.

[0066] Further, the system further utilizes the energy spectrum resolution capability of the photon counting CT device to realize calcified plaque quantification based on physical density, rather than the traditional Agatston integral based on CT value (Hounsfield unit );

[0067] To this end, the CT data acquisition module is further configured to:

[0068] In the image reconstruction process of the original X-ray projection data, the energy resolution characteristics of the photon counting CT device are utilized to divide the collected photons according to a preset energy threshold, to generate multi-energy bin image data (Multi-Energy Bin Images) of at least two (for example, a low energy spectrum window and a high energy spectrum window) or multiple preset energy windows, and to transmit the multi-energy bin image data to the coronary calcification identification module;

[0069] Correspondingly, the coronary artery calcification identification module is configured to perform the following operations to replace the screening method based on the preset calcification determination CT value threshold:

[0070] (1) Spectral data receiving and processing: receiving multi-spectral channel image data of the region where the coronary artery trunk and main branches are located;

[0071] (2) Basis material decomposition: processing the multi-spectral channel image data by using a preset basis material decomposition algorithm; the basis material decomposition algorithm (for example, a method based on maximum likelihood estimation or projection domain decomposition) is based on a preset basis material model, and the basis material model at least includes two basis materials of hydroxyapatite (HA) and water (or equivalent soft tissue); the algorithm solves the distribution of each basis material in the image space by analyzing the characteristic attenuation difference of different materials under different spectral windows;

[0072] (3) Quantitative density map generation: the output of the basis material decomposition algorithm is a quantitative hydroxyapatite density map and a water (or equivalent soft tissue) density map; each pixel value or voxel value in the quantitative hydroxyapatite density map corresponds to a value representing the physical density of hydroxyapatite, and the unit of the value is mass / volume (for example: );

[0073] (4) Calcification focus identification based on physical density: analyzing the quantitative hydroxyapatite density map, identifying and segmenting all regions meeting the conditions based on a preset physical density threshold (for example, 50 or 100 ) and a preset minimum lesion volume or area (for example, 3 voxels or 0.5 ), as effective calcification foci defined based on physical density;

[0074] Further, the integral calculation and evaluation module is further configured to calculate a physical calcification quantitative index based on the quantitative hydroxyapatite density map and the effective calcification foci, to replace the Agatston integral based on the CT value weight and the area; the physical calcification quantitative index is, for example:

[0075] (i) Total calcium mass: by multiplying the hydroxyapatite physical density value (ρHA) in each effective calcification focus voxel by its voxel volume (Vvoxel), the total calcium mass (Mtotal) of the coronary artery is calculated: ​) are multiplied and then all the voxels of all the valid calcifications are accumulated to obtain the total coronary calcification mass in milligrams ( );

[0076] (ii) or, the coronary calcification density score: another quantitative score calculated based on the physical density value ( ) and the lesion volume ( );

[0077] Finally, the risk assessment generation module is configured to determine the coronary calcification risk level based on the physical calcification quantitative indicator (such as the total coronary calcification mass) and the preset risk classification rule (such as the threshold based on the total mass), and generate a structured report containing the physical density quantitative result.

[0078] Compared with the prior art, the beneficial effects of the present application are:

[0079] The present application counts X-ray photons directly through the CT data acquisition module, coronary calcification identification module, integral calculation and evaluation module, and risk assessment generation module, eliminates electronic noise, and improves the signal-to-noise ratio and contrast-to-noise ratio. In the conventional chest low-dose CT, no electrocardiogram gating is required, and the automatic processing from low-dose lung CT plain scan data acquisition to structured report generation is realized, without the need for manual data transmission and intervention between each link. The coronary calcification evaluation is integrated into the conventional lung CT examination process, avoiding the additional burden of separate coronary CT scanning for the patient, improving the examination efficiency; reducing the errors that may be introduced by manual operation, ensuring the consistency and reliability of the evaluation results, and optimizing the clinical examination and diagnosis process. BRIEF DESCRIPTION OF DRAWINGS

[0080] Figure 1 The present application is a photon CT non-electrocardiogram gated lung CT plain scan coronary calcification integral evaluation system module diagram. DETAILED DESCRIPTION

[0081] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0082] To solve the technical problems that the prior art needs to be scanned twice, and the low-dose lung CT plain scan lacks electrocardiogram gating, has artifacts, resolution and contrast problems, the existing calcification identification method is insufficient for artifact filtering of non-gated images, and cannot realize one-stop evaluation, increasing the complexity of the clinical process and the burden of the patient, please refer toFigure 1 The embodiment provides the following technical solutions:

[0083] The photon CT non-electrocardiogram gating lung CT plain scanning coronary artery calcification score evaluation system comprises:

[0084] The CT data acquisition module is configured to control the photon counting CT to cover the full lung scanning range containing the coronary artery running region with preset low-dose scanning parameters under the condition of no cardiac cycle synchronization, acquire low-dose lung CT plain scanning data, and process the CT plain scanning data to convert the CT plain scanning data into standard medical image data;

[0085] The coronary artery calcification identification module is configured to perform image processing on the received standard medical image data to enhance the gray scale contrast of the coronary artery and calcification foci, locate and segment the region where the main stem and main branches of the coronary artery are located by using a deep learning model, identify the high-density region in the segmented region based on a preset calcification judgment CT value threshold, exclude isolated regions meeting the preset artifact characteristics by connected domain analysis, and screen effective calcification foci.

[0086] The integral calculation and evaluation module is configured to obtain the position information of the effective calcification foci, assign corresponding weights to the effective calcification foci according to a preset scoring standard, obtain the integral of each calcification focus in combination with the actual area of each effective calcification focus, accumulate the integral of each calcification focus under the same coronary branch to obtain the coronary branch calcification integral, and accumulate the coronary branch calcification integrals to obtain the total coronary calcification integral.

[0087] The risk evaluation generation module is configured to determine the coronary calcification risk level according to a preset risk level division rule based on the coronary branch calcification integrals and the total coronary calcification integral, and generate a structured report containing basic information, integral results, risk evaluation and clinical suggestions, extract patient basic information and CT preset low-dose scanning parameters, generate a calcification focus distribution diagram in combination with the position information of the effective calcification foci, and integrate the integral data, risk level, calcification focus distribution diagram and preset clinical suggestions.

[0088] In the embodiment, the preset low-dose scanning parameters are as follows: the tube voltage is controlled in the range of 80-120 kV, the tube current is controlled in the range of 50-100 mA, the layer thickness is set to 0.625-1.25 mm, and the pitch is adjusted to 1.0-1.5, so that the radiation dose is consistent with that of conventional low-dose lung CT plain scanning, that is, 1-3 mSv, without the need to additionally increase the radiation exposure of the patient.

[0089] In the embodiment, the preset scoring standard is evaluated according to the Agatston scoring standard.

[0090] In the embodiment, the X-ray photons are directly counted by the CT data acquisition module structure to eliminate electronic noise, and a non-electrocardiogram-gated continuous scanning design is combined, so that when the preset low-dose scanning parameters are used to acquire the lung CT plain scan data, the image quality is ensured without increasing the radiation dose, the signal-to-noise ratio and the contrast-to-noise ratio of the low-dose image are effectively improved, the dependence of the electrocardiogram gating on the heart rhythm of the patient is avoided, and the scope of the applicable population for the coronary calcification screening is expanded. The coronary calcification identification module automatically locates and segments the coronary trunk and main branch regions by using a deep learning model to exclude the interference of non-coronary structures, extracts the high-density regions based on a preset calcification judgment CT value threshold, and excludes isolated artifacts by using a connected domain analysis, so that the process of locating and identifying the calcification foci does not need manual participation, the artifact interference problem caused by slight heart motion in the non-electrocardiogram-gated scanning is solved, the effective calcification foci are accurately locked, and the accuracy and stability of the calcification foci identification are significantly improved. The integral calculation and evaluation module combines the weight assignment and branch integral accumulation structure designed based on the Agatston score standard to calculate the coronary branch calcification integral and the total calcification integral, respectively. The risk assessment generation module adjusts the risk level based on the total integral and the functional priority of the coronary branch to which the calcification foci belong, so that the risk assessment result is more suitable for the actual influence of the coronary branch calcification on the heart function, more targeted assessment basis is provided for the clinic, and the clinical guidance value of the coronary calcification risk assessment is improved.

[0091] In the embodiment, the process of acquiring the low-dose lung CT plain scan data by the CT data acquisition module includes:

[0092] Based on the non-heart motion cycle synchronization condition and the preset low-dose scanning parameters, the photon counting CT device is controlled to perform scanning;

[0093] The scanning range covering the main lung and the coronary running area is set, and the corresponding scanning field boundary is set based on the scanning range, wherein the scanning field boundary covers the running area of the coronary trunk and main branches.

[0094] The photon counting CT device is continuously scanned according to the preset low-dose scanning parameters and the scanning field boundary to obtain the original X-ray projection data.

[0095] The original X-ray projection data is subjected to image reconstruction processing to generate low-dose lung CT plain scan tomographic image data, and the low-dose lung CT plain scan tomographic image data is converted into standard medical image data.

[0096] In the embodiment, the process of converting the low-dose lung CT plain scan tomographic image data into standard medical image data includes:

[0097] The original X-ray projection data is preprocessed, including detector response correction, scatter interference suppression and noise filtering of the projection data, to eliminate system errors and interference signals in the data acquisition process; based on the preprocessed original X-ray projection data, a preset image reconstruction algorithm is used to reconstruct the tomographic image, to generate low-dose lung CT plain tomographic image data containing coronary artery running region information, the tomographic image data corresponding to the gray scale information of the anatomical structure within the scan field boundary;

[0098] The pixel gray scale information and spatial position information in the low-dose lung CT plain tomographic image data are extracted, combined with patient identification information and scan parameters, and converted into format data conforming to the Digital Imaging and Communications in Medicine standard, to form standard medical image format data, for subsequent calcification recognition processing.

[0099] In the embodiment, the scan can be started without relying on electrocardiogram signal synchronization, avoiding the limitation of patient heart rhythm in the existing electrocardiogram gating coronary artery scan, and the photon counting CT eliminates electronic noise through the hardware characteristic of directly counting X-ray photons, without the need to compensate for image quality by increasing scan parameters, to ensure that the coronary artery region details can still be retained under low-dose scanning, the scan field boundary is set through geometric constraint, the coronary artery calcification evaluation is directly integrated into the conventional lung CT examination process, and the clinical examination efficiency is greatly improved; through the combined operation of detector response correction, scatter interference suppression and noise filtering, the hardware noise reduction advantage of the photon counting CT is formed in cooperation, to further eliminate system errors and interference signals in the data acquisition process, to strengthen the gray scale contrast of the coronary artery and calcification, to avoid the structure detail loss problem caused by the existing single noise reduction processing, in the image reconstruction stage, the preset image reconstruction algorithm is guided by the geometric constraint to perform tomographic reconstruction according to the anatomical geometric relationship within the scan field boundary, to avoid the dislocation of the anatomical structure in the reconstructed image, to ensure the geometric accuracy of the coronary artery region in the tomographic image, to extract the pixel gray scale information and spatial position information, and to integrate the patient identification information and scan parameters, to ensure that the image data carries complete clinical correlation information.

[0100] In the embodiment, the coronary artery calcification recognition module comprises:

[0101] The image processing unit is configured to process the received standard medical image format data, to enhance the gray scale contrast of the coronary artery and calcification, including reducing image noise interference through a noise suppression algorithm, while strengthening the gray scale difference between the high-density region and the surrounding tissue, and outputting the optimized image data;

[0102] The coronary artery region segmentation unit is configured to extract features of the optimized image data using a preset deep learning model, to locate and segment the target region of the coronary artery trunk and main branches based on the feature extraction result, to exclude the interference of non-coronary structures such as lung tissue and sternum, and to output the image data of the target region;

[0103] A calcification candidate recognition unit is configured to perform gray value screening on the image data of the target region based on a preset calcification determination CT value threshold, extract a high-density region meeting the preset CT value threshold as a calcification candidate region, and associate the corresponding candidate region position and gray information;

[0104] An effective calcification screening unit is configured to perform connected domain feature analysis on the calcification candidate region, identify and exclude regions meeting preset artifact features, retain high-density regions consistent with the coronary artery anatomical course and meeting physiological structure connectivity, screen effective calcification, and associate the corresponding effective calcification position, actual area, and CT value; the preset artifact features include: isolated distribution and mismatch with the coronary artery course direction, area less than a preset minimum area threshold, and connectivity not meeting the coronary artery anatomical structure.

[0105] In this embodiment, the coronary artery region segmentation unit adopts a preset deep learning model, the deep learning model is a U-Net++ model, a clinically labeled image data set of the coronary artery trunk and main branches is used as a training sample, the target region is segmented by learning the coronary artery anatomical features, including the course direction, diameter change, branch connection relationship, etc., the target region image data output by the preset deep learning model is associated with the spatial coordinate information of the coronary artery trunk and main branches, and accurate segmentation of the coronary artery region is realized.

[0106] In this embodiment, the image processing unit performs image processing through an image processor, reduces noise interference of low-dose standard medical image format data by means of a noise suppression algorithm, and simultaneously enhances the gray difference between the high-density region and the surrounding tissue by means of a gray enhancement algorithm, the coronary artery region segmentation unit adopts a deep learning model and combines geometric constraint calibration of the target region spatial coordinates to accurately exclude non-coronary structure interference such as lung tissue and sternum, thereby defining a reliable target range for calcification recognition; the calcification candidate recognition unit extracts a high-density region based on a preset calcification determination CT value threshold and associates the position and gray information of the target region, thereby avoiding the problem of false positives caused by indiscriminate screening of the whole lung image and reducing subsequent invalid power consumption; the effective calcification screening unit excludes artifact regions with isolated distribution, mismatched course direction, and excessively small area by means of connected domain feature analysis combined with the coronary artery anatomical course and physiological structure constraint, retains effective calcification regions meeting physiological features and associates the position, actual area, and CT value, thereby ensuring the accuracy of calcification recognition and providing reliable data for the subsequent integral calculation and evaluation module.

[0107] In this embodiment, the integral calculation and evaluation module includes:

[0108] The weight assignment unit is configured to obtain CT values associated with the effective calcification foci, compare the CT values with CT value intervals of a preset scoring standard based on the preset scoring standard, and assign corresponding weights to each effective calcification focus based on a comparison result;

[0109] The single-focus score calculation unit is configured to obtain an actual area of the effective calcification focus, and obtain a score value of a single calcification focus in combination with the corresponding weight assigned to the effective calcification focus;

[0110] The branch score summary unit is configured to obtain a position of the effective calcification focus, obtain spatial coordinates of a coronary main stem and main branches based on image data of the target region, match the position of the effective calcification focus with the spatial coordinates, and determine a coronary branch to which each effective calcification focus belongs; and obtain a calcification score of each coronary branch based on the score value of the single calcification focus under the same coronary branch.

[0111] The total score calculation unit is configured to obtain the calcification scores of the coronary branches, cumulatively calculate all the calcification scores of the coronary branches, and generate a total coronary calcification score.

[0112] In this embodiment, in the preset scoring standard, the corresponding relationship between the weight and the CT value interval is that the first weight is assigned when the CT value is in a first preset interval, the second weight is assigned when the CT value is in a second preset interval, the third weight is assigned when the CT value is in a third preset interval, and the fourth weight is assigned when the CT value is in a fourth preset interval; the CT values of the first to fourth preset intervals are sequentially increased, and the values of the first to fourth weights are sequentially increased.

[0113] In this embodiment, according to the total calcification score, a corresponding grade is automatically matched, wherein 0 points of the total score is no calcification, which is low risk, 1-100 points is mild calcification, which is low-medium risk, 101-400 points is moderate calcification, which is high-medium risk, and >400 points is severe calcification, which is high risk.

[0114] In this embodiment, the process of determining the coronary calcification risk grade according to the preset risk grade division rule specifically includes:

[0115] The risk assessment element includes the total coronary calcification score and distribution position information of the effective calcification focus, the distribution position information is associated with a coronary branch identifier to which the calcification focus belongs, and includes a left main stem, a left anterior descending branch, a left circumflex branch, and a right coronary artery.

[0116] performing a preliminary risk level determination based on the total coronary calcification score, if the total calcification score matches a first preset score interval, the preliminary determination is low risk, no calcification; if the total calcification score matches a second preset score interval, the preliminary determination is low to moderate risk, mild calcification; if the total calcification score matches a third preset score interval, the preliminary determination is moderate to high risk, moderate calcification; if the total calcification score matches a fourth preset score interval, the preliminary determination is high risk, severe calcification;

[0117] combined with the effective calcification focus distribution position information and the coronary branch function priority, the preliminary risk level is dynamically adjusted, if the preliminary determination is low to moderate risk or moderate to high risk, and the calcification focus is distributed in a preset coronary key function branch, such as left main or left anterior descending branch, the preliminary risk level is adjusted by one risk level, the low to moderate risk is adjusted to moderate risk, and the moderate to high risk is adjusted to high risk;

[0118] if the calcification focus is distributed in a preset coronary non-key function branch, such as right coronary artery or left circumflex branch, and the number of calcification branches in the preset coronary non-key function branch set does not exceed the preset calcification value, the preliminary risk level is maintained, and the preliminary determination result and the dynamic adjustment result are integrated to generate the final coronary calcification risk level;

[0119] In this embodiment, the structured report includes:

[0120] obtain the patient's basic information and CT preset low dose scanning parameters, such as tube voltage, tube current, layer thickness, pitch, etc., integrate the obtained patient's basic information and scanning parameters into the basic information field of the structured report, the patient's basic information includes name, gender, age, examination ID;

[0121] fill in the calcification score result and risk assessment field of the structured report corresponding to each coronary branch score, total calcification score and determined risk level;

[0122] generate a calcification focus distribution diagram labeled with calcification focus position, area and belonging coronary branch based on the effective calcification focus position information, and embed it in the structured report;

[0123] According to the determined risk level, the corresponding preset clinical suggestion template is extracted, including: low risk suggestion regular follow-up, low to moderate to moderate risk suggestion combined with blood lipid, blood pressure and other indicators for comprehensive evaluation, high risk suggestion further coronary CT angiography examination, fill in the corresponding suggestion into the clinical suggestion field of the structured report to form a complete structured report.

[0124] In the embodiment, the integral calculation and evaluation module generates the branch calcification integral and the total coronary calcification integral, accurately presents the calcification degree of each coronary branch, completes the preliminary risk classification based on the matching result of the total coronary calcification integral and the preset integral interval, adjusts the risk level in combination with the coronary branch identifier and the branch function priority of the effective calcification focus distribution, makes the risk level determination result more suitable for the actual heart health risk condition in clinic, improves the clinical guidance value of risk assessment, and the structured report can enable medical staff to quickly obtain complete coronary calcification evaluation data and adaptive diagnosis and treatment guidelines, reduce information screening and interpretation time, and improve the accuracy of clinical diagnosis efficiency and subsequent diagnosis and treatment scheme.

[0125] In the embodiment, the integral calculation and evaluation module further includes a dynamic quality correction unit, which is configured to correct the integral value (hereinafter referred to as the original integral value ) of the single calcification focus obtained by the single focus integral calculation unit to generate a motion and noise corrected integral value , and the branch integral aggregation unit and the total integral calculation unit are further configured to perform subsequent integral aggregation and accumulation calculation based on the ;

[0126] The dynamic quality correction unit is specifically configured as:

[0127] For the first effective calcification focus filtered out by the coronary calcification identification module, the original integral value , the original measurement area and the average CT value are first obtained;

[0128] Further, the dynamic quality correction unit communicates with the image processing unit in the coronary calcification identification module to obtain the local image quality measurement parameter of the first effective calcification focus, and the local image quality measurement parameter at least includes:

[0129] (a) Local motion blur measurement : by analyzing the edge profile of the first effective calcification focus, the full width at half maximum (Full Width at Half Maximum, FWHM) of the edge spread function (Edge Spread Function, ESF) or line spread function (Line Spread Function, LSF) thereof is calculated, which is defined as the , which is used to quantitatively represent the degree of image blur caused by heart motion (without heart cycle synchronization condition); and

[0130] (b) local background noise standard deviation (σ ): the statistical standard deviation of CT pixel values within a pre-defined ROI around the i -th valid calcified lesion, which excludes other high-density regions and only contains background tissue (e.g. myocardium or blood), is calculated as the local background noise standard deviation σ , which is used to quantitatively characterize the local signal-to-noise ratio level caused by low-dose scanning;

[0131] Finally, the dynamic quality correction unit calculates the motion and noise corrected Agatston score A :

[0132]

[0133]

[0134] wherein:

[0135] is the index of the valid calcified lesion;

[0136] is the motion and noise corrected Agatston score (unitless) of the i -th calcified lesion;

[0137] is the original Agatston score (unitless) of the i -th calcified lesion, which is calculated by the single lesion score calculation unit based on σ and w (i) and their corresponding weights;

[0138] is the motion confidence factor (unitless) of the i -th calcified lesion, which ranges from 0 to 1;

[0139] is the noise confidence factor (unitless) of the i -th calcified lesion, which ranges from 0 to 1;

[0140] is the pre-defined motion correction tuning coefficient (unitless), which is obtained through phantom calibration experiments or based on training of clinical data sets;

[0141] is the local motion blur metric (unit: meter ), which characterizes the characteristic scale of motion artifacts;

[0142] is the i Original measured area of ​​each calcification foci (unit: square meters) ), Corresponding characteristic dimensions of calcification foci (unit: meters) );

[0143] It is a very small stabilized length constant (unit: meter) ,For example (), used to prevent the denominator from being zero;

[0144] The preset noise correction tuning coefficients (dimensionless) are obtained through phantom calibration experiments or training based on clinical datasets.

[0145] The standard deviation of the local background noise (unit: Heinz units) );

[0146] For the first Average CT value of each calcification (unit: Henle units) );

[0147] The preset CT value threshold for calcification determination used by the coronary artery calcification identification module (unit: Henle units) For example, 130 ), The effective intensity of the calcification signal above the threshold.

[0148] The integral calculation and evaluation module further includes a dynamic quality correction unit, used to adjust the original integral value of a single calcified foci obtained by the single-foci integral calculation unit. Quantization correction is performed to address motion blur introduced by non-ECG gating and integration inaccuracies caused by high noise introduced by low doses. This unit first obtains the first... Three key local image quality metrics for an effective calcification: the first is the local motion blur metric ( The parameter is obtained by the image processing unit for the first... The edge contours of each calcification foci are sampled at high frequency. The edge spread function (ESF) is calculated and differentiated to obtain the line spread function (LSF) of that edge. Finally, the full width at half maximum (FWHM) of the LSF is taken (unit: millimeters). As The value of is the physical characterization of the image blur scale caused by motion artifacts; the second is the standard deviation of local background noise ( The parameter is obtained as follows: the image processing unit in the first... Within a small region of interest (ROI) adjacent to each calcification and pre-defined as background tissue (such as myocardium or blood with CT values ​​in the range of 20-60 HU), calculate the statistical standard deviation (in Henle units) of the CT values ​​of all pixels. This value is then assigned as The first is used to characterize the local signal-to-noise ratio level; the second is the original measured area of ​​the calcification foci ( ). (Unit: square millimeters) ) and average CT value ( )(unit: All parameters are provided by the calcification candidate identification unit. After acquiring the parameters, the dynamic quality correction unit executes a coronary calcification integral dynamic correction formula. ,in This is the integral value after motion and noise correction. The predefined CT value threshold for calcification determination (e.g., 130) ), and The dimensionless correction tuning coefficients are determined in advance through phantom calibration experiments. For a very small stabilized length constant (e.g.) This is to prevent the denominator from being zero. This formula uses "relative ambiguity" (…). ) and "relative noise level" To dynamically calculate two confidence factors ( and ), thus affecting the original integral After adjusting the discount, the final result is It is transmitted to the branch integration unit for subsequent more accurate integration and risk assessment.

[0149] For example, in a non-ECG-gated low-dose lung CT scan, the system identified a valid calcification foci in the right coronary artery (RCA) (index). The single-burner integral calculation unit is based on its original measured area (); ) = 1.1 and average CT value ( ) = 175 The original integral value was calculated. ) = 1.3; at this time, the dynamic mass correction unit is activated to evaluate the credibility of this integral value; first, the image processing unit analyzes the edge of the calcification, due to the significant RCA motion caused by the beating heart, the calculated LSF is wider, and its FWHM is determined as the local motion blur metric (M) ) = 0.7 ; at the same time, the unit defines a 5x5 pixel background ROI in the pericardial fat area (CT value -50HU) beside the calcification, due to the low-dose scan, the CT value of this area fluctuates dramatically, and the calculated local background noise standard deviation (N ) = 30 ; the system uses a preset CT value threshold (T ) = 130 , and a calibrated correction tuning coefficient (C and ; the correction unit substitutes the formula to calculate: the motion confidence factor (C ); the noise confidence factor (C ); finally, the motion and noise corrected integral value (I ) = = ; this significantly reduced corrected integral (0.693) is passed to the branch integral aggregation unit, which more truly reflects the actual integral contribution of the calcification under the influence of motion and noise pollution, avoiding the overestimation of integral and misjudgment of risk level caused by artifacts.

[0150] In this embodiment, the system further utilizes the spectral resolution capability of the photon counting CT device to realize calcification quantification based on physical density, rather than the traditional Agatston integral based on CT value (Hounsfield Unit );

[0151] To this end, the CT data acquisition module is further configured to:

[0152] In the image reconstruction process of the original X-ray projection data, the energy resolution characteristic of the photon counting CT device is utilized to divide the collected photons according to a preset energy threshold, to generate multi-energy bin image data (Multi-Energy Bin Images) of at least two (for example, a low energy spectrum window and a high energy spectrum window) or multiple preset energy windows, and to transmit the multi-energy bin image data to the coronary calcification identification module;

[0153] Correspondingly, the coronary calcification identification module is configured to perform the following operations to replace the screening method based on the preset calcification determination CT value threshold:

[0154] (1) Spectral data reception and processing: receiving multi-spectral channel image data of the region where the coronary main stem and main branches are located;

[0155] (2) Basis material decomposition: processing the multi-spectral channel image data using a preset basis material decomposition algorithm; the basis material decomposition algorithm (for example, a method based on maximum likelihood estimation or projection domain decomposition) is based on a preset basis material model, which includes at least two basis materials of hydroxyapatite (HA) and water (or equivalent soft tissue); the algorithm solves the distribution of each basis material in the image space by analyzing the characteristic attenuation difference of different materials under different spectral windows;

[0156] (3) Quantitative density map generation: the output of the basis material decomposition algorithm is a quantitative hydroxyapatite density map and a water (or equivalent soft tissue) density map; each pixel value or voxel value in the quantitative hydroxyapatite density map corresponds to a value representing the physical density of hydroxyapatite, and the unit of the value is mass / volume (for example: );

[0157] (4) Calcification lesion identification based on physical density: analyzing the quantitative hydroxyapatite density map, identifying and segmenting all regions meeting the conditions based on a preset physical density threshold (for example, 50 or 100 ) and a preset minimum lesion volume or area (for example, 3 voxels or 0.5 ), as effective calcification lesions defined based on physical density;

[0158] Furthermore, the integral calculation evaluation module is further configured to calculate a physical calcification quantitative index based on the quantitative hydroxyapatite density map and the effective calcification lesions, to replace the Agatston integral based on CT value weight and area; the physical calcification quantitative index is, for example:

[0159] (i) Total coronary calcification mass: by multiplying the hydroxyapatite physical density value (ρ) in each effective calcification lesion voxel by its voxel volume (V), and then accumulating all voxels of all effective calcification lesions, the total coronary calcification mass in milligrams (mg) is obtained;

[0160] ​​​(ii) or, Coronary Calcium Density Score: other quantitative score based on the physical density value (ρ) ) and lesion volume (V) ) weighted calculation;

[0161] Finally, the risk assessment generation module is configured to determine the coronary calcium risk level based on the physical calcium quantitative indicator (e.g. total coronary calcium mass) and preset risk classification rules (e.g. total mass-based threshold), and generate a structured report containing the physical density quantitative results.

[0162] The system utilizes the energy spectrum resolution capability of the photon counting CT device to perform a physical density-based rather than CT value (HU)-based calcification lesion quantification method to fundamentally replace the Agatston score method based on CT value threshold and weight. To this end, the CT data acquisition module is configured to: during scanning, utilize the hardware function of the photon counting detector (PCD) to sort the received X-ray photons into at least two (or multiple) preset energy windows (Energy Bins) (e.g. low energy window: 20-50 keV; high energy window: 50-140 keV) in real time according to their energy, and generate multiple sets of parallel multi-energy spectrum channel image data based on the data reconstruction of each energy window. Subsequently, the coronary calcium identification module receives these multi-energy spectrum image data and applies a basis material decomposition algorithm (Basis Material Decomposition Algorithm). The working principle of this algorithm is: for each voxel in the image, the algorithm is based on a preset two-basis-material model (Two-Basis-Material Model) containing at least water (Water) and hydroxyapatite (Hydroxyapatite, HA); the goal of the algorithm is to solve the two unknown quantities of the physical density of the two basis materials in the voxel (Hydroxyapatite Density) and (Water Density) (both in or ); to achieve the solution, the algorithm first reads two measurement values of the voxel from the multi-energy spectrum channel image data: one is the low-energy window linear attenuation coefficient value (unit ), and the other is the high-energy window linear attenuation coefficient value (unit ); then, the algorithm solves and by solving a linear equation system composed of two equations:

[0163]

[0164]

[0165] In this system of equations, there are four mass attenuation coefficients—that is... (HA coefficient in low energy window) (Coefficient of water in low-energy window) (The coefficient of HA in the high-energy window) and (Coefficients of water in the high-energy window) – all are known physical constants (units) These constant values ​​were obtained by querying authoritative physics databases such as NIST XCOM to obtain the values ​​of water and hydroxyapatite at different energy points. The value, combined with the spectral response function of the CT system detector, is integrated and averaged within low-energy and high-energy windows (e.g., 20-50 keV and 50-140 keV), and finally pre-loaded into the system as a system calibration parameter; because and These are voxel measurements, four. The coefficients are known constants of the system, therefore the system of equations can be uniquely solved, thus yielding the solution. and These two unknown physical density values; the algorithm's output is not a CT (HU) image, but two (or more) quantitative density maps, including a hydroxyapatite quantitative density map (HA Map) with voxel values ​​in mass / volume units (e.g., Finally, the integral calculation and evaluation module performs quantification based on this HA Map: it uses a physical density threshold (e.g., 50). ) to replace 130 A threshold is set to identify effective calcification foci; and a quantitative indicator of physical calcification, such as total coronary calcification mass, is calculated by taking the density values ​​of all voxels in the HA Map that are above the physical density threshold. ) and its voxel volume ( ,For example Multiply by , and then sum the mass values ​​of all voxels to obtain the result in milligrams (mg). The total calcification mass, expressed in units of 1, is used by the risk assessment generation module to determine the risk level.

[0166] For example, in a single photon CT lung scan, the CT data acquisition module is set to dual energy window mode (low energy window 20-50 keV, high energy window 50-140 keV), two sets of image data are reconstructed and output; the coronary calcification identification module starts the base material decomposition algorithm; for a specific voxel in the coronary region, the algorithm reads the linear attenuation coefficient value of the voxel in the low energy window image , and reads the linear attenuation coefficient value of the voxel in the high energy window image ; the algorithm retrieves the preset mass attenuation coefficient value from the system calibration library (these values have been calculated by NIST XCOM data and system spectral integration): 、 、 、 ; the algorithm solves the equations (1) and (2) , and obtains two unknown densities of the voxel: (i.e. 150 ) and ; the algorithm repeats this process for all voxels to generate a hydroxyapatite quantitative density map (HA Map); then, the integral calculation evaluation module applies a physical density threshold of 50 to analyze the HA Map, and finds that the 150 density value of the voxel is higher than the threshold, and the voxel is determined to be calcified; the module further identifies that a total of 2500 voxels belong to the calcification, and the voxel volume of this scan is known ; the module calculates the total mass of the coronary calcification of the calcification by accumulating the (density volume) value of all 2500 calcified voxels, and the total mass of the coronary calcification of the calcification is 12.2 ; the physical index (12.2 ) is sent to the risk assessment generation module to generate a structured report that is independent of CT (HU) value, has more explicit physical meaning, and is more robust to motion and low dose artifacts.

[0167] The above merely describes the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can make equivalent replacements or changes to the technical solutions and inventive concepts of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A photon CT non-electrocardiogram-gated lung CT scan for coronary artery calcification score evaluation system, characterized in that, include: The CT data acquisition module is used to acquire low-dose lung CT plain scan data under conditions of no cardiac cycle synchronization, and to process the CT plain scan data and convert it into standard medical image data. The coronary artery calcification identification module is used to process the received standard medical image data, locate and segment the area where the main coronary artery trunk and major branches are located, identify high-density areas within the segmented area, and screen for effective calcification foci. The integral calculation and evaluation module is used to obtain the location information of effective calcifications, assign corresponding weights to effective calcifications according to preset scoring criteria, obtain the integral value of each effective calcification based on its actual area, perform motion and noise correction, and obtain the calcification integral of each coronary artery branch and the total coronary artery calcification integral based on the corrected integral value. The correction includes: For the screened out first effective calcification focus, its corresponding original integral value , original measurement area and average CT value are obtained; acquiring a local image quality metric parameter for the first effective calcification focus, the local image quality metric parameter comprising at least: Local motion blur metric : by analyzing the edge profile of the first effective calcified lesion, calculating a full width at half maximum of an edge spread function or a line spread function of the edge profile, the full width at half maximum being defined as the , for quantitatively characterizing a degree of image blur caused by a motion of the heart. local background noise standard deviation : around each of the effective calcified lesions, a statistical standard deviation of CT pixel values within a pre-defined neighborhood around the lesion is calculated, which is defined as the local signal-to-noise ratio, for quantitatively characterizing the local signal-to-noise level resulting from the low-dose scan; Based on the preset coronary calcification score dynamic correction formula, the motion and noise corrected score value is calculated : wherein: is the index of the effective calcification; is the motion and noise corrected integral value of the th calcification; is the original integral value of the th calcification, calculated based on and and their corresponding weights; is the motion confidence factor of the th calcification; is the noise confidence factor of the th calcification; is the preset motion correction tuning coefficient, obtained by phantom calibration experiment or based on clinical dataset training; is the local motion blur metric, representing the characteristic scale of motion artifact; is the original measured area of the th calcification, corresponding to the characteristic size of the calcification; is the stabilization length constant; is the preset noise correction tuning coefficient, obtained by phantom calibration experiment or based on clinical dataset training; is the local background noise standard deviation; is the average CT value of the th calcification; is the preset calcification determination CT value threshold adopted by the coronary calcification identification module, representing the effective intensity of the calcification signal being higher than the threshold. The risk assessment generation module is used to determine the coronary artery calcification risk level based on the calcification scores of each coronary artery branch and the total coronary artery calcification score, according to preset risk level classification rules, and generate a structured report.

2. The photon CT non-electrocardiogram-gated lung CT plain scan versus coronary artery calcification score evaluation system of claim 1, wherein, The process of acquiring low-dose lung CT plain scan data by the CT data acquisition module includes: Based on the condition of no cardiac cycle synchronization, combined with preset low-dose scanning parameters, the photon counting CT equipment is controlled to perform scanning. A scanning range is set with the whole lung as the main body and covering the coronary artery course area. The corresponding scanning field boundary is set based on the scanning range, wherein the scanning field boundary covers the course area of ​​the main coronary artery trunk and major branches. According to the preset low-dose scanning parameters and scanning field boundaries, the photon counting CT equipment is used to continuously scan and obtain the raw X-ray projection data. The original X-ray projection data is processed for image reconstruction to generate low-dose lung CT tomographic image data, and then the low-dose lung CT tomographic image data is converted into standard medical image data. 3.The photon CT non-ECG-gated lung CT scan to coronary artery calcification score evaluation system of claim 2, wherein, The process of converting low-dose lung CT tomographic image data into standard medical image data includes: The original X-ray projection data is preprocessed, and based on the preprocessed original X-ray projection data, a preset image reconstruction algorithm is used to reconstruct the tomographic image, generating low-dose lung CT plain scan tomographic image data containing information on the coronary artery course region. The pixel grayscale information and spatial location information of the low-dose lung CT tomographic image data are extracted, and combined with patient identification information and scanning parameters, they are converted into format data that conforms to medical digital imaging and communication standards to form standard medical image format data.

4. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score evaluation system of claim 3, wherein, The coronary artery calcification recognition module includes: The image processing unit is used to process the received standard medical image format data, including reducing image noise interference through noise suppression algorithms, while enhancing the grayscale difference between high-density areas and surrounding tissues, and outputting optimized image data. The coronary artery region segmentation unit is used to extract features from the optimized image data, locate and segment the target regions of the coronary artery trunk and main branches based on the feature extraction results, and output the image data of the target regions. The calcification candidate identification unit is configured to perform gray value screening on the image data of the target region based on a preset calcification determination CT value threshold, extract a high-density region meeting the preset CT value threshold as a calcification candidate region, and associate corresponding candidate region position and gray information; The effective calcification focus screening unit is configured to perform connected domain feature analysis on the calcification candidate region, identify and exclude regions meeting a preset artifact feature, screen effective calcification foci, and associate corresponding effective calcification focus positions, actual areas, and CT values.

5. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score evaluation system of claim 4, wherein, The coronary region segmentation unit adopts a preset deep learning model, takes a clinically labeled image data set of coronary trunks and main branches as training samples, performs target region segmentation by learning coronary anatomical feature, and associates the target region image data output by the preset deep learning model with spatial coordinate information of the coronary trunks and main branches.

6. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score evaluation system of claim 5, wherein, The integral calculation and evaluation module includes: The weight assignment unit is configured to obtain the CT value associated with the effective calcification focus, compare the CT value with a CT value interval of a preset scoring standard based on the preset scoring standard, and assign a corresponding weight to each effective calcification focus based on the comparison result; The single-focus integral calculation unit is configured to obtain the actual area of the effective calcification focus, and obtain the integral value of a single calcification focus in combination with the corresponding weight assigned to the effective calcification focus; The branch integral aggregation unit is configured to obtain the position of the effective calcification focus, obtain the spatial coordinates of the coronary trunks and main branches based on the image data of the target region, match the position of the effective calcification focus with the spatial coordinates, determine the coronary branch to which each effective calcification focus belongs, and obtain the calcification integral of each coronary branch based on the integral value of a single calcification focus under the same coronary branch. The total integral calculation unit is configured to obtain the calcification integral of each coronary branch, cumulatively calculate all the calcification integrals of the coronary branches, and generate a coronary total calcification integral.

7. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score evaluation system of claim 6, wherein, The process of determining the coronary calcification risk level according to the preset risk level division rule specifically includes: Obtain risk assessment elements, including the coronary total calcification integral and the distribution position information of the effective calcification focus, and the distribution position information is associated with the coronary branch identifier to which the calcification focus belongs; Perform preliminary risk level determination based on the coronary total calcification integral. If the total calcification integral matches a first preset integral interval, it is preliminarily determined as low risk. If the total calcification integral matches a second preset integral interval, it is preliminarily determined as low-medium risk. If the total calcification integral matches a third preset integral interval, it is preliminarily determined as medium-high risk. If the total calcification integral matches a fourth preset integral interval, it is preliminarily determined as high risk. Adjust the preliminary risk level dynamically in combination with the distribution position information of the effective calcification focus and the functional priority of the coronary branch. If the preliminary determination is low-medium risk or medium-high risk, and the calcification focus is distributed in a preset coronary key functional branch, the preliminary risk level is adjusted up by one risk level. If the calcification focus is distributed in a preset coronary non-key functional branch, and the number of calcified branches in the preset coronary non-key functional branch set does not exceed a preset calcification value, the preliminary risk level is maintained, and the preliminary determination result and the dynamic adjustment result are integrated to generate a final coronary calcification risk level.

8. The photon CT non-electrocardiogram-gated lung CT scan to coronary artery calcification score evaluation system of claim 7, wherein, The structured report comprises: Obtaining patient basic information and scanning parameters, and integrating the obtained patient basic information and scanning parameters into the basic information field of the structured report; Filling in the calcification score results and risk assessment field of the structured report according to the scores of each coronary branch, the total calcification score and the determined risk level; Generating a calcification distribution diagram labeled with the positions, areas and coronary branches of the calcification foci based on the position information of the effective calcification foci, and embedding the diagram in the structured report; According to the determined risk level, extracting the corresponding preset clinical suggestion template, filling in the corresponding suggestions into the clinical suggestion field of the structured report, and forming a complete structured report.

9. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score assessment system of claim 6, wherein, The integral calculation evaluation module further comprises a dynamic quality correction unit, which is configured to correct the integral value of the single calcification focus obtained by the single-focus integral calculation unit, i.e. the original integral value to generate a motion and noise corrected integral value The branch integral aggregation unit and the total integral calculation unit are further configured to perform subsequent integral aggregation and accumulation calculation based on the ​ The dynamic quality correction unit communicates with an image processing unit in the coronary calcification identification module to obtain a local image quality metric parameter for the first effective calcification focus.

10. The photon CT non-electrocardiogram gating pulmonary CT scan to coronary artery calcification score assessment system of claim 6, wherein, The system further utilizes the energy spectrum resolution capability of the photon counting CT device to realize calcification quantification based on physical density instead of the traditional Agatston score based on CT value; The CT data acquisition module is further configured to: During the image reconstruction process of the original X-ray projection data, the energy resolution characteristic of the photon counting CT device is utilized to divide the collected photons according to a preset energy threshold, to generate multi-energy spectrum channel image data of at least two or more preset energy windows, and to transmit the multi-energy spectrum channel image data to the coronary calcification identification module; Correspondingly, the coronary calcification identification module is configured to perform the following operations to replace the screening method based on the preset calcification determination CT value threshold: Receiving the multi-energy spectrum channel image data of the regions where the coronary trunks and main branches are located; Using a preset base material decomposition algorithm to process the multi-energy spectrum channel image data; the base material decomposition algorithm is based on a preset base material model, and the base material model at least includes hydroxyapatite and water as two base materials; The output of the base material decomposition algorithm is a hydroxyapatite quantitative density map and a water density map; each pixel value or voxel value in the hydroxyapatite quantitative density map corresponds to a value representing the physical density of hydroxyapatite; Analyzing the hydroxyapatite quantitative density map, identifying and segmenting all regions meeting the conditions based on a preset physical density threshold and a preset minimum lesion volume or area as effective calcification foci defined based on physical density; And the integral calculation and evaluation module is further configured to calculate a physical calcification quantitative index based on the hydroxyapatite quantitative density map and the effective calcification foci, to replace the Agatston score based on CT value weight and area; Finally, the risk assessment generation module is configured to determine the coronary calcification risk level based on the physical calcification quantitative index and a preset risk division rule, and to generate a structured report containing the physical density quantitative result.

Images