Brain image-based hemorrhagic area identification method and related equipment

Abstract

The invention discloses a hemorrhagic area identification method based on a brain image and related equipment, and aims to solve the problem that the existing hemorrhagic area identification method is low in identification accuracy. The method comprises the following steps: acquiring a dynamic multi-phase brain CT scanning image and clinical physiological parameters of a patient; based on the dynamic multi-phase brain CT scanning image, skull stripping preprocessing is carried out to extract a brain tissue area image; dividing the brain tissue area image into a plurality of anatomical partition images based on a standard Talairach coordinate system; acquiring image statistical characteristics, hemodynamic parameters and enhancement parameters of each anatomical partition image; inputting the clinical physiological parameters, the hemodynamic parameters and the enhancement parameters into the dynamic threshold prediction model, and generating a CT value distribution reference interval corresponding to each anatomical partition image; and marking a bleeding area in the dynamic multi-phase brain CT scanning image based on the CT value distribution reference interval, and outputting an image processing result containing position information of the bleeding area.

Description

Technical Field

[0001] This application relates to the field of medical imaging technology, and in particular to a method and related equipment for identifying hemorrhage areas based on brain images. Background Technology

[0002] Intracranial hemorrhage is a common neurological emergency characterized by rapid onset and progression. If not diagnosed and treated promptly, it can lead to serious consequences and even death. Non-contrast computed tomography (CT) of the head has become the preferred imaging modality for screening for intracranial hemorrhage due to its advantages of rapid acquisition, high resolution, and sensitivity to bleeding.

[0003] In clinical practice and existing related methods, the global threshold method is usually used for preliminary identification of hemorrhage areas. This method is based on the high density (i.e., high CT value) imaging feature of hemorrhage areas in CT images. By setting a fixed CT value threshold range (e.g., 40–80 Hounsfield (HU)), it distinguishes hemorrhage tissue areas from normal brain tissue areas.

[0004] However, existing methods for detecting hemorrhagic tissue regions using fixed CT value threshold ranges fail to adequately consider differences in tissue density across different brain regions and physiological differences among patients, which can easily lead to misidentification and low accuracy. Summary of the Invention

[0005] This application provides a method for identifying hemorrhage regions based on brain images, which solves the problem that existing methods for detecting hemorrhage regions by using a fixed CT value threshold range fail to fully consider the differences in tissue density in different brain regions and the physiological differences between patients, easily leading to misidentification and thus low identification accuracy.

[0006] This application also provides a device for identifying hemorrhage areas based on brain images, an electronic device, a computer-readable storage medium, and a computer program product.

[0007] The embodiments of this application adopt the following technical solutions: In a first aspect, this application provides a method for identifying hemorrhage regions based on brain imaging, comprising: To obtain dynamic multiphase brain CT scan images and clinical physiological parameters of patients; Preprocessing of cranial dissection based on dynamic multi-phase brain CT scan images to extract brain tissue region images; The brain tissue region images are divided into multiple anatomical partition images based on the standard Talairach coordinate system; Image statistical features, hemodynamic parameters, and enhancement parameters were obtained for each anatomical region. The hemodynamic parameters included the CT value rise slope in the arterial phase, the CT value fall rate in the venous phase, and the vascular impedance index. The enhancement parameters included cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor. Clinical physiological parameters, hemodynamic parameters, and enhancement parameters are input into the dynamic threshold prediction model to generate the corresponding CT value distribution reference interval for each anatomical region image. Based on the CT value distribution reference interval, the hemorrhage area in the dynamic multi-phase brain CT scan image is marked, and the image processing result containing the location information of the hemorrhage area is output.

[0008] Optionally, hemodynamic parameters for each anatomical region image can be obtained, including: Determine whether dynamic multiphase brain CT scan images include a sequence of continuously acquired CT perfusion images; If the sequence includes continuously acquired CT perfusion images, target voxels are selected in each anatomical partition image based on the statistical characteristics of the CT value distribution of each anatomical partition image. Target voxels include voxels in each anatomical partition image whose CT values ​​are close to the mean, or the geometric center voxels in each anatomical partition image located in the region of maximum density. Continuous CT value data of the target voxel in CT perfusion image sequences are obtained to construct the time and density change curves of the target voxel; The slope of the arterial phase CT value increase, the rate of decrease of the venous phase CT value, and the vascular impedance index were calculated based on the time and density change curves of the target voxel.

[0009] Optionally, the continuously acquired CT perfusion image sequences include arterial phase image sequences, venous phase image sequences, and delayed phase image sequences; The slope of the arterial phase CT value increase, the rate of decrease of the venous phase CT value, and the vascular impedance index were calculated based on the time and density change curves of the target voxel, including: Within the time window of the arterial phase, the rising segment of the time and density change curves was linearly fitted using the least squares method to obtain the slope of the rising CT value during the arterial phase. Within the time window from the venous phase to the delayed phase, a double exponential decay fitting was performed on the descending segment of the time and density change curves to obtain the rate of decrease in CT values ​​during the venous phase. The vascular resistance index is obtained by dividing the difference between the peak value of the arterial phase and the trough value of the venous phase in the time and density change curve by the duration of the arterial phase.

[0010] Optional, if not including sequentially acquired CT perfusion image sequences; Then, obtain the hemodynamic parameters of each anatomical region image, including: The image statistical features corresponding to each anatomical partition image, the patient's clinical physiological parameters, and the spatial location number of the anatomical partition corresponding to the anatomical partition image in the standard Talairach coordinate system are used as inputs to the parameter estimation model for processing, to obtain the arterial phase CT value rise slope, venous phase CT value fall rate, and vascular impedance index of each anatomical partition image. Among them, the spatial location number is mapped to a fixed-length vector embedding through the partition number of the dissecting partition, which is used to represent the location information of the dissecting partition; The parameter estimation model employs a feedforward fully connected neural network structure, including multiple input layers, hidden layers, and output layers, to predict the corresponding hemodynamic parameters from the input information.

[0011] Optionally, the dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence; The feature fusion layer is used to generate partition features for each anatomical partition image based on clinical physiological parameters, hemodynamic parameters, and enhancement parameters; The interval calculation layer is used to generate an initial CT value distribution reference interval for each anatomical partition image based on the partition features output by the feature fusion layer; The output layer is optimized to adjust the initial CT value distribution reference interval and output the final CT value distribution reference interval.

[0012] Optionally, the dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence; Clinical physiological parameters, hemodynamic parameters, and enhancement parameters are input into the dynamic threshold prediction model to generate a reference interval for the CT value distribution of each anatomical region image, including: The feature fusion layer receives clinical physiological parameters, hemodynamic parameters, and enhancement parameters, and performs the following operations for each anatomical region image based on these parameters to generate region features for each anatomical region image: Based on the arterial phase CT value rise slope, venous phase CT value fall rate, intracranial pressure score, and CT grayscale correction factor of the anatomical partition images, the spatiotemporal coupling parameters of the anatomical partition images are calculated; the spatiotemporal coupling parameters are used to describe the spatiotemporal evolution of blood flow features in the anatomical partition images. Based on the cerebral oxygen metabolism rate, the average cerebral oxygen metabolism rate of the whole brain, the vascular density of the anatomical partition image, the average vascular density of the whole brain, and the CT value variability of the adjacent anatomical partition image, the vascular impedance weight of the anatomical partition image is calculated. The vascular impedance weight is used to describe the degree of influence of the vascular impedance index of the anatomical partition image on the CT value distribution reference interval. Based on the calcification score, cerebral oxygen metabolism rate, cerebral small vessel disease score, and intracranial pressure score of the anatomical partition images, the metabolic pressure compensation factor of the anatomical partition images is calculated. The metabolic pressure compensation factor is a composite pathological correction parameter used to compensate for the CT value deviation caused by metabolic abnormalities or increased intracranial pressure. Calculate the patient's metabolic age coefficient and clinical risk factors based on clinical physiological parameters; The interval calculation layer receives the partition features output by the feature fusion layer, and performs the following operations for each anatomical partition image based on the partition features to generate an initial CT value distribution reference interval for each anatomical partition image: Baseline CT values ​​of anatomical region images are calculated based on the average CT value of the anatomical region images, cerebral oxygen metabolism rate, normal reference value of cerebral oxygen metabolism rate, and metabolic age coefficient. Based on baseline CT values, spatiotemporal coupling parameters, vascular impedance weights, and intracranial pressure scores, the upper limit of the reference interval for the distribution of initial CT values ​​is calculated. Based on baseline CT values, cerebral small vessel disease scores, vascular impedance index, and clinical risk factors, the lower limit of the initial CT value distribution reference interval was calculated. The optimized output layer performs the following operations for each anatomical region image to generate a final CT value distribution reference interval for each anatomical region image: If the cerebral oxygen metabolism rate of the anatomical partition image is less than the initial CT value distribution reference interval, the lower limit of the initial CT value distribution reference interval is adjusted based on the cerebral oxygen metabolism rate and blood oxygen saturation in the clinical physiological parameters. Calculate the first ratio of the width of the adjusted initial CT value distribution reference interval to the cerebral oxygen metabolism rate of the anatomical partition image; The average ratio of the interval width to the brain oxygen metabolism rate of all anatomical partition images is calculated based on the first ratio of each anatomical partition image. If the first ratio of each anatomical partition image is less than or equal to two standard deviations of the ratio mean, and the adjusted initial CT value distribution reference interval of each anatomical partition image satisfies the preset CT value distribution reference interval condition, then the adjusted initial CT value distribution reference interval is used as the corresponding CT value distribution reference interval for each anatomical partition image, and the corresponding CT value distribution reference interval for each anatomical partition image is output.

[0013] Optional, preset CT value distribution reference interval conditions include: The CT value distribution reference interval is determined by the base width of the CT value distribution reference interval, the heart rate compensation amount in the patient's clinical physiological parameters, and the respiratory rate compression amount in the patient's clinical physiological parameters. The base width of the CT value distribution reference interval refers to the range of fluctuation of the standard CT values ​​for anatomical divisions in healthy individuals.

[0014] Secondly, this application provides a hemorrhage region identification device based on brain imaging, comprising a first information acquisition module, a skull dissection processing module, an anatomical region division module, a second information acquisition module, an information processing module, and a hemorrhage region identification module, wherein: The first information acquisition module is used to acquire dynamic multi-phase brain CT scan images and clinical physiological parameters of the patient; The cranial dissection processing module is used to perform cranial dissection preprocessing based on dynamic multi-phase brain CT scan images to extract brain tissue region images; The anatomical partitioning module is used to divide brain tissue region images into multiple anatomical partition images based on the standard Talairach coordinate system. The second information acquisition module is used to acquire the image statistical features, hemodynamic parameters, and enhancement parameters of each anatomical region image. Among them, the hemodynamic parameters include the CT value rise slope in the arterial phase, the CT value fall rate in the venous phase, and the vascular impedance index; the enhancement parameters include the cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor. The information processing module is used to input clinical physiological parameters, hemodynamic parameters and enhancement parameters into the dynamic threshold prediction model to generate the corresponding CT value distribution reference interval for each anatomical partition image; The hemorrhage area identification module is used to annotate the hemorrhage areas in dynamic multi-phase brain CT scan images based on the CT value distribution reference interval, and output image processing results containing the location information of the hemorrhage areas.

[0015] Thirdly, this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the method for identifying hemorrhage regions based on brain images as described above.

[0016] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the brain image-based hemorrhage region identification method as described above.

[0017] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the brain image-based hemorrhage region identification method described above.

[0018] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: The method provided in this application extracts brain tissue from dynamic multi-phase brain CT images and partitions the brain tissue images in the Talairach standard coordinate system. By combining the image statistical features, hemodynamic parameters, enhancement parameters, and the patient's clinical physiological parameters of each anatomical partition, a dynamic threshold prediction model is constructed to generate personalized CT value distribution reference intervals. Compared to existing technologies that rely on fixed CT value thresholds for hemorrhage area detection, this method can fully consider the differences in tissue density in different brain regions and the physiological differences between individuals, thus effectively solving the problems of misidentification and low accuracy of hemorrhage areas caused by fixed thresholds. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A schematic diagram illustrating the implementation process of a method for identifying hemorrhage regions based on brain images, provided in an embodiment of this application; Figure 2 A schematic diagram of the specific structure of a hemorrhage region identification device based on brain imaging is provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0022] Example 1 To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] As will be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application.

[0025] Currently, in clinical practice and existing methods, the global threshold method is commonly used for preliminary identification of hemorrhage areas. This method is based on the high-density (i.e., high CT value) imaging feature of hemorrhage areas in CT images. By setting a fixed CT value threshold range (e.g., 40–80 Hounsfield (HU)), it distinguishes hemorrhage tissue areas from normal brain tissue areas.

[0026] However, in practical applications, the normal CT values ​​of different brain regions (such as the frontal lobe, parietal lobe, temporal lobe, occipital lobe, brainstem, and cerebellum) vary. Therefore, using a single fixed threshold in existing technologies is difficult to adapt to all anatomical structures, easily leading to false detections. Secondly, this method ignores individual physiological differences, such as the patient's age, gender, and basal metabolic state, which further reduces the accuracy of the detection.

[0027] To address the problem that existing methods for detecting hemorrhagic tissue regions using fixed CT value threshold ranges fail to adequately consider differences in tissue density across different brain regions and physiological differences among patients, leading to frequent misidentifications and low accuracy, this application provides a method for identifying hemorrhagic regions based on brain images.

[0028] The execution subject of this method can be various types of computing devices, or it can be an application or app installed on the computing device. The computing device can be a user terminal such as a mobile phone, tablet computer, or smart wearable device, or it can be a server.

[0029] For ease of description, this application uses a server as the execution subject of the method in its embodiments to illustrate the method. Those skilled in the art will understand that this embodiment uses a server as an example to describe the method, which is merely an illustrative example and does not limit the scope of protection of the corresponding claims.

[0030] Specifically, the implementation flow of the method provided in this application embodiment is as follows: Figure 1 As shown, it includes the following steps: Step 11: Obtain dynamic multiphase brain CT scan images and clinical physiological parameters of the patient.

[0031] The patient's dynamic multiphase brain CT scan images include arterial phase image sequences, venous phase image sequences, and delayed phase image sequences.

[0032] Clinical physiological parameters include baseline vital signs, biochemical and laboratory parameters, demographic and chronic indicators, and treatment-related parameters. Biochemical and laboratory parameters include information such as serum creatinine, blood glucose, and arterial oxygen partial pressure. Demographic and chronic indicators include information such as age, BMI, and sex.

[0033] Basic vital signs parameters, including information such as the patient's systolic blood pressure, blood oxygen saturation, heart rate, and respiratory rate.

[0034] It should be noted that the basic vital signs parameters, biochemical and laboratory parameters, and demographic and chronic indicators listed above are merely exemplary descriptions of this application and do not impose any limitations on the embodiments of this application.

[0035] Step 12: Perform cranial dissection preprocessing based on dynamic multi-phase brain CT scan images to extract brain tissue region images.

[0036] Skull dissection preprocessing refers to the process of removing non-brain tissues such as the skull and dura mater based on dynamic multi-phase brain CT scan images, while preserving images of the brain parenchyma.

[0037] In this embodiment of the application, considering that there are grayscale differences in different tissues in CT images, such as brain tissue being approximately 20-80 HU, while the HU threshold of the skull is usually higher than that of other tissues (typically, the skull > 200 HU), a HU threshold can be set to separate the skull. For example, the HU threshold can be set to 200 HU, so that structures greater than 200 HU are regarded as skull region images, and the rest are brain tissue region images.

[0038] Alternatively, skull boundaries can be extracted from dynamic multi-phase brain CT scan images using operators such as Canny and Sobel, and then combined with level set evolution segmentation of dynamic multi-phase brain CT scan images to obtain brain tissue region images.

[0039] In the process of obtaining brain tissue region images by combining level set evolution segmentation of dynamic multi-phase brain CT scan images, the contours of high-density structures such as the skull can first be extracted from the dynamic multi-phase brain CT scan images using edge detection operators (such as Canny or Sobel) as initial segmentation boundaries. Then, this boundary is embedded into a pre-defined high-dimensional level set function.

[0040] Secondly, during the evolutionary segmentation process, the level set function is dynamically updated based on the image gradient, curvature constraints, and regional grayscale features of dynamic multi-phase brain CT scan images, driving the initial contour to approximate the true boundary of the brain tissue: the curvature term ensures a smooth contour, the image gradient term guides the contour to fit the edge of the brain parenchyma, and the grayscale constraint utilizes the difference in HU values ​​between the brain tissue and surrounding structures (e.g., 20-80 HU) to enhance segmentation accuracy. Finally, the zero contour line of the level set function converges to the brain tissue boundary, thus completing the segmentation.

[0041] Alternatively, this application can also fill the cranial gaps through morphological closure operations, then extract the cranial mask and reverse-engineer the brain tissue region.

[0042] It should be noted that the above-described example of preprocessing the skull based on dynamic multi-phase brain CT scan images to extract brain tissue region images is merely an exemplary description of this application and does not limit the embodiments of this application in any way.

[0043] Step 13: Divide the brain tissue region image into multiple anatomical partition images based on the standard Talairach coordinate system.

[0044] In this embodiment of the application, anatomical landmarks of the anterior commissure (AC) and posterior commissure (PC) can be detected; Then, a standard Talairach coordinate system is constructed with the midpoint of the AC-PC line as the origin. Finally, the brain tissue region images are aligned to the standard Talairach brain template through affine transformation to obtain multiple anatomical partition images.

[0045] In this embodiment, the identification and localization results of brain hemorrhage generally include intraventricular hemorrhage and parenchymal hemorrhage. Preferably, the brain region includes ventricles and brain parenchyma. The ventricles include the left and right ventricles, and the brain parenchyma includes the left frontal lobe, right frontal lobe, left temporal lobe, right temporal lobe, left parietal lobe, right parietal lobe, left occipital lobe, right occipital lobe, left centrum ovale, right centrum ovale, left basal ganglia, right basal ganglia, left cerebellum, right cerebellum, and brainstem. Therefore, the brain region can be considered to include 17 subregions, named as follows: left ventricle, right ventricle, left frontal lobe, right frontal lobe, left temporal lobe, right temporal lobe, left parietal lobe, right parietal lobe, left occipital lobe, right occipital lobe, left centrum ovale, right centrum ovale, left basal ganglia, right basal ganglia, left cerebellum, right cerebellum, and brainstem.

[0046] Correspondingly, in an optional implementation, the brain tissue region image can be divided into 17 anatomical partition images based on the standard Talairach coordinate system, including the left ventricle anatomical partition image, the right ventricle anatomical partition image, the left frontal lobe anatomical partition image, the right frontal lobe anatomical partition image, the left temporal lobe anatomical partition image, the right temporal lobe anatomical partition image, the left parietal lobe anatomical partition image, the right parietal lobe anatomical partition image, the left occipital lobe anatomical partition image, the right occipital lobe anatomical partition image, the left centrum ovale anatomical partition image, the right centrum ovale anatomical partition image, the left basal ganglia anatomical partition image, the right basal ganglia anatomical partition image, the left cerebellum anatomical partition image, and the right cerebellum and brainstem anatomical partition images.

[0047] Alternatively, in another alternative implementation, the brain tissue region images can be divided into nine anatomical regions based on the standard Talairach coordinate system, including ventricular anatomical regions, frontal lobe anatomical regions, temporal lobe anatomical regions, parietal lobe anatomical regions, occipital lobe anatomical regions, centrum semiovale anatomical regions, basal ganglia anatomical regions, cerebellum anatomical regions, and brainstem anatomical regions.

[0048] It should be noted that the above-described method for dividing brain tissue region images into multiple anatomical partition images is merely an exemplary description of this application and does not impose any limitation on the embodiments of this application.

[0049] In this embodiment, the corresponding division method can be selected according to actual needs. For example, the brain tissue region image can be divided into 9 anatomical regions based on the standard Talairach coordinate system. Alternatively, the brain tissue region image can be divided into 17 anatomical regions based on the standard Talairach coordinate system.

[0050] Step 14: Obtain the image statistical features, hemodynamic parameters, and enhancement parameters for each anatomical region image.

[0051] Among them, hemodynamic parameters include the slope of the CT value increase during the arterial phase, the rate of decrease of the CT value during the venous phase, and the vascular impedance index.

[0052] In this embodiment, the vascular impedance index is equal to the difference between the peak value of the arterial phase CT and the trough value of the venous phase CT in the time and density change curve, divided by the duration of the arterial phase.

[0053] Enhanced parameters include cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor.

[0054] In one alternative implementation, the hemodynamic parameters of each anatomical region image can be obtained as follows: Determine whether dynamic multiphase brain CT scan images include a sequence of continuously acquired CT perfusion images; If the sequence includes continuously acquired CT perfusion images, target voxels are selected in each anatomical partition image based on the statistical characteristics of the CT value distribution of each anatomical partition image. Target voxels include voxels in each anatomical partition image whose CT values ​​are close to the mean, or the geometric center voxels in each anatomical partition image located in the region of maximum density. Continuous CT value data of the target voxel in CT perfusion image sequences are obtained to construct the time and density change curves of the target voxel; The slope of the arterial phase CT value increase, the rate of decrease of the venous phase CT value, and the vascular impedance index were calculated based on the time and density change curves of the target voxel.

[0055] In one alternative implementation, when acquiring continuous CT value data of the target voxel in a CT perfusion image sequence to construct the time-density variation curve of the target voxel, the physical coordinates of the selected target voxel (such as the geometric center voxel) can be converted into a matrix index in each temporal image. Then, based on this matrix index, the CT values ​​of the voxel are cyclically read along the time axis in all temporal phases to construct the original data pair of time and CT values. Finally, the time-density variation curve of the target voxel is constructed based on this original data pair.

[0056] Optionally, if the continuously acquired CT perfusion image sequence includes arterial phase image sequences, venous phase image sequences, and delayed phase image sequences, then the calculation of the arterial phase CT value rise slope, venous phase CT value fall rate, and vascular impedance index based on the time and density change curves of the target voxels can be achieved as follows: Within the time window of the arterial phase, the rising segment of the time and density change curves was linearly fitted using the least squares method to obtain the slope of the rising CT value during the arterial phase. Within the time window from the venous phase to the delayed phase, a double exponential decay fitting was performed on the descending segment of the time and density change curves to obtain the rate of decrease in CT values ​​during the venous phase. The vascular resistance index is obtained by dividing the difference between the peak value of CT in the arterial phase and the trough value of CT in the venous phase in the time and density change curve by the duration of the arterial phase.

[0057] Alternatively, if a sequence of continuously acquired CT perfusion images is not available, the hemodynamic parameters for each anatomical region can be obtained as follows: The image statistical features corresponding to each anatomical partition image, the patient's clinical physiological parameters, and the spatial location number of the anatomical partition corresponding to the anatomical partition image in the standard Talairach coordinate system are used as inputs to the parameter estimation model for processing, to obtain the arterial phase CT value rise slope, venous phase CT value fall rate, and vascular impedance index of each anatomical partition image. Among them, the spatial location number is mapped to a fixed-length vector embedding through the partition number of the dissecting partition, which is used to represent the location information of the dissecting partition; The parameter estimation model employs a feedforward fully connected neural network structure, including multiple input layers, hidden layers, and output layers, to predict the corresponding hemodynamic parameters from the input information.

[0058] Step 15: Input clinical physiological parameters, hemodynamic parameters, and enhancement parameters into the dynamic threshold prediction model to generate the corresponding CT value distribution reference interval for each anatomical partition image; The dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence. The feature fusion layer is used to generate partition features for each anatomical partition image based on clinical physiological parameters, hemodynamic parameters, and enhancement parameters; The interval calculation layer is used to generate an initial CT value distribution reference interval for each anatomical partition image based on the partition features output by the feature fusion layer; The output layer is optimized to adjust the initial CT value distribution reference interval and output the final CT value distribution reference interval.

[0059] In one alternative implementation, a reference interval for the CT value distribution of each anatomical region image can be generated based on clinical physiological parameters, hemodynamic parameters, and enhancement parameters using a dynamic threshold prediction model, as follows: The feature fusion layer receives clinical physiological parameters, hemodynamic parameters, and enhancement parameters, and performs the following operations for each anatomical region image based on these parameters to generate region features for each anatomical region image: Based on the arterial phase CT value rise slope, venous phase CT value fall rate, intracranial pressure score, and CT grayscale correction factor of the anatomical partition images, the spatiotemporal coupling parameters of the anatomical partition images are calculated; the spatiotemporal coupling parameters are used to describe the spatiotemporal evolution of blood flow features in the anatomical partition images. Based on the cerebral oxygen metabolism rate, the average cerebral oxygen metabolism rate of the whole brain, the vascular density of the anatomical partition image, the average vascular density of the whole brain, and the CT value variability of the adjacent anatomical partition image, the vascular impedance weight of the anatomical partition image is calculated. This vascular impedance weight is used to describe the degree of influence of the vascular impedance index of the anatomical partition image on the CT value distribution reference interval. Based on the calcification score, cerebral oxygen metabolism rate, cerebral small vessel disease score, and intracranial pressure score of the anatomical partition images, the metabolic pressure compensation factor of the anatomical partition images is calculated. The metabolic pressure compensation factor is a composite pathological correction parameter used to compensate for the CT value deviation caused by metabolic abnormalities or increased intracranial pressure. The patient's metabolic age coefficient and clinical risk factors were calculated based on clinical physiological parameters.

[0060] It should be noted that the reason for calculating the spatiotemporal coupling parameters of anatomical region images is because cerebral blood flow has significant spatiotemporal coupling characteristics. Ignoring the transition relationship between the arterial and venous phases, especially the flow velocity regulation behavior under the background of intracranial pressure changes, will lead to misidentification of risks such as cerebral hemorrhage. Moreover, in existing related technologies, when identifying hemorrhage areas based on brain images, the dynamic changes in CT values ​​of anatomical regions across multiple time phases are ignored, resulting in a lack of temporal continuity and spatial sensitivity in the assessment of cerebral blood flow status. In this application embodiment, in order to avoid the lack of temporal continuity and spatial sensitivity in the assessment of cerebral blood flow status and to reduce identification errors, the method of calculating the spatiotemporal coupling parameters of anatomical region images is proposed.

[0061] In one optional implementation, when calculating the spatiotemporal coupling parameters of the anatomical partition image based on the arterial phase CT value rise slope, venous phase CT value fall rate, intracranial pressure score, and CT grayscale correction factor, the following implementation method can be used:

[0062] in, These are spatiotemporal coupling parameters; It is the slope of the CT value increase during the arterial phase; It is the rate of decrease in CT value during the venous phase; Indicates intracranial pressure score; It is the CT grayscale correction factor; It is an intracranial pressure sensitivity factor; This is a reference value for central intracranial pressure in clinical settings.

[0063] Secondly, in this embodiment, considering the different blood flow resistance in different anatomical regions, especially in areas where there are differences in cerebral vascular density, metabolic demand, and spatial variability, changes in blood flow impedance have a significant impact on the CT value distribution. Therefore, if changes in vascular impedance cannot be accurately reflected, it can easily lead to distortion in the interpretation of CT values, thereby affecting the accuracy of identifying brain hemorrhage areas. Based on this, to avoid these problems, in an optional embodiment, the vascular impedance weight of the anatomical region image can be calculated according to the cerebral oxygen metabolism rate, the average cerebral oxygen metabolism rate of the whole brain, the vascular density of the anatomical region image, the average vascular density of the whole brain, and the CT value variability of adjacent anatomical region images, as follows:

[0064] in, It is the vascular impedance weight; This represents the brain's oxygen metabolism rate in anatomical region images. This represents the average brain oxygen metabolism rate of the entire brain. This represents the blood vessel density in an anatomical region image. This represents the average vascular density of the entire brain. This indicates the variability of CT values ​​in adjacent anatomical partition images compared to the anatomical partition image. , , These are the preset adjustment weight parameters.

[0065] Furthermore, CT images often exhibit grayscale shifts due to abnormal tissue metabolism, calcification, or abnormal intracranial pressure. Increased metabolic pressure is frequently accompanied by perfusion abnormalities, calcification, and small vessel disease, directly affecting CT values. Failure to correct for the impact of metabolic pressure on CT values ​​may lead to misjudgments when identifying hemorrhage areas based on brain images. In this application, to avoid this problem, in one optional embodiment, a pressure function F can be constructed first based on the calcification score, cerebral oxygen metabolism rate, cerebral small vessel disease score, and intracranial pressure score of the anatomical region images; then, a metabolic pressure compensation factor for the anatomical region images is calculated based on the pressure function F.

[0066] Represents the pressure function; Indicates calcification score; Indicates brain oxygen metabolism rate; Standard reference values ​​for brain oxygen metabolism; Indicates the cerebral small vessel disease score; to These are the weight parameters for each score, which can be obtained by fitting the training data.

[0067]

[0068] in, Metabolic stress compensation factor representing anatomical partition images.

[0069] The interval calculation layer receives the partition features output by the feature fusion layer, and performs the following operations for each anatomical partition image based on the partition features to generate an initial CT value distribution reference interval for each anatomical partition image: Baseline CT values ​​of anatomical region images are calculated based on the average CT value of the anatomical region images, cerebral oxygen metabolism rate, normal reference value of cerebral oxygen metabolism rate, and metabolic age coefficient. Based on baseline CT values, spatiotemporal coupling parameters, vascular impedance weights, and intracranial pressure scores, the upper limit of the reference interval for the distribution of initial CT values ​​is calculated. The lower limit of the reference interval for the distribution of initial CT values ​​was calculated based on baseline CT values, cerebral small vessel disease score, vascular impedance index, and clinical risk factors.

[0070] The optimized output layer performs the following operations for each anatomical region image to generate a final CT value distribution reference interval for each anatomical region image: If the cerebral oxygen metabolism rate of the anatomical partition image is less than the initial CT value distribution reference interval, the lower limit of the initial CT value distribution reference interval is adjusted based on the cerebral oxygen metabolism rate and blood oxygen saturation in the clinical physiological parameters. Calculate the first ratio of the width of the adjusted initial CT value distribution reference interval to the cerebral oxygen metabolism rate of the anatomical partition image; The average ratio of the interval width to the brain oxygen metabolism rate of all anatomical partition images is calculated based on the first ratio of each anatomical partition image. If the first ratio of each anatomical partition image is less than or equal to two standard deviations of the ratio mean, and the adjusted initial CT value distribution reference interval of each anatomical partition image satisfies the preset CT value distribution reference interval condition, then the adjusted initial CT value distribution reference interval is used as the corresponding CT value distribution reference interval for each anatomical partition image, and the corresponding CT value distribution reference interval for each anatomical partition image is output.

[0071] Optionally, the preset CT value distribution reference interval conditions are: The CT value distribution reference interval is determined by the base width of the CT value distribution reference interval, the heart rate compensation amount in the patient's clinical physiological parameters, and the respiratory rate compression amount in the patient's clinical physiological parameters. The basic width of the CT value distribution reference interval refers to the range of fluctuation of the standard CT values ​​for anatomical divisions in healthy individuals.

[0072] In one alternative implementation, the CT value distribution reference interval can be determined by first taking the baseline width of the reference interval, then adding the heart rate compensation from the patient's clinical physiological parameters, and finally subtracting the respiratory rate compression from the patient's clinical physiological parameters.

[0073] Step 16: Based on the CT value distribution reference interval, the hemorrhage area in the dynamic multi-phase brain CT scan image is marked, and the image processing result containing the location information of the hemorrhage area is output.

[0074] In this embodiment, the hemorrhage area can be superimposed with a red outline on a dynamic multi-phase brain CT scan image.

[0075] For example, suppose a hemorrhage region is identified in the anatomical region of the left frontal lobe based on the CT value distribution reference interval, and the location information of this hemorrhage region is: Talairach standard coordinate system: x=120, y=200, z=45. Then the location information of this hemorrhage region can be output.

[0076] Optionally, the image processing results may also include information such as the bleeding volume of the bleeding area and the confidence level of identifying the bleeding area.

[0077] The method provided in this application extracts brain tissue from dynamic multi-phase brain CT images and partitions the brain tissue images in the Talairach standard coordinate system. By combining the image statistical features, hemodynamic parameters, enhancement parameters, and the patient's clinical physiological parameters of each anatomical partition, a dynamic threshold prediction model is constructed to generate personalized CT value distribution reference intervals. Compared to existing technologies that rely on fixed CT value thresholds for hemorrhage area detection, this method can fully consider the differences in tissue density in different brain regions and the physiological differences between individuals, thus effectively solving the problems of misidentification and low accuracy of hemorrhage areas caused by fixed thresholds.

[0078] Example 2 To address the problem that existing methods for detecting hemorrhagic tissue regions using fixed CT value thresholds fail to adequately consider differences in tissue density across different brain regions and physiological variations among patients, leading to frequent misidentifications and low accuracy, this application provides a hemorrhagic region identification device based on brain imaging. A schematic diagram of the device's specific structure is shown below. Figure 2 As shown, it includes a first information acquisition module 21, a skull dissection processing module 22, an anatomical region division module 23, a second information acquisition module 24, an information processing module 25, and a hemorrhage area identification module 26. The functions of each module are as follows: The first information acquisition module 21 is used to acquire dynamic multi-phase brain CT scan images and clinical physiological parameters of the patient; The cranial dissection processing module 22 is used to perform cranial dissection preprocessing based on dynamic multi-phase brain CT scan images to extract brain tissue region images. Anatomical partitioning module 23 is used to divide brain tissue region images into multiple anatomical partition images based on the standard Talairach coordinate system; The second information acquisition module 24 is used to acquire the image statistical features, hemodynamic parameters, and enhancement parameters of each anatomical region image; among which, the hemodynamic parameters include the CT value rise slope in the arterial phase, the CT value fall rate in the venous phase, and the vascular impedance index; the enhancement parameters include the cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor. The information processing module 25 is used to input clinical physiological parameters, hemodynamic parameters and enhancement parameters into the dynamic threshold prediction model to generate the corresponding CT value distribution reference interval for each anatomical partition image; The hemorrhage area identification module 26 is used to mark the hemorrhage area in the dynamic multi-phase brain CT scan image based on the CT value distribution reference interval, and output the image processing result containing the location information of the hemorrhage area.

[0079] Optionally, the second information acquisition module 24 includes: The judgment unit is used to determine whether the dynamic multi-phase brain CT scan image includes a sequence of continuously acquired CT perfusion images; The selection unit is used to select target voxels in each anatomical partition image based on the statistical characteristics of the CT value distribution of each anatomical partition image if the image contains a sequence of continuously acquired CT perfusion images. The target voxels include voxels in each anatomical partition image whose CT values ​​are close to the mean, or the geometric center voxels in each anatomical partition image located in the region of maximum density. The construction unit is used to acquire continuous CT value data of the target voxel in the CT perfusion image sequence to construct the time and density change curve of the target voxel; The calculation unit is used to calculate the slope of the arterial phase CT value increase, the rate of the venous phase CT value decrease, and the vascular impedance index based on the time and density change curves of the target voxel.

[0080] Optionally, the continuously acquired CT perfusion image sequences include arterial phase image sequences, venous phase image sequences, and delayed phase image sequences; Computational unit, used for: Within the time window of the arterial phase, the rising segment of the time and density change curves was linearly fitted using the least squares method to obtain the slope of the rising CT value during the arterial phase. Within the time window from the venous phase to the delayed phase, a double exponential decay fitting was performed on the descending segment of the time and density change curves to obtain the rate of decrease in CT values ​​during the venous phase. The vascular resistance index is obtained by dividing the difference between the peak value of the arterial phase and the trough value of the venous phase in the time and density change curve by the duration of the arterial phase.

[0081] Optional, if not including sequentially acquired CT perfusion image sequences; Then, the calculation unit is used for: The image statistical features corresponding to each anatomical partition image, the patient's clinical physiological parameters, and the spatial location number of the anatomical partition corresponding to the anatomical partition image in the standard Talairach coordinate system are used as inputs to the parameter estimation model for processing, to obtain the arterial phase CT value rise slope, venous phase CT value fall rate, and vascular impedance index of each anatomical partition image. Among them, the spatial location number is mapped to a fixed-length vector embedding through the partition number of the dissecting partition, which is used to represent the location information of the dissecting partition; The parameter estimation model employs a feedforward fully connected neural network structure, including multiple input layers, hidden layers, and output layers, to predict the corresponding hemodynamic parameters from the input information.

[0082] Optionally, the dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence; The feature fusion layer is used to generate partition features for each anatomical partition image based on clinical physiological parameters, hemodynamic parameters, and enhancement parameters; The interval calculation layer is used to generate an initial CT value distribution reference interval for each anatomical partition image based on the partition features output by the feature fusion layer; The output layer is optimized to adjust the initial CT value distribution reference interval and output the final CT value distribution reference interval.

[0083] Optionally, the dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence; Information processing module 25 is used for: The feature fusion layer receives clinical physiological parameters, hemodynamic parameters, and enhancement parameters, and performs the following operations for each anatomical region image based on these parameters to generate region features for each anatomical region image: Based on the arterial phase CT value rise slope, venous phase CT value fall rate, intracranial pressure score, and CT grayscale correction factor of the anatomical partition images, the spatiotemporal coupling parameters of the anatomical partition images are calculated; the spatiotemporal coupling parameters are used to describe the spatiotemporal evolution of blood flow features in the anatomical partition images. Based on the cerebral oxygen metabolism rate, the average cerebral oxygen metabolism rate of the whole brain, the vascular density of the anatomical partition image, the average vascular density of the whole brain, and the CT value variability of the adjacent anatomical partition image, the vascular impedance weight of the anatomical partition image is calculated. The vascular impedance weight is used to describe the degree of influence of the vascular impedance index of the anatomical partition image on the CT value distribution reference interval. Based on the calcification score, cerebral oxygen metabolism rate, cerebral small vessel disease score, and intracranial pressure score of the anatomical partition images, the metabolic pressure compensation factor of the anatomical partition images is calculated. The metabolic pressure compensation factor is a composite pathological correction parameter used to compensate for the CT value deviation caused by metabolic abnormalities or increased intracranial pressure. Calculate the patient's metabolic age coefficient and clinical risk factors based on clinical physiological parameters; The interval calculation layer receives the partition features output by the feature fusion layer, and performs the following operations for each anatomical partition image based on the partition features to generate an initial CT value distribution reference interval for each anatomical partition image: Baseline CT values ​​of anatomical region images are calculated based on the average CT value of the anatomical region images, cerebral oxygen metabolism rate, normal reference value of cerebral oxygen metabolism rate, and metabolic age coefficient. Based on baseline CT values, spatiotemporal coupling parameters, vascular impedance weights, and intracranial pressure scores, the upper limit of the reference interval for the distribution of initial CT values ​​is calculated. The lower limit of the reference interval for the distribution of initial CT values ​​was calculated based on baseline CT values, cerebral small vessel disease score, vascular impedance index, and clinical risk factors.

[0084] The optimized output layer performs the following operations for each anatomical region image to generate a final CT value distribution reference interval for each anatomical region image: If the cerebral oxygen metabolism rate of the anatomical partition image is less than the initial CT value distribution reference interval, the lower limit of the initial CT value distribution reference interval is adjusted based on the cerebral oxygen metabolism rate and blood oxygen saturation in the clinical physiological parameters. Calculate the first ratio of the width of the adjusted initial CT value distribution reference interval to the cerebral oxygen metabolism rate of the anatomical partition image; The average ratio of the interval width to the brain oxygen metabolism rate of all anatomical partition images is calculated based on the first ratio of each anatomical partition image. If the first ratio of each anatomical partition image is less than or equal to two standard deviations of the ratio mean, and the adjusted initial CT value distribution reference interval of each anatomical partition image satisfies the preset CT value distribution reference interval condition, then the adjusted initial CT value distribution reference interval is used as the corresponding CT value distribution reference interval for each anatomical partition image, and the corresponding CT value distribution reference interval for each anatomical partition image is output.

[0085] Optional, preset CT value distribution reference interval conditions include: The CT value distribution reference interval is determined by the base width of the CT value distribution reference interval, the heart rate compensation amount in the patient's clinical physiological parameters, and the respiratory rate compression amount in the patient's clinical physiological parameters.

[0086] In one alternative implementation, the CT value distribution reference interval can be determined by first taking the baseline width of the reference interval, then adding the heart rate compensation from the patient's clinical physiological parameters, and finally subtracting the respiratory rate compression from the patient's clinical physiological parameters.

[0087] The base width of the CT value distribution reference interval refers to the range of fluctuation of the standard CT values ​​for anatomical divisions in healthy individuals.

[0088] The device provided in this application extracts brain tissue based on dynamic multi-phase brain CT images and partitions the brain tissue images in the Talairach standard coordinate system. It then combines the image statistical features, hemodynamic parameters, enhancement parameters, and the patient's clinical physiological parameters of each anatomical partition to construct a dynamic threshold prediction model to generate personalized CT value distribution reference intervals. Compared to existing technologies that rely on fixed CT value thresholds for hemorrhage area detection, this method fully considers the differences in tissue density in different brain regions and individual physiological differences, thereby effectively solving the problems of misidentification and low accuracy of hemorrhage areas caused by fixed thresholds.

[0089] Example 3 Figure 3 To illustrate the hardware structure of an electronic device according to various embodiments of this application, the electronic device may include a processor 301 and a memory 302 storing computer program instructions. Specifically, the processor 301 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of this application.

[0090] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 302 may include removable or non-removable (or fixed) media. Where appropriate, memory 302 may be internal or external to an electronic device. In a particular embodiment, memory 302 may be a non-volatile solid-state memory.

[0091] In one embodiment, memory 302 may be read-only memory (ROM). In one embodiment, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0092] The processor 301 reads and executes computer program instructions stored in the memory 302 to implement any of the brain image-based hemorrhage region identification methods in the above embodiments.

[0093] In one example, the electronic device may also include a communication interface 303 and a bus 310. For example, Figure 3 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 310 and complete communication with each other.

[0094] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0095] Bus 310 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 310 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0096] Furthermore, in conjunction with the brain image-based hemorrhage region identification method in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the brain image-based hemorrhage region identification methods in the above embodiments.

[0097] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0098] The above description is merely a specific implementation example of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0099] Secondly, those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0104] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0105] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0106] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0107] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for identifying hemorrhage regions based on brain imaging, characterized in that, include: To obtain dynamic multiphase brain CT scan images and clinical physiological parameters of patients; Based on the aforementioned dynamic multi-phase brain CT scan images, cranial dissection preprocessing is performed to extract brain tissue region images; The brain tissue region image is divided into multiple anatomical partition images based on the standard Talairach coordinate system; Image statistical features, hemodynamic parameters, and enhancement parameters are acquired for each anatomical region image; wherein, the hemodynamic parameters include the CT value rise slope in the arterial phase, the CT value fall rate in the venous phase, and the vascular impedance index; the enhancement parameters include cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor. The clinical physiological parameters, hemodynamic parameters, and enhancement parameters are input into the dynamic threshold prediction model to generate a corresponding CT value distribution reference interval for each anatomical partition image; Based on the CT value distribution reference interval, the hemorrhage area in the dynamic multi-phase brain CT scan image is marked, and the image processing result containing the location information of the hemorrhage area is output.

2. The method as described in claim 1, characterized in that, Obtain hemodynamic parameters for each of the anatomical regions, including: Determine whether the dynamic multi-phase brain CT scan images include a sequence of continuously acquired CT perfusion images; If the continuously acquired CT perfusion image sequence is included, then a target voxel is selected in each anatomical partition image based on the statistical characteristics of the CT value distribution of each anatomical partition image. The target voxel includes voxels in each anatomical partition image whose CT value is close to the mean, or the geometric center voxel in each anatomical partition image located in the region of maximum density. Continuous CT value data of the target voxel in the CT perfusion image sequence are obtained to construct the time and density change curve of the target voxel; The slope of the arterial phase CT value increase, the rate of the venous phase CT value decrease, and the vascular impedance index are calculated based on the time and density change curves of the target voxel.

3. The method as described in claim 2, characterized in that, The continuously acquired CT perfusion image sequence includes arterial phase image sequence, venous phase image sequence, and delayed phase image sequence; The slope of the arterial phase CT value increase, the rate of the venous phase CT value decrease, and the vascular impedance index are calculated based on the time and density change curves of the target voxel, including: Within the time window of the arterial phase, the rising segment of the time and density change curves is linearly fitted using the least squares method to obtain the slope of the rising CT value during the arterial phase. Within the time window from the venous phase to the delayed phase, a double exponential decay fitting is performed on the descending segment of the time and density change curves to obtain the CT value decrease rate during the venous phase. The vascular resistance index is obtained by dividing the difference between the peak value of the arterial phase and the trough value of the venous phase in the time and density change curve by the duration of the arterial phase.

4. The method as described in claim 2, characterized in that, If it does not include the continuously acquired CT perfusion image sequence; Then, the hemodynamic parameters of each anatomical region image are obtained, including: The image statistical features corresponding to each anatomical partition image, the patient's clinical physiological parameters, and the spatial location number of the anatomical partition corresponding to the anatomical partition image in the standard Talairach coordinate system are used as inputs to the parameter estimation model for processing, to obtain the arterial phase CT value rise slope, venous phase CT value fall rate, and vascular impedance index of each anatomical partition image; The spatial location number is mapped to a fixed-length vector embedding through the partition number of the dissecting partition, which is used to represent the location information of the dissecting partition; The parameter estimation model employs a feedforward fully connected neural network structure, including multiple input layers, hidden layers, and output layers, to predict the corresponding hemodynamic parameters from the input information.

5. The method as described in claim 1, characterized in that, The dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence. The feature fusion layer is used to generate partition features for each anatomical partition image based on the clinical physiological parameters, the hemodynamic parameters, and the enhancement parameters; The interval calculation layer is used to generate an initial CT value distribution reference interval for each anatomical partition image based on the partition features output by the feature fusion layer; The optimized output layer is used to adjust the initial CT value distribution reference interval and output the final CT value distribution reference interval.

6. The method as described in claim 1 or 5, characterized in that, The dynamic threshold prediction model includes a feature fusion layer, an interval calculation layer, and an optimized output layer connected in sequence. The clinical physiological parameters, hemodynamic parameters, and enhancement parameters are input into a dynamic threshold prediction model to generate a corresponding CT value distribution reference interval for each anatomical region image, including: The feature fusion layer receives the clinical physiological parameters, the hemodynamic parameters, and the enhancement parameters, and performs the following operations for each anatomical region image based on the clinical physiological parameters, hemodynamic parameters, and enhancement parameters to generate region features for each anatomical region image: Based on the rising slope of the arterial phase CT value, the falling rate of the venous phase CT value, the intracranial pressure score, and the CT grayscale correction factor of the anatomical partition image, the spatiotemporal coupling parameters of the anatomical partition image are calculated; the spatiotemporal coupling parameters are used to describe the spatiotemporal evolution of blood flow features in the anatomical partition image. Based on the cerebral oxygen metabolism rate, the average cerebral oxygen metabolism rate of the whole brain, the vascular density of the anatomical partition image, the average vascular density of the whole brain, and the CT value variability of the adjacent anatomical partition images, the vascular impedance weight of the anatomical partition image is calculated. The vascular impedance weight is used to describe the degree of influence of the vascular impedance index of the anatomical partition image on the CT value distribution reference interval. Based on the calcification score, cerebral oxygen metabolism rate, cerebral small vessel disease score, and intracranial pressure score of the anatomical partition image, a metabolic pressure compensation factor is calculated for the anatomical partition image. The metabolic pressure compensation factor is a composite pathological correction parameter used to compensate for CT value deviations caused by metabolic abnormalities or increased intracranial pressure. Calculate the patient's metabolic age coefficient and clinical risk factors based on the clinical physiological parameters. The interval calculation layer receives the partition features output by the feature fusion layer, and performs the following operations for each anatomical partition image based on the partition features to generate an initial CT value distribution reference interval for each anatomical partition image: The baseline CT value of the anatomical region image is calculated based on the average CT value of the anatomical region image, the cerebral oxygen metabolism rate, the normal reference value of the cerebral oxygen metabolism rate, and the metabolic age coefficient. Based on the baseline CT value, the spatiotemporal coupling parameter, the vascular impedance weight, and the intracranial pressure score, calculate the upper limit of the reference interval for the initial CT value distribution; Based on the baseline CT value, the cerebral small vessel disease score, the vascular impedance index, and the clinical risk factor, calculate the lower limit of the reference interval for the distribution of the initial CT value; The optimized output layer performs the following operations for each anatomical partition image to generate the final CT value distribution reference interval for each anatomical partition image: If the cerebral oxygen metabolism rate of the anatomical partition image is less than the initial CT value distribution reference interval, then the lower limit of the initial CT value distribution reference interval is adjusted based on the cerebral oxygen metabolism rate and the blood oxygen saturation in the clinical physiological parameters. Calculate the first ratio of the width of the adjusted initial CT value distribution reference interval to the cerebral oxygen metabolism rate of the anatomical partition image; The average ratio of the interval width to the brain oxygen metabolism rate of all the anatomical partition images is calculated based on the first ratio of each of the anatomical partition images. If the first ratio of each of the anatomical partition images is less than or equal to two standard deviations of the average of the ratios, and the adjusted initial CT value distribution reference interval of each of the anatomical partition images satisfies the preset CT value distribution reference interval conditions, then the adjusted initial CT value distribution reference interval is used as the corresponding CT value distribution reference interval for each of the anatomical partition images, and the corresponding CT value distribution reference interval for each of the anatomical partition images is output.

7. The method as described in claim 6, characterized in that, The preset CT value distribution reference interval conditions include: The CT value distribution reference interval is determined by the base width of the CT value distribution reference interval, the heart rate compensation amount in the patient's clinical physiological parameters, and the respiratory rate compression amount in the patient's clinical physiological parameters. The basic width of the CT value distribution reference interval refers to the fluctuation range of the standard CT values ​​for anatomical divisions in healthy individuals.

8. A device for identifying hemorrhage regions based on brain imaging, characterized in that, It includes a first information acquisition module, a skull dissection processing module, an anatomical region division module, a second information acquisition module, an information processing module, and a bleeding area identification module, wherein: The first information acquisition module is used to acquire dynamic multi-phase brain CT scan images and clinical physiological parameters of the patient; The cranial dissection processing module is used to perform cranial dissection preprocessing based on the dynamic multi-phase brain CT scan images to extract brain tissue region images. The anatomical partitioning module is used to divide the brain tissue region image into multiple anatomical partition images based on the standard Talairach coordinate system. The second information acquisition module is used to acquire the image statistical features, hemodynamic parameters, and enhancement parameters of each anatomical partition image; wherein, the hemodynamic parameters include the CT value rise slope in the arterial phase, the CT value fall rate in the venous phase, and the vascular impedance index; the enhancement parameters include cerebral oxygen metabolism rate, calcification score, cerebral small vessel disease score, intracranial pressure score, and CT grayscale correction factor. The information processing module is used to input the clinical physiological parameters, the hemodynamic parameters and the enhancement parameters into the dynamic threshold prediction model to generate the corresponding CT value distribution reference interval for each anatomical partition image; The hemorrhage region identification module is used to mark the hemorrhage region in the dynamic multi-phase brain CT scan image based on the CT value distribution reference interval, and output the image processing result containing the location information of the hemorrhage region.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the method for identifying hemorrhage regions based on brain images as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for identifying hemorrhage regions based on brain images as described in any one of claims 1 to 7.

11. A computer program product, characterized in that, The method includes a computer program that, when executed by a processor, implements the method for identifying hemorrhage regions based on brain images as described in any one of claims 1 to 7.

Images