Brain hemorrhage DTI image matching method and system based on multi-modal fusion empowerment CT

Abstract

The application provides a CT-based brain hemorrhage DTI image matching method and system based on multi-modal fusion, and relates to the technical field of medical image processing and analysis, wherein a multi-modal feature database is constructed according to physiological parameter information, skull feature information and DTI images of multiple stroke individuals, and a candidate individual is quickly and accurately matched from the multi-modal feature database according to the physiological parameter information and CT images of a target individual through a first difference index and a second difference index, the first difference index is used to represent the difference degree between the skull feature information, and the second difference index is used to represent the difference degree between the physiological parameter information. The DTI image of the candidate individual is determined as the DTI image of the target individual, and then the self-limitation and equipment limitation of the target individual can be eliminated, the DTI image can be quickly determined only through the CT image of the target individual, and the diagnosis and treatment time can be avoided due to the acquisition delay of the DTI image, and the diagnosis and treatment efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical image processing and analysis, and particularly relates to a CT brain hemorrhage DTI image matching method and system based on multi-modal fusion empowerment. BACKGROUND

[0002] In the clinical diagnosis and treatment of neurological diseases, the DTI (diffusion tensor imaging) technology can directly show the shape and integrity of the white matter fiber bundle, and is an important tool for evaluating the damage degree of the corticospinal tract (CST) caused by diseases such as cerebral hemorrhage.

[0003] However, the DTI scan usually needs more than 30 minutes of acquisition time, and due to the low coverage of the current detection equipment, it is limited by the popularization degree of 3.0T high-field MRI, and in addition, individuals may be restricted by their own conditions such as physical function, metal implants or claustrophobia, it is difficult to quickly obtain high-quality DTI images, which has the defect of long acquisition time, which seriously affects the timing of diagnosis and treatment.

[0004] Therefore, there is an urgent need for a CT brain hemorrhage DTI image matching method and system based on multi-modal fusion empowerment, which can quickly and accurately obtain the DTI image of the target individual, reduce the acquisition time, and improve the acquisition efficiency. Avoid the impact of delayed DTI image acquisition on diagnosis and treatment time, and improve the diagnosis and treatment efficiency. SUMMARY

[0005] The present application provides a CT brain hemorrhage DTI image matching method and system based on multi-modal fusion empowerment, which can quickly and accurately match the DTI image of the target individual based on the pre-constructed database, can effectively eliminate the self-condition restriction and equipment restriction of the target individual, reduce the time of obtaining the DTI image, and improve the acquisition efficiency. Avoid the impact of delayed DTI image acquisition on diagnosis and treatment time, and improve the diagnosis and treatment efficiency.

[0006] To achieve the above purpose, the embodiments of the present application adopt the following technical solutions:

[0007] In a first aspect, a CT-based brain hemorrhage DTI image matching method based on multi-modal fusion is provided. The method comprises: obtaining physiological parameter information, CT images and DTI images of a head region of a plurality of stroke individuals, the physiological parameter information including age, gender, onset duration, hematoma location, hematoma shape and hematoma volume; determining skull feature information of each of the plurality of stroke individuals according to the CT images of the head region of the plurality of stroke individuals, the skull feature information including the thickness of different parts of the skull and the three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system; constructing a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals; determining the DTI image of a candidate individual as the DTI image of a target individual from the multi-modal feature database based on the physiological parameter information and the skull feature information of the target individual, the candidate individual being one of the plurality of stroke individuals, a first difference index between the target individual and the candidate individual being less than a preset threshold, and a second difference index between the target individual and the candidate individual being the minimum value of a plurality of second difference indexes between the target individual and the plurality of stroke individuals, the first difference index being positively correlated with the difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index being positively correlated with the difference degree of the physiological parameter information between the target individual and the candidate individual.

[0008] The method provided by the present application constructs a multi-modal feature database according to the physiological parameter information, the skull feature information and the DTI images of a plurality of stroke individuals, and quickly and accurately matches a candidate individual from the multi-modal feature database according to the physiological parameter information and the CT image of a target individual through a first difference index and a second difference index, wherein the first difference index is used to represent the difference degree between the skull feature information, and the second difference index is used to represent the difference degree between the physiological parameter information. Determining the DTI image of the candidate individual as the DTI image of the target individual can effectively eliminate the self-condition limitation and the equipment limitation of the target individual, reduce the time for obtaining the DTI image, and improve the acquisition efficiency. The method avoids the influence of the delay in obtaining the DTI image on the diagnosis and treatment time, and improves the diagnosis and treatment efficiency.

[0009] In a possible implementation manner of the first aspect, the different parts of the skull include frontal bone, parietal bone, temporal bone, occipital bone and sphenoid bone, and the landmark points include the front point, the back point, the left temporal point, the right temporal point, the vertex point and the cranial base point of the skull cavity, and the hemorrhage point, the interbrow point, the fontanel point, the person point, the center point of the external auditory meatus and the occipital external protuberance point.

[0010] The method provided by the application can quickly and accurately determine the thickness of different parts of the skull and the three-dimensional coordinates of different mark points by limiting the composition and mark points of the skull, effectively reduces the ambiguity of feature extraction, accurately obtains the skull feature information of different individuals, and further improves the matching degree between the DTI image of the candidate individual and the DTI image of the target individual, thereby improving the matching accuracy.

[0011] In a possible implementation form of the first aspect, the skull feature information of each of the plurality of stroke individuals is determined according to the CT images of the head regions of the plurality of stroke individuals, including: determining a skull binary image of each of the plurality of stroke individuals according to the CT images of the head regions of the plurality of stroke individuals; determining a skull three-dimensional model corresponding to the skull binary image of each of the plurality of stroke individuals based on a three-dimensional reconstruction algorithm; determining the thickness of different parts of the skull according to the distance between each point corresponding to the three-dimensional profile of each part of the skull on the skull three-dimensional model and the center point of the skull three-dimensional model; and determining the three-dimensional coordinates of each mark point of the skull in a preset spatial rectangular coordinate system, wherein the center point of the skull three-dimensional model is the origin of the preset spatial rectangular coordinate system.

[0012] The method provided by the application can ensure that the skull features of different individuals have a unified reference by converting the CT images into skull binary images and then establishing a skull three-dimensional model corresponding to each skull binary image, and eliminate the influence of positioning deviation on thickness calculation and mark point coordinate extraction; the method provided by the application can realize quantitative description of the skull morphology based on three-dimensional profile calculation of the skull thickness and accurate coordinates of the mark points, provide a high-precision data basis for subsequent difference calculation, and further improve the reliability and accuracy of matching.

[0013] In a possible implementation form of the first aspect, the determination formula of the first difference index D1 is:

[0014] D1=α·D thickness +(1-α)·D landmark ;

[0015] Wherein, α is a total weight coefficient, D thickness is a skull thickness difference score, and D landmark is a mark point position difference score; the determination formula of the skull thickness difference score D thickness is:

[0016]

[0017] N is the number of parts of the skull, T target,i is the thickness of the i-th part of the skull of the target individual; and T candidate,i is the thickness of the i-th part of the skull of the stroke individual.

[0018] landmark position difference score D landmark The determination formula is:

[0019]

[0020] M is the number of landmarks, P target,j is the three-dimensional coordinate of the jth landmark of the target individual; P candidate,j is the three-dimensional coordinate of the jth landmark of the stroke individual, normalize_factor j is a normalization factor.

[0021] The method provided by the application separates the skull feature difference into thickness difference and landmark position difference, and obtains a first difference index by weighted summation. Among them, the thickness difference is calculated by comparing the thickness of each skull part of the target individual and the candidate individual, and the landmark difference is evaluated based on the Euclidean distance of the three-dimensional coordinates, and a normalization factor is introduced to eliminate the influence of the measurement scale. In this way, the different dimensions of the skull features have different contribution degrees to the matching result, and separating the thickness and landmark difference can realize fine evaluation; the weighted summation gives reasonable weight to both, so that the method provided by the application can comprehensively consider the double differences of shape and spatial position; the normalization processing ensures that the difference scores of different landmarks are comparable, avoids that some features are overestimated or ignored due to the scale problem, and finally realizes scientific quantification and accurate matching of the skull feature difference.

[0022] In a possible implementation manner of the first aspect, different parts of the skull are respectively configured with a first weight coefficient, and each landmark is respectively configured with a second weight coefficient;

[0023] Skull thickness difference score D thickness The determination formula is:

[0024]

[0025] Landmark position difference score D landmark The determination formula is:

[0026]

[0027] Among them, w i is the first weight coefficient of the i th part of the skull; v i is the second weight coefficient of the i th landmark.

[0028] The method provided by the application assigns different weights to different parts and landmark points of the skull, thereby improving the determination mode of the thickness difference and the landmark point position difference, and enhancing the matching priority of key structures and weakening the interference of secondary features by weight configuration in the case that different skull regions and landmark points have different importance in the DTI image. Meanwhile, the adjustability of the weight makes the algorithm adapt to different use scenarios, and improves the flexibility and practicality.

[0029] In a possible implementation of the first aspect, a determination formula of the preset threshold μ is as follows:

[0030] μ = μ 0 · ε ;

[0031] wherein μ 0 is a skull difference degree reference threshold, and ε is a skull complexity weight.

[0032] In the case that the comprehensive complexity index of the target user is less than 0.7, the skull complexity weight ε = 0.8; in the case that the comprehensive complexity index of the target user is greater than or equal to 0.7 and less than 1.3, the skull complexity weight ε = 1; and in the case that the comprehensive complexity index of the target user is greater than 1.3, the skull complexity weight ε = 1.2.

[0033] A determination formula of the comprehensive complexity index v is as follows:

[0034]

[0035] wherein A is a balance coefficient, D ref is a dispersion reference value; CV thickness is a skull thickness variation coefficient, η thickness is a standard deviation of different parts of the skull, σ thickness is a mean value of different parts of the skull, DV landmark is a skull landmark point dispersion; P j is a three-dimensional coordinate of the j th landmark point of the target user, and P is a three-dimensional coordinate of a geometric center of all landmark points of the target user.

[0036] The method provided by the application can dynamically adjust the preset threshold based on the skull complexity of the target individual, calculate the skull thickness variation coefficient and the landmark dispersion of the target individual, construct a comprehensive complexity index, and then map different skull complexity weights according to the index interval, so as to realize the dynamic adjustment of the preset threshold of the target individual. The logic is that the individual difference of skull morphology naturally exists, and the fixed threshold cannot adapt to the matching demand of complex skull structure; the dynamic threshold mechanism quantifies the skull complexity, adopts a strict threshold for individuals with simple structure to ensure high matching accuracy, and appropriately relaxes the standard for complex individuals to avoid no available matching cases due to too strict threshold. This adaptive adjustment strategy expands the application range while ensuring the matching quality, and significantly improves the success rate and feasibility of matching, especially when dealing with special individuals.

[0037] In a possible implementation of the first aspect, a determination formula of the second difference index D2 is:

[0038]

[0039] K is the number of parameters included in the physiological parameter information; β k is a weight coefficient of the kth parameter included in the physiological parameter information, Score k is a difference score of the kth parameter included in the physiological parameter information.

[0040] The difference score Scoreage of the age included in the physiological parameter information is: age

[0041]

[0042] A target is the age of the target individual, A candidate is the age of the stroke individual, A max and A min are the maximum and minimum values of the ages of the plurality of stroke individuals.

[0043] The difference score Scoregender of the gender included in the physiological parameter information is: gender Scoregender=0 when the target individual and the stroke individual have the same gender; and Scoregender=1 when the target individual and the stroke individual have different genders. gender

[0044] The difference score Scoreduration of the duration of onset included in the physiological parameter information is: time

[0045]

[0046] T target is the duration of onset of the target individual; T​​​candidate the onset length of the target individual;

[0047] T max and T min are the maximum and minimum of the onset lengths of the plurality of stroke individuals;

[0048] the difference score Score of the hematoma location included in the physiological parameter information location is: Score = 0 in the case that the hematoma region of the target individual and the stroke individual are located in the same region; Score = 0.5 in the case that the hematoma region of the target individual and the stroke individual are located in adjacent regions; Score = 1 in the case that the hematoma region of the target individual and the stroke individual are located in different and non-adjacent regions; location location location

[0049] the difference score Score of the hematoma shape included in the physiological parameter information shape is:

[0050]

[0051] L target is the length of the long axis of the hematoma of the target individual, L candidate is the length of the long axis of the hematoma of the stroke individual; L max and L min are the maximum and minimum of the lengths of the long axis of the hematoma of the plurality of stroke individuals; W target is the length of the short axis of the hematoma of the target individual, W candidate is the length of the short axis of the hematoma of the stroke individual; W max and W min are the maximum and minimum of the lengths of the short axis of the hematoma of the plurality of stroke individuals; H target is the height of the hematoma of the target individual, H candidate is the height of the hematoma of the stroke individual; H max and H min are the maximum and minimum of the heights of the hematoma of the plurality of stroke individuals;

[0052] the difference score Score of the hematoma volume included in the physiological parameter information volume is:

[0053]

[0054] V target is the volume of the hematoma of the target individual; V candidate is the volume of the hematoma of the stroke individual.

[0055] ​​​The method provided by the application quantifies the differences of physiological parameters such as age, gender, and onset time, and obtains a second difference index by weighted summation. The parameters such as age and onset time are calculated by normalized difference score, the gender is evaluated by 0-1 binary, and the hematoma position is scored according to regional correlation. The physiological parameters and skull characteristics are complementary, and the single skull matching has poor accuracy. By quantifying the differences of each parameter, the method provided by the application can evaluate the similarity of the physiological parameter dimension of the candidate individual and the target individual, and can effectively eliminate the self-condition limitation and equipment limitation of the target individual, reduce the time of obtaining the DTI image, and improve the acquisition efficiency. The method avoids the influence of the delay of obtaining the DTI image on the diagnosis and treatment time, and improves the diagnosis and treatment efficiency.

[0056] In a possible implementation manner of the first aspect, the method further includes: determining a timeliness weight coefficient corresponding to each stroke individual according to the storage time of the skull characteristic information, the physiological parameter information, and the DTI image of each stroke individual in the multi-modal feature database;

[0057] The timeliness weight coefficient γ of the i th stroke individual is determined as follows: norm，i The determination formula of the timeliness weight coefficient γ of the i th stroke individual is as follows:

[0058]

[0059] Wherein, t is the current time, t0 is the storage time, and λ is the decay coefficient.

[0060] The determination formula of the second difference index D2 is as follows:

[0061]

[0062] Wherein, θ is a basic difference threshold.

[0063] The method provided by the application introduces a timeliness weight coefficient, calculates the decay degree according to the data storage time, and adjusts the second difference index in combination with a basic difference threshold. The old data weight is reduced by an exponential decay formula, the algorithm preferentially selects recent matching cases, and the timeliness of the result is ensured. At the same time, the constraint of the basic threshold avoids ignoring the essential difference of the physiological parameters due to excessive dependence on new data, for example, prevents new data from being misjudged as the best match due to data acquisition error. The dynamic balance of new and old data and matching quality is realized by combination of the two, and the timeliness and reliability of the matching result are improved.

[0064] In a second aspect, the present application provides a CT-based human brain hemorrhage DTI image matching system based on multi-modal fusion, comprising: a data acquisition module, configured to acquire physiological parameter information, CT images and DTI images of a head region of a plurality of stroke individuals, wherein the physiological parameter information comprises age, gender, onset duration, hematoma position, hematoma shape and hematoma volume; an information determination module, configured to determine skull feature information of each of the plurality of stroke individuals according to the CT images of the head region of the plurality of stroke individuals, wherein the skull feature information comprises thicknesses of different parts of the skull and three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system; a database construction module, configured to construct a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals; and an image matching module, configured to determine, based on physiological parameter information and skull feature information of a target individual, a DTI image of a candidate individual as the DTI image of the target individual from the multi-modal feature database, wherein the candidate individual is one of the plurality of stroke individuals, a first difference index between the target individual and the candidate individual is less than a preset threshold, and a second difference index between the target individual and the candidate individual is the minimum value of a plurality of second difference indexes between the target individual and the plurality of stroke individuals, the first difference index is positively correlated with a difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index is positively correlated with a difference degree of the physiological parameter information between the target individual and the candidate individual.

[0065] In a third aspect, an electronic device is provided, comprising a memory and one or more processors; the memory is coupled to the processors; and the memory stores computer program codes including computer instructions, which, when executed by the processors, cause the electronic device to perform the method in any implementation manner of the first aspect.

[0066] In a fourth aspect, a computer-readable storage medium is provided, comprising computer instructions, which, when executed on an electronic device, cause the electronic device to perform the method in any implementation manner of the first aspect.

[0067] In a fifth aspect, a computer program product is provided, which, when executed on a computer, causes the computer to perform the method in any implementation manner of the first aspect.

[0068] It can be understood that the system of the second aspect, the electronic device of the third aspect, the computer-readable storage medium of the fourth aspect, and the computer program product of the fifth aspect provided above can achieve the beneficial effects as described in the first aspect and any possible design manner thereof, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1A structural schematic diagram of an electronic device provided by an embodiment of the present application is shown in FIG. 1.

[0070] Figure 2 A flowchart of a CT-based brain hemorrhage DTI image matching method based on multi-modal fusion enabled by an embodiment of the present application is shown in FIG. 2.

[0071] Figure 3 A flowchart of another CT-based brain hemorrhage DTI image matching method based on multi-modal fusion enabled by an embodiment of the present application is shown in FIG. 3.

[0072] Figure 4 An application scenario diagram of a CT-based brain hemorrhage DTI image matching method based on multi-modal fusion enabled by an embodiment of the present application is shown in FIG. 4.

[0073] Figure 5 A structural schematic diagram of an image matching system provided by an embodiment of the present application is shown in FIG. 5. DETAILED DESCRIPTION

[0074] The technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings. In the description of the present application, unless otherwise specified, " / " represents an "or" relationship between the objects before and after the " / " symbol, for example, A / B can represent A or B; in the present application, "or" is only a description of the relationship between the associated objects, and can represent three relationships, for example, A or B, which can represent three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, in the description of the present application, unless otherwise specified, "multiple" means two or more than two. "At least one of the following" or similar expressions means any combination of these items, including any combination of single item or multiple items.

[0075] In addition, in order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, "first", "second", etc. are used to distinguish the same items or similar items with basically the same function and effect. Those skilled in the art can understand that "first", "second", etc. do not limit the quantity and execution order, and "first", "second", etc. also do not necessarily mean different.

[0076] Meanwhile, in the embodiments of the present application, the words "exemplary" or "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as more excellent or more advantageous than other embodiments or design schemes. Rather, the words "exemplary" or "for example" are intended to present the relevant concept in a specific manner, and to facilitate understanding.

[0077] In the clinical diagnosis and treatment of neurological diseases, DTI (diffusion tensor imaging) technology can directly show the shape and integrity of white matter fiber bundles, and is an important tool for evaluating the damage degree of the corticospinal tract (CST) in diseases such as cerebral hemorrhage.

[0078] However, DTI scanning usually requires more than 30 minutes of acquisition time, and due to the low coverage of current detection equipment, it is limited by the popularity of 3.0T high-field MRI, and individuals may be limited by their own conditions such as physical function, metal implants, or claustrophobia, making it difficult to quickly obtain high-quality DTI images, which has the disadvantage of long acquisition time, which seriously affects the timing of diagnosis and treatment.

[0079] Therefore, there is an urgent need for a CT-enabled brain hemorrhage DTI image matching method based on multi-modal fusion, which can quickly and accurately obtain the DTI image of the target individual, reduce the acquisition time, and improve the acquisition efficiency. Avoid the impact of DTI image acquisition delay on diagnosis and treatment time, and improve the efficiency of diagnosis and treatment.

[0080] Therefore, there is an urgent need for a CT-enabled brain hemorrhage DTI image matching method based on multi-modal fusion, which can quickly and accurately obtain the DTI image of the target individual, reduce the acquisition time, and improve the acquisition efficiency. Avoid the impact of DTI image acquisition delay on diagnosis and treatment time, and improve the efficiency of diagnosis and treatment.

[0081] The method provided by the application constructs a multi-modal feature database according to physiological parameter information, skull feature information and DTI images of a plurality of stroke individuals, and quickly and accurately matches a candidate individual from the multi-modal feature database according to physiological parameter information and CT images of a target individual by using a first difference index and a second difference index, wherein the first difference index is used to represent the difference degree between the skull feature information, and the second difference index is used to represent the difference degree between the physiological parameter information. The DTI image of the candidate individual is determined as the DTI image of the target individual, which can effectively eliminate the self-condition limitation and device limitation of the target individual, reduce the time for obtaining the DTI image, and improve the acquisition efficiency. The method avoids the influence of the delay in obtaining the DTI image on the diagnosis and treatment time, and improves the diagnosis and treatment efficiency.

[0082] In some embodiments, the CT-based human brain hemorrhage DTI image matching method based on multi-modal fusion can be executed by a CT-based human brain hemorrhage DTI image matching system 100 (hereinafter referred to as the image matching system 100).

[0083] As an example, the image matching system 100 can be any electronic device 200 with data processing capability, such as a general-purpose computer, a personal computer, a notebook computer, a switch or a tablet computer, etc. The specific implementation of the image matching system 100 is not limited here.

[0084] Figure 1 A hardware structure schematic diagram of an electronic device provided by the embodiments of the application is shown. The electronic device 200 includes a processor 210, a memory 220 and a communication interface 230.

[0085] The processor 210 can include one or more processing cores. The processor 210 connects various parts in the electronic device 200 by using various interfaces and lines, executes various functions of the electronic device 200 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 220, and calling data stored in the memory 220. Optionally, the processor 210 can be implemented in at least one of a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processing (DSP), a field-programmable gate array (FPGA) and a programmable logic array (PLA).

[0086] The memory 220 can include a random access memory (RAM) and can also include a read-only memory (ROM). Optionally, the memory 220 includes a non-transitory computer-readable storage medium. The memory 220 can be used to store instructions, programs, codes, code sets, or instruction sets. The memory 220 can include a program storage area. The program storage area can store instructions for implementing an operating system, instructions for implementing at least one function (such as a data acquisition function, a database construction function, and an image matching function), instructions for implementing each of the above-mentioned method embodiments, and the like.

[0087] The communication interface 230 is configured to communicate with other devices, equipment, or communication networks, such as data storage devices, image processing equipment, or Ethernet, a radio access network (RAN), a wireless local area network (WLAN), and the like.

[0088] In terms of physical implementation, each of the above-mentioned devices (such as the processor 210, the memory 220, and the communication interface 230) can be a device in the same device (such as a notebook computer). Alternatively, at least two of the devices can be arranged in the same device as different devices in the device, such as a deployment manner similar to that of devices or components in a distributed system.

[0089] It can be understood that the structure illustrated in the embodiment does not constitute a specific limitation on the electronic device 200. In other embodiments of the present application, the electronic device 200 can include more or fewer components than those illustrated, or combine certain components, or split certain components, or different component arrangements. The illustrated components can be implemented in hardware, software, or a combination of software and hardware.

[0090] An image matching method for CT-based multi-modal fusion-enabled brain hemorrhage DTI provided by an embodiment of the present application is described below in conjunction with the accompanying drawings.

[0091] Figure 2 A flowchart of an image matching method for CT-based multi-modal fusion-enabled brain hemorrhage DTI provided by an embodiment of the present application is shown. Optionally, the method can be executed by the electronic device 200 shown in the figure. The method can include the following steps: Figure 1

[0092] S1, physiological parameter information, CT images, and DTI images of a plurality of stroke individuals are obtained. The physiological parameter information includes age, gender, onset time, hematoma location, hematoma shape, and hematoma volume.​

[0093] Specifically, the hematoma position, the hematoma shape and the hematoma volume of the plurality of stroke individuals are determined based on CT images of head regions of the plurality of stroke individuals, and specific determination manners of the hematoma position, the hematoma shape and the hematoma volume are not particularly limited in the embodiments of the present application.

[0094] S2, determine skull feature information of each of the plurality of stroke individuals according to CT images of head regions of the plurality of stroke individuals.

[0095] Specifically, the skull feature information includes thicknesses of different parts of the skull and three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system.

[0096] In a possible implementation, the different parts of the skull include frontal bone, parietal bone, temporal bone, occipital bone and sphenoid bone, and the landmark points include an anterior point, a posterior point, a left temporal point, a right temporal point, a vertex point and a skull base point of a skull cavity, and a bleeding point, a glabella point, a fontanel point, a person point, an external auditory meatus center point and an external occipital protuberance point.

[0097] Specifically, the anterior point is the most anterior point of the skull cavity corresponding to the frontal bone, the posterior point is the most posterior point of the skull cavity corresponding to the occipital bone, the left temporal point and the right temporal point are the most lateral points of the skull cavity corresponding to the left temporal bone and the right temporal bone. The vertex point is the highest point of the skull cavity corresponding to the parietal bone. The skull base point is the lowest point of the skull cavity corresponding to the sphenoid bone.

[0098] It should be understood that the above landmark points are only exemplary, and more or fewer landmark points can be set by an operator based on different use scenarios, and the embodiments of the present application do not particularly limit this.

[0099] The method provided by the embodiments of the present application can quickly and accurately determine the thicknesses of different parts of the skull and the three-dimensional coordinates of different landmark points by limiting the constituent parts and the landmark points of the skull, can effectively reduce the ambiguity of feature extraction, can accurately acquire the skull feature information of different individuals, and can further improve the matching degree between the DTI image of the candidate individual and the DTI image of the target individual, thereby improving the accuracy of matching.

[0100] In some embodiments, S2 includes:

[0101] determine a skull binary image of each of the plurality of stroke individuals according to CT images of head regions of the plurality of stroke individuals; determine a skull three-dimensional model corresponding to the skull binary image of each of the plurality of stroke individuals based on a three-dimensional reconstruction algorithm; determine thicknesses of different parts of the skull according to distances between points corresponding to the different parts of the skull on the skull three-dimensional model and a center point of the skull three-dimensional model; and determine three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system, wherein the preset spatial rectangular coordinate system takes the center point of the skull three-dimensional model as an origin.

[0102] According to the CT image of the head region of the plurality of cerebral apoplexy individuals, the skull binary image of each cerebral apoplexy individual is determined; based on a three-dimensional reconstruction algorithm, the skull three-dimensional model corresponding to the skull binary image of each cerebral apoplexy individual is determined, including:

[0103] The CT image of the head region of the plurality of cerebral apoplexy individuals is denoised, a median filter algorithm is used to remove noise in the image, and a Gaussian filter is used to smooth the image. At the same time, a histogram equalization technique is used to enhance the contrast of the image, expand the gray dynamic range of the skull tissue and the surrounding soft tissue, air and other regions in the CT image, so that the skull part is more clear and distinguishable in the image. A threshold segmentation method is used to separate the skull tissue from the CT image according to the specific gray value range of the skull tissue in the CT image to obtain an initial skull region. A region growing algorithm is used to select a point with uniform gray in the skull region as a seed point, and according to a set gray similarity threshold, the surrounding pixel points that meet the conditions are gradually merged into the skull region. The segmented skull region is binarized, the pixel points corresponding to the skull tissue are assigned a value of 1, and the background region is assigned a value of 0 to form a binary image containing only two states of skull and background. Morphological operations are used to remove small burrs and isolated noise points on the edge of the skull region by erosion operation, and to fill smaller holes in the interior by dilation operation to smooth the edge of the skull profile. Based on the processed skull binary image, a three-dimensional reconstruction algorithm is used to construct a skull three-dimensional model according to the two-dimensional skull binary image. The geometric center point of the skull three-dimensional model is taken as the origin of the preset spatial rectangular coordinate system, and according to the anatomical standard direction of the skull, the front-back direction of the skull is set as the X axis, the left-right direction is set as the Y axis, and the up-down direction is set as the Z axis.

[0104] The method provided in the embodiments of the present application converts the CT image into a skull binary image, and then establishes a skull three-dimensional model corresponding to each skull binary image, so as to ensure that the skull features of different individuals have a unified reference, and eliminate the influence of positioning deviation on thickness calculation and landmark point coordinate extraction. The method provided in the embodiments of the present application calculates the skull thickness based on a three-dimensional profile, and combines the accurate coordinates of the landmark points, so as to realize quantitative description of the skull shape, provide a high-precision data basis for subsequent difference calculation, and further improve the reliability and accuracy of matching.

[0105] S3, constructing a multi-modal feature database according to the skull feature information, physiological parameter information and DTI image of the plurality of cerebral apoplexy individuals.

[0106] S4, determining the DTI image of the candidate individual as the DTI image of the target individual from the multi-modal feature database based on the physiological parameter information and the skull feature information of the target individual.

[0107] Specifically, the candidate individual is one of the plurality of stroke individuals, the first difference index between the target individual and the candidate individual is less than a preset threshold, and the second difference index between the target individual and the candidate individual is the minimum value of a plurality of second difference indexes between the target individual and the plurality of stroke individuals. The first difference index is positively correlated with the difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index is positively correlated with the difference degree of the physiological parameter information between the target individual and the candidate individual.

[0108] In one example, the preset threshold has a value range of greater than or equal to 0.15 and less than or equal to 0.3.

[0109] In some embodiments, referring to Figure 3 , the S4 includes:

[0110] S41, based on the skull feature information of the target individual, determining an initial candidate individual set from the multi-modal feature database, and the first difference index between the target individual and each initial candidate individual in the initial candidate individual set is less than a preset threshold.

[0111] In one possible implementation, the determination formula of the first difference index D1 is:

[0112] D1 = a D thickness + (1-a) D landmark ;

[0113] Wherein, a is an overall weight coefficient, D thickness is a skull thickness difference score, and D landmark is a landmark position difference score; in one example, the overall weight coefficient a is 0.4.

[0114] The determination formula of the skull thickness difference score D thickness is:

[0115]

[0116] N is the number of parts constituting the skull, T target,i is the thickness of the i-th part of the target individual constituting the skull; T candidate,i is the thickness of the i-th part of the stroke individual constituting the skull.

[0117] The determination formula of the landmark position difference score D landmark is:

[0118]

[0119] M is the number of landmarks, P target,j is the three-dimensional coordinates of the j-th landmark of the target individual; P candidate,ja three-dimensional coordinate of a jth landmark point of the individual with stroke, normalize_factor j is a normalization factor.

[0120] Specifically, the normalization factor is the overall size of the skull or the average distance from the landmark point to the center of the skull, which can avoid the influence of scale on different individuals.

[0121] The method provided by the embodiment of the application separates the skull feature difference into thickness difference and landmark point position difference, and obtains a first difference index through weighted summation. The thickness difference is calculated by comparing the thickness of each skull part of the target individual and the candidate individual, and the landmark point difference is evaluated based on the Euclidean distance of the three-dimensional coordinates, and a normalization factor is introduced to eliminate the influence of the measurement scale. In this way, the different dimensions of the skull features have different contribution degrees to the matching result, and separating the thickness and landmark point difference can achieve fine evaluation; the weighted summation gives reasonable weights to the two, so that the method provided by the embodiment of the application can comprehensively consider the double differences of the shape and the spatial position; the normalization processing ensures that the difference scores of different landmark points are comparable, avoids that some features are overestimated or ignored due to the scale problem, and finally realizes the scientific quantification and accurate matching of the skull feature difference.

[0122] In another possible implementation manner, different parts of the skull are respectively configured with first weight coefficients, and each landmark point is respectively configured with a second weight coefficient.

[0123] In one example, the first weight coefficient of the frontal bone is 0.25, the first weight coefficient of the parietal bone is 0.25, the first weight coefficient of the temporal bone is 0.2, the first weight coefficient of the occipital bone is 0.2, and the first weight coefficient of the sphenoid bone is 0.10.

[0124] In another example, the second weight coefficient of the anterior point of the skull cavity is 0.1, the second weight coefficient of the posterior point is 0.1, the second weight coefficient of the left temporal point is 0.07, the second weight coefficient of the right temporal point is 0.07, the second weight coefficient of the vertex point is 0.08, the second weight coefficient of the skull base point is 0.03, the second weight coefficient of the bleeding point is 0.2, the second weight coefficient of the glabella point is 0.1, the second weight coefficient of the anterior fontanel point is 0.05, the second weight coefficient of the metopic point is 0.05, the second weight coefficient of the external auditory meatus center point is 0.1, and the second weight coefficient of the external occipital protuberance point is 0.05.

[0125] The skull thickness difference score D thickness is determined by the following formula:

[0126]

[0127] The determination formula of the landmark point position difference score D landmark is as follows:

[0128]

[0129] wherein w i is a first weight coefficient of the i-th part of the skull; v i is a second weight coefficient of the i-th landmark point.

[0130] The method provided by the embodiments of the present application assigns different weights to different parts of the skull and landmark points, thereby improving the determination manner of the thickness difference and the landmark position difference, and enhancing the matching priority of the key structure and weakening the interference of the secondary feature by weight configuration in the case that the importance of different skull regions and landmark points in the DTI image is different. Meanwhile, the adjustability of the weight enables the algorithm to adapt to different use scenarios, thereby improving the flexibility and practicability.

[0131] Due to the difference between different individuals, the same preset threshold leads to the decline of the matching accuracy.

[0132] To solve this problem, in some embodiments, the method provided by the present application further comprises:

[0133] determining the preset threshold corresponding to the target individual.

[0134] Specifically, the determination formula of the preset threshold μ is:

[0135] μ=μ0·ε;

[0136] wherein μ0 is a skull difference degree reference threshold, and ε is a skull complexity weight.

[0137] In the case that the comprehensive complexity index of the target user is less than 0.7, the skull complexity weight ε=0.8, in the case that the comprehensive complexity index of the target user is greater than or equal to 0.7 and less than 1.3, the skull complexity weight ε=1, and in the case that the comprehensive complexity index of the target user is greater than 1.3, the skull complexity weight ε=1.2.

[0138] The determination formula of the comprehensive complexity index τ is:

[0139]

[0140]

[0141] wherein A is a balance coefficient, Dref is a dispersion reference value; CV thickness is a skull thickness variation coefficient, η thickness is a standard deviation of different parts of the skull, σ thickness is a mean value of different parts of the skull, DV landmark is a skull landmark point dispersion; P jP is the three-dimensional coordinate of the geometric center of all landmark points of the target user.

[0142] The method provided by the embodiment of the application can dynamically adjust the preset threshold based on the skull complexity of the target individual, construct a comprehensive complexity index by calculating the skull thickness coefficient of variation and the landmark point dispersion of the target individual, and then map different skull complexity weights according to the index interval, so as to realize dynamic adjustment of the preset threshold of the target individual. The logic is that individual differences in skull morphology naturally exist, and a fixed threshold cannot adapt to the matching requirements of complex skull structures; the dynamic threshold mechanism quantifies the skull complexity, uses a strict threshold for individuals with simple structures to ensure high matching accuracy, and appropriately relaxes the standard for complex individuals to avoid no available matching cases due to too strict threshold. This adaptive adjustment strategy expands the application range while ensuring matching quality, and significantly improves the success rate and feasibility of matching, especially when dealing with special individuals.

[0143] In S42, a second difference index between the target individual and each initial candidate individual in the initial candidate individual set is determined based on the physiological parameter information of the target individual, the initial candidate individual with the minimum second difference index is determined as the candidate individual, and the DTI image of the candidate individual is determined as the DTI image of the target individual.

[0144] In a possible implementation, the determination formula of the second difference index D2 is as follows:

[0145]

[0146] K is the number of parameters included in the physiological parameter information; β k is the weight coefficient of the kth parameter included in the physiological parameter information, Score k is the difference score of the kth parameter included in the physiological parameter information.

[0147] Specifically, the weight coefficient of the age is 0.2, the weight coefficient of the gender is 0.1, the weight coefficient of the onset duration is 0.25, the weight coefficient of the hematoma position is 0.2, the weight coefficient of the hematoma shape is 0.1, and the weight coefficient of the hematoma volume is 0.15.

[0148] The difference score Score age of the age included in the physiological parameter information is as follows:

[0149]

[0150] A target is the age of the target individual, A candidate is the age of the stroke individual, A max and A minmax and min of the age of the plurality of stroke individuals;

[0151] a difference score Scoregender included in the physiological parameter information gender is:

[0152] Scoregender = 0 in case the target individual and the stroke individual are of the same gender; gender

[0153] Scoregender = 1 in case the target individual and the stroke individual are of different gender;

[0154] a difference score ScoreT included in the physiological parameter information time is:

[0155]

[0156] T target is the onset time of the target individual; T candidate is the onset time of the stroke individual;

[0157] T max and T min are max and min of the onset time of the plurality of stroke individuals;

[0158] a difference score Scorehematoma included in the physiological parameter information location is:

[0159] Scorehematoma = 0 in case the target individual and the stroke individual have hematoma in the same region; location

[0160] Scorehematoma = 0.5 in case the target individual and the stroke individual have hematoma in adjacent regions; location

[0161] Scorehematoma = 1 in case the target individual and the stroke individual have hematoma in different and non-adjacent regions; location

[0162] a difference score Scoreshape included in the physiological parameter information shape is:

[0163]

[0164] L target is the length of the long axis of the hematoma of the target individual; L candidate is the length of the long axis of the hematoma of the stroke individual; L max min are max and min of the length of the long axis of the hematoma of the plurality of stroke individuals; W​​​​​target W is a hematoma short axis length of the target individual; candidate W is a hematoma short axis length of the stroke individual; max W and W min H and H target H is a hematoma height of the target individual; candidate H is a hematoma height of the stroke individual; max H and H min H and H

[0165] The physiological parameter information includes a difference score Score of a hematoma volume volume is:

[0166]

[0167] V target V is a hematoma volume of the target individual; candidate V is a hematoma volume of the stroke individual.

[0168] As can be seen from the above, the method provided by the embodiments of the present application quantifies the differences in physiological parameters such as age, gender, and onset time, and obtains a second difference index by weighted summation. Among them, the parameters such as age and onset time are calculated by difference score through normalization, gender is evaluated by 0-1 binary, and the hematoma position is scored according to regional correlation. The physiological parameters and the skull characteristics are complementary, and the single skull matching has the problem of poor accuracy; by quantifying the differences of each parameter, the method provided by the present application can evaluate the similarity of the candidate individual and the target individual in the physiological parameter dimension, and can effectively eliminate the self-condition limitation and the equipment limitation of the target individual, reduce the time of obtaining the DTI image, and improve the acquisition efficiency. Avoid the influence of the delay of the acquisition of the DTI image on the diagnosis and treatment time, and improve the diagnosis and treatment efficiency.

[0169] In another possible implementation manner, the method provided by the embodiments of the present application further includes:

[0170] According to the storage time of the skull characteristic information, the physiological parameter information and the DTI image of each stroke individual in the multi-modal feature database, the time-effectiveness weight coefficient corresponding to each stroke individual is determined;

[0171] The time-effectiveness weight coefficient γ of the i-th stroke individual is determined according to the following formula: norm，i

[0172]

[0173] Wherein, t is the current time, t0 is the storage time, and λ is the decay coefficient.

[0174] ​Further, the determination formula of the second difference index D2 is:

[0175]

[0176] Wherein, θ is a basic difference threshold.

[0177] In one example, the attenuation coefficient is 0.001-0.01 / day, and the basic difference threshold is 0.2.

[0178] The method provided by the application introduces a timeliness weight coefficient, calculates the attenuation degree according to the data storage time, and adjusts the second difference index in combination with the basic difference threshold. The old data weight is reduced by the exponential attenuation formula, the algorithm preferentially selects recent matching cases, and the timeliness of the result is ensured. At the same time, the constraint of the basic threshold avoids ignoring the essential difference of the physiological parameters due to excessive dependence on new data, for example, preventing new data from being misjudged as the best match due to data acquisition errors. The combination of the two realizes the dynamic balance of data new and old and matching quality, and improves the timeliness and reliability of the matching result.

[0179] From the above S1-S4, the method provided by the embodiment of the application constructs a multi-modal feature database according to the physiological parameter information, skull feature information and DTI image of a plurality of stroke individuals, and quickly and accurately matches candidate individuals from the multi-modal feature database according to the physiological parameter information, CT image of the target individual through the first difference index and the second difference index, wherein the first difference index is used to represent the difference degree between the skull feature information, and the second difference index is used to represent the difference degree between the physiological parameter information. The DTI image of the candidate individual is determined as the DTI image of the target individual, which can effectively eliminate the self-condition limitation and device limitation of the target individual, reduce the time of obtaining the DTI image, and improve the acquisition efficiency. Avoiding the influence of the delay of obtaining the DTI image on the diagnosis and treatment time, and improving the diagnosis and treatment efficiency.

[0180] In one example, referring to Figure 4 , Figure 4 is an application scenario diagram of a multi-modal fusion enabled CT-based human brain hemorrhage DTI image matching method provided by the embodiment of the application. The method provided by the embodiment of the application is applied to an image matching system, and the above-mentioned image matching system includes a server 400 and a plurality of terminal devices 410. The server 400 communicates with the plurality of terminal devices 410 respectively.

[0181] The server 400 acquires physiological parameter information of a plurality of stroke individuals, CT images and DTI images of head regions, the physiological parameter information including age, gender, onset duration, hematoma position, hematoma shape and hematoma volume; the server 400 determines skull feature information of each of the plurality of stroke individuals according to the CT images of the head regions of the plurality of stroke individuals, the skull feature information including thicknesses of different parts of the skull and three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system; and the server 400 constructs a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals.

[0182] For any terminal device 410 sending an image matching request to the server 400, the image matching request carrying physiological parameter information and a CT image of a head region of a target individual, the server 400 responds to the image matching request and determines skull feature information of the target individual according to the CT image of the head region of the target individual. Then the server 400 determines a candidate individual from the multi-modal feature database based on the physiological parameter information and the skull feature information of the target individual, determines a DTI image of the candidate individual as the DTI image of the target individual, and finally sends the DTI image of the target individual to the terminal device 410.

[0183] The above mainly describes the scheme of the embodiments of the present application from the perspective of the method. It can be understood that the image matching system 100 comprises at least one of the corresponding hardware structure and software module for implementing each function. Those skilled in the art should easily realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is implemented in hardware or computer software driven hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of the present application.

[0184] The embodiments of the present application can divide the image matching system 100 into functional units according to the above method examples. For example, the image matching system 100 can be divided into functional units corresponding to each function, or two or more functions can be integrated into one processing unit. The integrated units can be implemented in the form of hardware or software functional units. It should be noted that the division of units in the embodiments of the present application is illustrative, and is only a logical functional division. Actual implementation can have another division manner.

[0185] Exemplarily, Figure 5A hardware structure schematic diagram of an image matching system provided by an embodiment of the present application is shown. The image matching system 100 comprises: a data acquisition module 110, configured to acquire physiological parameter information of a plurality of stroke individuals, CT images and DTI images of head regions of the plurality of stroke individuals, the physiological parameter information comprising age, gender, onset duration, hematoma position, hematoma shape and hematoma volume; an information determination module 120, configured to determine skull feature information of each of the plurality of stroke individuals according to the CT images of the head regions of the plurality of stroke individuals, the skull feature information comprising thicknesses of different parts of the skull and three-dimensional coordinates of each landmark point of the skull in a preset spatial orthogonal coordinate system; a database construction module 130, configured to construct a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals; and an image matching module 140, configured to determine, based on physiological parameter information and skull feature information of a target individual, a DTI image of a candidate individual as the DTI image of the target individual from the multi-modal feature database, the candidate individual being one of the plurality of stroke individuals, a first difference index between the target individual and the candidate individual being less than a preset threshold, and a second difference index between the target individual and the candidate individual being a minimum value of a plurality of second difference indexes between the target individual and the plurality of stroke individuals, the first difference index being positively correlated with a difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index being positively correlated with a difference degree of the physiological parameter information between the target individual and the candidate individual.

[0186] It should be understood that the specific description of the optional manners above can refer to the method embodiments described above, which will not be described herein again. In addition, the explanation and beneficial effect of any of the image matching systems 100 provided above can refer to the corresponding method embodiments described above, which will not be described herein again.

[0187] The embodiment of the present application further provides a computer readable storage medium, and at least one computer instruction is stored in the computer readable storage medium, and the at least one computer instruction is loaded and executed by a processor to realize the method of each of the above embodiments. The explanation and beneficial effect of the related content in any of the computer readable storage media provided above can refer to the corresponding embodiments described above, which will not be described herein again.

[0188] The embodiment of the present application further provides a chip. The chip integrates a control circuit and one or more ports for realizing the functions of the image matching system 100 described above. Optionally, the functions supported by the chip can refer to the above, which will not be described herein again.

[0189] Those skilled in the art can understand that all or part of the steps of the above-mentioned embodiments can be completed by programs instructing relevant hardware. The programs can be stored in a computer readable storage medium. The storage medium mentioned above can be a read-only memory, a random access memory, etc. The processing unit or processor mentioned above can be a central processing unit, a general-purpose processor, an application specific integrated circuit (ASIC), a digital signal processor (DSP), a field programmable gate array (FPGA) or other programmable logic device, a transistor logic device, a hardware component or any combination thereof.

[0190] The embodiments of the present application also provide a computer program product containing instructions, which, when executed on a computer, cause the computer to perform any of the methods described above. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, all or part of the processes or functions according to the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable apparatus. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (such as infrared, wireless, microwave, etc. ) way. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media sets. The available media can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as an SSD), etc.

[0191] It should be noted that the devices for storing computer instructions or computer programs provided by the embodiments of the present application, such as but not limited to the above-mentioned memories, computer readable storage media, communication chips and the like, are all non-transitory. Those skilled in the art should be aware that in one or more examples described above, the functions described by the embodiments of the present application can be implemented by hardware, software, firmware or any combination thereof. When implemented by software, these functions can be stored in a computer readable storage medium or transmitted as one or more instructions or codes on a computer readable storage medium. The computer readable storage medium includes computer storage medium and communication medium, wherein the communication medium includes any medium that facilitates the transfer of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0192] Although the embodiments of the present application have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-mentioned embodiments within the scope of the present application.

Claims

1. A multi-modal fusion-based CT-enabled human brain hemorrhage DTI image matching method, characterized in that, The method comprises: acquiring physiological parameter information, CT images and DTI images of head regions of a plurality of stroke individuals, the physiological parameter information including age, gender, onset duration, hematoma position, hematoma shape and hematoma volume; determining skull feature information of each of the plurality of stroke individuals according to the CT images of the head regions of the plurality of stroke individuals, the skull feature information including thicknesses of different parts of the skull and three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system; constructing a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals; determining, from the multi-modal feature database, a DTI image of a candidate individual as a DTI image of a target individual based on the physiological parameter information and the skull feature information of the target individual, the candidate individual being one of the plurality of stroke individuals, a first difference index between the target individual and the candidate individual being less than a preset threshold, and a second difference index between the target individual and the candidate individual being the minimum value of a plurality of second difference indexes between the target individual and the plurality of stroke individuals, the first difference index being positively correlated with a difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index being positively correlated with a difference degree of the physiological parameter information between the target individual and the candidate individual.

2. The method of claim 1, wherein, The different parts of the skull include frontal bone, parietal bone, temporal bone, occipital bone and sphenoid bone, and the landmark points include anterior point, posterior point, left temporal point, right temporal point, vertex point and basilar point of the skull cavity, and hemorrhage point, glabella point, fontanelle point, person point, center point of external auditory meatus and external occipital protuberance point.

3. The method of claim 2, wherein, The method comprises: determining a skull binary image of each of the plurality of stroke individuals according to the CT images of the head regions of the plurality of stroke individuals; determining a skull three-dimensional model corresponding to the skull binary image of each of the plurality of stroke individuals based on a three-dimensional reconstruction algorithm; determining the thicknesses of the different parts of the skull according to distances between each point corresponding to the three-dimensional profile of each of the different parts of the skull on the skull three-dimensional model and a center point of the skull three-dimensional model; determining the three-dimensional coordinates of each landmark point of the skull in the preset spatial rectangular coordinate system, wherein the preset spatial rectangular coordinate system takes the center point of the skull three-dimensional model as an origin.

4. The method of claim 3, wherein, The determination formula of the first difference index D1 is: D1 = a - D thickness + (1 - a) - D landmark ; wherein a is an overall weight coefficient, D thickness is a skull thickness difference score, D landmark is a landmark position difference score; the skull thickness difference score D thickness is determined by the formula: N is the number of parts that make up the skull, T target,i is the thickness of the i-th part of the skull of the target individual; T candidate,i is the thickness of the i-th part of the skull of the stroke individual; Landmark position difference score D landmark The determination formula is: M is the number of landmarks, P target,j is the three-dimensional coordinate of the jth landmark of the target individual; P candidate,j is the three-dimensional coordinate of the jth landmark of the stroke individual, normalize_factor j is the normalization factor.

5. The method of claim 4, wherein, The different parts of the skull are respectively configured with first weight coefficients, and each of the landmark points is respectively configured with a second weight coefficient; Skull thickness difference score D thickness The determination formula is: Landmark position difference score D landmark The determination formula is: wherein w i is a first weight coefficient for the i-th portion of the skull; v i is a second weight coefficient for the i-th landmark.

6. The method of claim 5, wherein, The determination formula of the preset threshold μ is: μ = μ0·ε; wherein μ0 is a skull difference degree reference threshold, and ε is a skull complexity weight. In a case where the comprehensive complexity index of the target user is less than 0.7, the skull complexity weight ε is 0.8, in a case where the comprehensive complexity index of the target user is greater than or equal to 0.7 and less than 1.3, the skull complexity weight ε is 1, and in a case where the comprehensive complexity index of the target user is greater than 1.3, the skull complexity weight ε is 1.2; A determination formula of the comprehensive complexity index τ is: where A is a balance coefficient, Dref is a reference value of dispersion; CV thickness is a coefficient of variation of skull thickness, η thickness is a standard deviation of different parts of the skull, σ thickness is a mean value of different parts of the skull, DV landmark is a dispersion of skull landmarks; P j is a three-dimensional coordinate of the jth landmark of the target user, and P is a three-dimensional coordinate of the geometric center of all landmarks of the target user.

7. The method of claim 6, wherein, A determination formula of the second difference index D2 is: K is the number of parameters included in the physiological parameter information; β k is the weight coefficient of the kth parameter included in the physiological parameter information, Score k is the difference score of the kth parameter included in the physiological parameter information; physiological parameter information includes a difference score for age age is: A target is an age of the target individual, A candidate is an age of the stroke individual, A max and A min is a maximum and a minimum of the ages of the plurality of stroke individuals; The physiological parameter information includes a gender difference score Score gender is: Score = 0 in the case where the target individual is of the same gender as the stroke individual gender = 0; In a case where the gender of the target individual is different from that of the stroke individual, Scoregender is 1. physiological parameter information includes a difference score Score of the length of the episode time is: T target is the length of the onset of the target individual; T candidate is the length of the onset of the stroke individual; T max and T min are the maximum and minimum values of the length of the stroke for the plurality of stroke individuals; physiological parameter information includes a difference score Score of the hematoma location location is: In the case where the target individual and the stroke individual have the same region as the hematoma region, Score location = 0; Score = 0.5 in case the blood clot region of the target individual and the blood clot region of the stroke individual are located in adjacent regions; Score = 1 in case the blood clot region of the target individual and the blood clot region of the stroke individual are located in different and not adjacent regions. location location Score = 0.5 in case the blood clot region of the target individual and the blood clot region of the stroke individual are located in adjacent regions; Score = 1 in case the blood clot region of the target individual and the blood clot region of the stroke individual are located in different and not adjacent regions.​ The physiological parameter information includes a difference score Score of the hematoma morphology shape is: L target L is the length of the long axis of the hematoma for the target individual candidate L is the length of the long axis of the hematoma for the stroke individual max L is the length of the long axis of the hematoma for the target individual min L is the maximum and minimum of the length of the long axis of the hematoma for the plurality of stroke individuals W target W is a hematoma short axis length of the target individual candidate W is a hematoma short axis length of the stroke individual; W max and W min W and W are a maximum and a minimum of the hematoma short axis lengths of the plurality of stroke individuals; H target H is a hematoma height of the target individual candidate H is a hematoma height of the stroke individual; H max and H min H and H are a maximum and a minimum of the hematoma heights of the plurality of stroke individuals; Physiological parameter information includes a difference score Score of hematoma volume volume Is: V target is the volume of a hematoma for a target individual; V candidate is the volume of a hematoma for a stroke individual.

8. The method of claim 7, wherein, The method further comprises: Determining a timeliness weight coefficient corresponding to each stroke individual according to a storage time of the skull feature information, the physiological parameter information and the DTI image of each stroke individual in the multi-modal feature database; The time-effectiveness weight coefficient γ of the i-th stroke individual norm，i The determination formula is: Wherein, t is the current time, t0 is the storage time, and λ is the decay coefficient; A determination formula of the second difference index D2 is: Wherein, θ is a basic difference threshold.

9. A multi-modal fusion enabled CT-based human brain hemorrhage DTI image matching system, characterized in that, The system comprises: A data acquisition module configured to acquire physiological parameter information, CT images and DTI images of a head region of a plurality of stroke individuals, the physiological parameter information including age, gender, onset duration, hematoma location, hematoma shape and hematoma volume; An information determination module configured to determine skull feature information of each stroke individual in the plurality of stroke individuals according to the CT images of the head region of the plurality of stroke individuals, the skull feature information including the thickness of different parts of the skull and the three-dimensional coordinates of each landmark point of the skull in a preset spatial rectangular coordinate system; A database construction module configured to construct a multi-modal feature database according to the skull feature information, the physiological parameter information and the DTI images of the plurality of stroke individuals; An image matching module configured to determine, based on the physiological parameter information and the skull feature information of a target individual, a DTI image of a candidate individual as the DTI image of the target individual from the multi-modal feature database, the candidate individual being one of the plurality of stroke individuals, a first difference index between the target individual and the candidate individual being less than a preset threshold, and a second difference index between the target individual and the candidate individual being the smallest of a plurality of second difference indexes between the target individual and the plurality of stroke individuals, the first difference index being positively correlated with the difference degree of the skull feature information between the target individual and the candidate individual, and the second difference index being positively correlated with the difference degree of the physiological parameter information between the target individual and the candidate individual.

10. An electronic device, comprising: Comprise: A processor; A memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1-8.

Images