A method, device, equipment and medium for monitoring a stroke lesion volume

Abstract

The application discloses a kind of monitoring method, device and equipment and storage medium of cerebral apoplexy focus volume.The method comprises: according to the first brain image to be detected corresponding with baseline acquisition time, determine baseline focus volume, according to at least one vessel image sequence of interest corresponding with the cerebral apoplexy type to be detected, determine at least one vessel space-time characteristic corresponding to each vessel space-time characteristic, baseline focus volume and volume monitoring length are input into the focus volume prediction model obtained by pre-training, obtain the output corresponding to the monitoring focus volume of target monitoring time, wherein, volume monitoring length represents the time length between baseline acquisition time and target monitoring time, the present application embodiment solves the problem that the focus volume of cerebral apoplexy needs to be monitored by medical imaging equipment, reduces the medical cost and radiation dose of cerebral apoplexy condition monitoring.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, and in particular to a method, device, equipment and storage medium for monitoring the volume of stroke lesions. Background Technology

[0002] Cerebral stroke, also known as apoplexy, is an acute cerebrovascular disease caused by the sudden rupture of a blood vessel in the brain or by a blockage of a blood vessel in the brain, which prevents blood from flowing into the brain and causes damage to brain tissue. It includes ischemic stroke and hemorrhagic stroke.

[0003] Stroke has a high mortality and recurrence rate, and is associated with the risk of other neurological complications. Clinical studies have shown that the progression of stroke lesion volume is positively correlated with neurological dysfunction.

[0004] In current clinical practice, medical imaging equipment is needed to monitor changes in the volume of stroke lesions. However, multiple medical imaging examinations can significantly increase medical costs, and some medical imaging examinations can expose patients to high radiation doses. Summary of the Invention

[0005] This invention provides a method, device, equipment, and storage medium for monitoring the volume of stroke lesions, thereby addressing the issue that the volume of stroke lesions needs to be monitored using medical imaging equipment, and reducing the medical costs and radiation dose associated with stroke monitoring.

[0006] According to one embodiment of the present invention, a method for monitoring the volume of a stroke lesion is provided, the method comprising:

[0007] Acquire a dataset of images to be detected corresponding to the type of stroke to be detected; wherein, the dataset of images to be detected contains a first brain image to be detected corresponding to the baseline acquisition time and a sequence of images of at least one vessel of interest;

[0008] Based on the first brain image to be detected, determine the baseline lesion volume corresponding to the baseline acquisition time;

[0009] For each vessel of interest, the spatiotemporal features of the vessel of interest are determined based on the image sequence of the vessel of interest.

[0010] The spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration are input into a pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0011] The volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time.

[0012] According to another embodiment of the present invention, a device for monitoring the volume of a stroke lesion is provided, the device comprising:

[0013] The image dataset acquisition module is used to acquire an image dataset corresponding to the type of stroke to be detected; wherein, the image dataset to be detected includes a first brain image to be detected corresponding to the baseline acquisition time and an image sequence of at least one vessel of interest;

[0014] The baseline lesion volume determination module is used to determine the baseline lesion volume corresponding to the baseline acquisition time based on the first brain image to be detected.

[0015] The vascular spatiotemporal feature determination module is used to determine the vascular spatiotemporal features of each vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest.

[0016] The lesion volume monitoring output module is used to input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0017] The volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time.

[0018] According to another embodiment of the present invention, an electronic device is provided, the electronic device comprising:

[0019] At least one processor; and

[0020] A memory that is communicatively connected to the at least one processor;

[0021] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the method for monitoring the volume of stroke lesions according to any embodiment of the present invention.

[0022] According to another embodiment of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the method for monitoring the volume of stroke lesions according to any embodiment of the present invention.

[0023] The technical solution of this invention determines the baseline lesion volume based on a first brain image to be detected corresponding to the baseline acquisition time, and determines the spatiotemporal features of at least one vessel of interest based on an image sequence of at least one vessel of interest corresponding to the type of stroke to be detected. The spatiotemporal features of each vessel, the baseline lesion volume, and the volume monitoring duration are input into a pre-trained lesion volume prediction model to obtain the output monitored lesion volume corresponding to the target monitoring time. The volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time. This embodiment utilizes the correlation between the spatiotemporal features of vessels and the changes in the lesion volume of stroke, solving the problem that the lesion volume of stroke needs to be monitored using medical imaging equipment, and reducing the medical cost and radiation dose of stroke monitoring.

[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating a method for monitoring the volume of a stroke lesion according to an embodiment of the present invention;

[0027] Figure 2 A model architecture diagram of a self-encoder provided in one embodiment of the present invention;

[0028] Figure 3 This is a network architecture diagram of a fully connected network provided in one embodiment of the present invention;

[0029] Figure 4 A flowchart illustrating another method for monitoring the volume of a stroke lesion, provided in one embodiment of the present invention;

[0030] Figure 5 This is a schematic diagram illustrating the process of transforming a blood vessel image of interest into a straightened blood vessel image, provided as an embodiment of the present invention.

[0031] Figure 6 A flowchart illustrating another method for monitoring the volume of a stroke lesion, provided in one embodiment of the present invention;

[0032] Figure 7A flowchart illustrating a specific example of a method for monitoring the volume of a stroke lesion provided in an embodiment of the present invention;

[0033] Figure 8 This is a schematic diagram of the structure of a device for monitoring the volume of a stroke lesion according to an embodiment of the present invention;

[0034] Figure 9 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present invention. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0036] It should be noted that the terms "first," "second," "original," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0037] Figure 1 This is a flowchart illustrating a method for monitoring the volume of a stroke lesion according to an embodiment of the present invention. This embodiment is applicable to monitoring the progression and changes in the volume of a stroke lesion. The method can be executed by a stroke lesion volume monitoring device, which can be implemented in hardware and / or software and can be configured in a terminal device. Figure 1 As shown, the method includes:

[0038] S110. Obtain the image dataset corresponding to the type of stroke to be detected.

[0039] In this embodiment, the type of stroke to be detected is ischemic stroke or hemorrhagic stroke. Ischemic stroke refers to brain tissue necrosis caused by insufficient blood supply to the brain due to narrowing or occlusion of the blood supply arteries to the brain (such as the carotid artery and / or vertebral artery). Hemorrhagic stroke, also known as cerebral hemorrhage, intracranial hemorrhage, or cerebral hemorrhage, refers to brain tissue damage caused by intracranial vascular rupture and bleeding due to non-traumatic causes.

[0040] In this embodiment, the image dataset to be detected includes a first brain image to be detected corresponding to the baseline acquisition time and a sequence of images of at least one vessel of interest.

[0041] Specifically, the first brain image to be detected is used to characterize the brain image containing the stroke lesion acquired at the baseline acquisition time, and the baseline lesion volume is used to characterize the volume of the stroke lesion in the first brain image to be detected.

[0042] Specifically, the vessel of interest (ROI) image sequence includes at least two time-series-based ROI images. These ROI images characterize vascular images containing vessels of interest corresponding to the type of stroke to be detected. For example, at least one ROI corresponding to ischemic stroke includes, but is not limited to, the carotid artery and vertebral artery; at least one ROI corresponding to hemorrhagic stroke includes, but is not limited to, the inferior anastomotic vein (Labbe), the superficial middle cerebral vein (SMCV), and the basal vein of Rosenthal (BVR). The correspondence between the stroke type to be detected and each ROI is not limited here; the ROI can be customized according to actual needs.

[0043] The image type of the first brain image to be detected can be the same as or different from the image of the vessel of interest. For example, the image type of the first brain image to be detected or the image of the vessel of interest includes, but is not limited to, direct digital radiography (DR), computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), or ultrasound images. The image type of the first brain image to be detected or the image of the vessel of interest is not limited here, and can be customized according to actual needs.

[0044] S120. Based on the first brain image to be detected, determine the baseline lesion volume corresponding to the baseline acquisition time.

[0045] In one optional embodiment, determining the baseline lesion volume corresponding to the baseline acquisition time based on the first brain image to be detected includes: inputting the first brain image to be detected into a pre-trained volume extraction model to obtain the output baseline lesion volume corresponding to the baseline acquisition time.

[0046] Specifically, the method further includes: inputting the trained brain image into the untrained volume extraction model to obtain the output predicted volume features; determining the first loss function based on the predicted volume features and the standard volume features; adjusting the model parameters of the volume extraction model based on the first loss function; and taking the volume extraction model in the current iteration as the trained volume extraction model when the first loss function converges.

[0047] For example, the function type of the first loss function includes, but is not limited to, squared loss function, logarithmic loss function, exponential loss function, mean squared error loss function, logistic regression loss function, Huber loss function, cross-entropy loss function, or Kullback-Leibler divergence loss function, etc. The function type of the first loss function is not limited here, and can be customized according to actual needs.

[0048] In another optional embodiment, determining the baseline lesion volume corresponding to the baseline acquisition time based on the first brain image to be detected includes: performing a segmentation operation on the first brain image to be detected to obtain a stroke lesion image, and determining the baseline lesion volume corresponding to the baseline acquisition time based on the lesion voxel size corresponding to the stroke lesion image.

[0049] For example, the segmentation tool used in the segmentation operation can be The Medical Imaging Interaction Toolkit (MITK). There is no limitation on the segmentation tool used in the segmentation operation here, and the specific settings can be customized according to actual needs.

[0050] S130. For each vessel of interest, determine the spatiotemporal characteristics of the vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest.

[0051] Specifically, vascular spatiotemporal features are used to characterize the temporal and spatial features of blood vessels of interest. For example, vascular temporal features include, but are not limited to, hemodynamic features, and vascular spatial features include, but are not limited to, vascular morphological features, vascular structural features, and vascular elasticity features.

[0052] In one optional embodiment, determining the spatiotemporal features of the blood vessel of interest based on the image sequence of the blood vessel of interest includes: inputting the image sequence of the blood vessel of interest into a pre-trained feature extraction model to obtain the output spatiotemporal features of the blood vessel of interest.

[0053] For example, the model type of the feature extraction model can be the encoder in an autoencoder. Specifically, the method further includes: inputting a training vascular image sequence into an untrained autoencoder; extracting vascular spatiotemporal features from the input training vascular image sequence using the encoder in the autoencoder; performing a decoding operation on the input vascular spatiotemporal features using the decoder in the autoencoder to obtain a predicted vascular image sequence; determining a second loss function based on the predicted vascular image sequence and the training vascular image sequence; and adjusting the model parameters of the autoencoder based on the second loss function; until the second loss function converges, using the encoder in the autoencoder during the current iteration as the feature extraction model.

[0054] For example, the function type of the second loss function includes, but is not limited to, squared loss function, logarithmic loss function, exponential loss function, mean squared error loss function, logistic regression loss function, Huber loss function, cross-entropy loss function, or Kullback-Leibler divergence loss function, etc. The function type of the second loss function is not limited here, and can be customized according to actual needs.

[0055] Figure 2 This is a model architecture diagram of an autoencoder provided in one embodiment of the present invention. Specifically, Figure 2 The left-hand model structure is the encoder in the autoencoder, and the right-hand model structure is the decoder in the autoencoder. The model structures of the encoder and decoder are basically symmetrical, consisting of a stack-like structure composed of multiple convolutional components. The decoder has one more convolutional output layer than the encoder.

[0056] S140. Input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0057] In this embodiment, the volume monitoring duration represents the length of time between the baseline acquisition time and the target monitoring time. For example, the volume monitoring duration can be 1 day, 7 days, or 30 days. There is no limitation on the volume monitoring duration here, and it can be customized according to actual needs.

[0058] For example, the model types for lesion volume prediction models include, but are not limited to, CNN (Convolutional Neural Networks), FCN (Fully Convolutional Networks), ResNet, DNN (Deep Neural Networks), RNN (Recurrent Neural Networks), or Transformer networks, etc. The model type for lesion volume prediction is not limited here; it can be customized according to actual needs.

[0059] In one alternative embodiment, the lesion volume prediction model is a fully connected network.

[0060] Figure 3 This is a network architecture diagram of a fully connected network provided in one embodiment of the present invention. Specifically, the fully connected network consists of 5 fully connected layers.

[0061] Specifically, the method further includes: inputting the training volume features, volume training time, and spatiotemporal features of blood vessels corresponding to at least one training blood vessel image sequence into the untrained lesion volume prediction model to obtain the output predicted lesion volume corresponding to the training monitoring time; determining a third loss function based on the predicted lesion volume and the standard lesion volume; adjusting the model parameters of the lesion volume prediction model based on the third loss function; and taking the lesion volume prediction model in the current iteration process as the trained lesion volume prediction model when the third loss function converges.

[0062] For example, the function type of the third loss function includes, but is not limited to, squared loss function, logarithmic loss function, exponential loss function, mean squared error loss function, logistic regression loss function, Huber loss function, cross-entropy loss function, or Kullback-Leibler divergence loss function, etc. The function type of the third loss function is not limited here, and can be customized according to actual needs.

[0063] The technical solution of this embodiment determines the baseline lesion volume based on the first brain image to be detected corresponding to the baseline acquisition time, and determines the spatiotemporal features of at least one vessel of interest based on the image sequence of at least one vessel of interest corresponding to the type of stroke to be detected. The spatiotemporal features of each vessel, the baseline lesion volume, and the volume monitoring duration are input into a pre-trained lesion volume prediction model to obtain the output monitored lesion volume corresponding to the target monitoring time. The volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time. This embodiment utilizes the correlation between the spatiotemporal features of vessels and the changes in the lesion volume of stroke to solve the problem that the lesion volume of stroke needs to be monitored using medical imaging equipment, thereby reducing the medical cost and radiation dose of stroke monitoring.

[0064] Figure 4 This is a flowchart of another method for monitoring the volume of a stroke lesion according to an embodiment of the present invention. This embodiment further refines the step of "determining the spatiotemporal characteristics of the vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest" in the above embodiment. Figure 4 As shown, the method includes:

[0065] S210. Obtain the image dataset corresponding to the type of stroke to be detected.

[0066] S220. Based on the first brain image to be detected, determine the baseline lesion volume corresponding to the baseline acquisition time.

[0067] S210-S220 in this embodiment are the same as those in the above embodiment. Figure 1 The S110-S120 shown are the same or similar, and will not be described again in this embodiment.

[0068] S230. For each vessel of interest, perform a straightening operation on at least two vessels of interest images in the corresponding vessel of interest image sequence to obtain at least two straightened vessel images.

[0069] In one optional embodiment, a straightening operation is performed on at least two blood vessel images in the blood vessel image sequence corresponding to the blood vessel of interest to obtain at least two straightened blood vessel images. This includes: for each blood vessel image of interest, obtaining the coordinates of the center points of at least two blood vessel center points on the blood vessel centerline corresponding to the blood vessel image of interest; determining the tangent vectors corresponding to each blood vessel center point based on the coordinates of each center point; determining the normal plane coordinates corresponding to at least two blood vessel center points based on the coordinates of each center point and the tangent vectors; and determining the straightened blood vessel image corresponding to the blood vessel image of interest based on the normal plane coordinates.

[0070] For example, the tool for obtaining the center point coordinates can be The Vascular Modeling Toolkit (VMTK). There is no limitation on the tool used here, and it can be selected according to the actual needs.

[0071] Figure 5 This is a schematic diagram illustrating a process from an image of a blood vessel of interest to a straightened blood vessel image, provided as an embodiment of the present invention. Specifically, Figure 5 Figure 'a' in the diagram represents the image of the vessel of interest, with each normal plane represented by {P1, P2, ..., P...}. n} represents the center point of each blood vessel, using {a1, a2, ..., a...} L}express, This represents the tangent vector corresponding to the center point of the i-th blood vessel. Specifically, L represents the number of vessel center points on the vessel centerline.

[0072] like Figure 5 As shown in Figure b, for two adjacent normal planes on the center line of the blood vessel, such as normal plane P... i-1 and normal plane P i Normal plane P i It can be viewed as being formed by the normal plane P i-1 After rotation, we obtain plane P′ i Then from plane P′ i The result is obtained after translation.

[0073] Specifically, the normal plane P i ={v ij |1≤j≤m}, where m represents the normal plane P. i The number of points contained within it. And the normal plane P. i-1 Corresponding tangent vector With normal plane P i Corresponding tangent vector The included angle between them satisfies the formula:

[0074]

[0075] Specifically, tangent vector with tangent vector The axis of rotation between them is a unit vector, satisfying the formula:

[0076]

[0077] Specifically, the normal plane P i-1 The plane P′ obtained after rotation i The process can be viewed as the normal plane P i-1 All points v on(i-1)j Along a plane P that passes through the normal i-1 The center point of the blood vessel a i-1 rotation axis u (i-1)i Rotation θ (i-1)i The plane P′ is obtained. i All corresponding points v′ ij .

[0078] Specifically, the rotation of a vector in three-dimensional space is used to represent the rotation of the corresponding point. For example... Figure 5 As shown in Figure c, Along an axis of rotation passing through point a Rotate θ to obtain During this rotation process, the axis of rotation It is a quantity that has both magnitude and direction, but its magnitude (length) is not important here. To eliminate the axis of rotation... The modulus length is an unnecessary degree of freedom, and for ease of calculation, a rotation axis is defined. The module length is: Right now It is a unit vector.

[0079] Specifically, will and Decomposed into parallel axes of rotation and perpendicular (or orthogonal) to the axis of rotation Two component vectors ( and ),Right now in, yes On the axis of rotation Therefore, by orthogonal projection onto the surface, we can obtain:

[0080]

[0081] because and Simultaneously perpendicular (or orthogonal) to the axis of rotation Therefore, it can be about the axis of rotation Rotation to obtain This can be viewed as a rotation within a plane. Because the rotation does not change the modulus, the trajectory of the rotation can be represented by a circle. For example... Figure 5 As shown in diagram d, a structure can be constructed that is simultaneously perpendicular (or orthogonal) to the axis of rotation. and vector Right now

[0082]

[0083]

[0084] in, and The modulus lengths are equal, indicating that Also on the circle. Decompose into perpendicular (orthogonal) to and perpendicular (orthogonal) to Two component vectors Right now

[0085]

[0086]

[0087] Specifically, the normal plane P i-1 The plane P′ obtained after rotation i The process can be viewed as the normal plane P i-1 All vectors on Along a plane P that passes through the normal i-1 The center point of the blood vessel a i-1 Rotation axis (Unit vector) rotated θ (i-1)i The plane P′ is obtained. i All vectors corresponding to the above, i.e.

[0088]

[0089]

[0090]

[0091]

[0092] Specifically, plane P′ i The normal plane P obtained after translation i The process can be viewed as plane P′ i All vectors on The center point a of the blood vessel along the center line of the blood vessel i-1 and the center point of the blood vessel a i Rotate and translate in the direction to obtain the normal plane P. i All vectors corresponding to the above Right now

[0093]

[0094] Through the above derivation process, the normal plane P can be obtained. i-1 and normal plane P iThe rotational and translational relationships between them are known, so we only need to construct any known plane P0 to obtain all the normal planes.

[0095] Specifically, let the known plane P0 be a plane that passes through the origin of the coordinate system and is perpendicular to the z-axis, with its center point a0 = [0, 0, 0] and its normal vector... Each point {v} on it 01 v 02 , ..., v 0m Given the coordinates of the vessel centerline, according to the above formula, we can obtain the coordinates of all normal planes on the vessel centerline, P = {P1, P2, ..., P}. n Using an interpolation algorithm, P is introduced into the voxel values ​​of the four-dimensional blood vessel image of interest to obtain the coordinates of all normal planes in the straightened blood vessel image, V = {V1, V2, ..., V...}. n}

[0096] like Figure 5 As shown in Figure e, V is introduced into a new Cartesian coordinate system and arranged sequentially on the z-axis to obtain the straightened blood vessel image. At this time, the straightened blood vessel image is in a four-dimensional matrix of size [T, L, 16, 16], where T represents the number of images of the blood vessel of interest in the sequence of images of the blood vessel of interest.

[0097] S240. Input each straightened blood vessel image into the pre-trained feature extraction model to obtain the spatiotemporal features of the blood vessel of interest.

[0098] The feature extraction model in this embodiment is the same as or similar to the feature extraction model used in the above embodiment for feature extraction of the blood vessel image sequence of interest, and will not be described again in this embodiment.

[0099] S250. Input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0100] S250 in this embodiment is the same as in the above embodiment. Figure 1 The S140 shown is the same or similar, and will not be described again in this embodiment.

[0101] The technical solution of this embodiment involves performing a straightening operation on at least two images of the vessel of interest in the image sequence corresponding to each vessel of interest, thereby obtaining at least two straightened vessel images. Each straightened vessel image is then input into a pre-trained feature extraction model to obtain the spatiotemporal features of the output vessel of interest. This eliminates the interference of vascular morphology features in traditional vascular images on the prediction results of lesion volume prediction models, and improves the accuracy of monitoring lesion volume in stroke.

[0102] Figure 6 This is a flowchart illustrating another method for monitoring the volume of a stroke lesion according to an embodiment of the present invention. This embodiment further refines the step of "acquiring the image dataset corresponding to the type of stroke to be detected" in the above embodiment. For example... Figure 6 As shown, the method includes:

[0103] S310. Obtain the vessel type of at least one vessel of interest corresponding to the type of stroke to be detected.

[0104] In this embodiment, when the type of stroke to be detected is ischemic stroke, the blood vessel type is arterial; when the type of stroke to be detected is hemorrhagic stroke, the blood vessel type is venous.

[0105] S320. Based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, perform a segmentation operation on the second brain image sequence to be detected to obtain a cerebral vascular image sequence corresponding to the vascular type.

[0106] Specifically, the second brain image sequence to be detected includes at least two second brain images to be detected based on time series, which are used to characterize a brain image containing at least one blood vessel of interest.

[0107] In an optional embodiment, the method further includes: obtaining the skull coordinates in the first frame of the original brain image in the original brain image sequence, and performing a rigid registration operation on the original brain images other than the first frame of the original brain image in the original brain image sequence according to the skull coordinates to obtain a registered brain image sequence; performing a segmentation operation on the first frame of the registered brain image in the registered brain image sequence according to a preset skull threshold to obtain a skull mask image; and performing a deletion operation on each registered brain image in the registered brain image sequence according to the skull mask image to obtain a second brain image sequence to be detected.

[0108] Specifically, rigid registration operation includes registration operations in 6 degrees of freedom: translation along the x-axis, translation along the y-axis, translation along the z-axis, rotation around the x-axis, rotation around the y-axis, and rotation around the z-axis.

[0109] For example, when the original brain image is a CT image, the preset skull threshold can be 155 HU. Specifically, the image composed of image voxels with a voxel value less than 155 HU in the first frame of the registered brain image is used as the skull mask image. The preset skull threshold is not limited here and can be customized according to actual needs.

[0110] The advantage of this setup is that it can reduce the alignment error of the brain image sequence and the impact of the skull on the segmentation effect of subsequent blood vessel images of interest, improve the image quality of the second brain image to be detected, and thus improve the accuracy of monitoring the lesion volume of stroke.

[0111] For example, the time decay curve can be generated using a Gaussian function. In an optional embodiment, based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, a segmentation operation is performed on the second brain image sequence to be detected to obtain a cerebral vascular image sequence corresponding to the blood vessel type. This includes: determining the sum sequence of curves corresponding to each brain voxel based on at least one high-order time decay curve corresponding to each brain voxel in the second brain image sequence to be detected; taking the image sequence composed of at least one brain voxel whose sum sequence of curves in the second brain image sequence to be detected satisfies a preset sequence condition as the cerebral vascular image sequence; determining the peak time corresponding to each cerebral vascular voxel based on the zero-order time decay curve corresponding to each cerebral vascular voxel in the cerebral vascular image sequence; and taking the image sequence composed of at least one cerebral vascular voxel whose peak time satisfies a preset time range corresponding to the blood vessel type as the cerebral vascular image sequence corresponding to the blood vessel type.

[0112] For example, each time decay higher-order curve can be a time decay first-order curve and a time decay second-order curve. Specifically, the curve sum sequence includes the sum of the absolute values ​​of the curves corresponding to at least one time decay higher-order curve.

[0113] For example, the sum of the absolute values ​​FG of the curves corresponding to the first-order time decay curves satisfies the formula: The sum of the absolute values ​​SG of the curves corresponding to the second-order time decay curves satisfies the formula: Where T represents the number of images of the second brain image to be detected in the second brain image sequence, and v′ i v″ represents the first-order curve parameter value of the brain voxel in the second brain image to be detected in the i-th frame. i This represents the second-order curve parameter value of the brain voxel in the second brain image to be detected in the i-th frame.

[0114] Specifically, the preset sequence conditions include curve parameter ranges corresponding to at least one higher-order time decay curve. For example, the curve parameter range corresponding to the first-order time decay curve is [0 21], and the curve parameter range corresponding to the second-order time decay curve is [0 2.15]. The preset sequence conditions are not limited here, and can be customized according to actual needs.

[0115] Specifically, time to peak (TTP) is used to characterize the time it takes for the zero-order curve parameter value of the cerebrovascular voxel in the time decay zero-order curve to reach its peak value.

[0116] In an optional embodiment, the method further includes: generating a cerebral vascular voxel-peak time histogram based on the peak time corresponding to each cerebral vascular voxel; using the minimum peak time between the peak times of two peaks in the cerebral vascular voxel-peak time histogram as a preset time threshold; and constructing a preset time range based on the blood vessel type and the preset time threshold; wherein, when the blood vessel type is a venous blood vessel, the minimum boundary value in the preset time range is the preset time threshold, and when the blood vessel type is an arterial blood vessel, the maximum boundary value in the preset time range is the preset time threshold.

[0117] Specifically, cerebral vascular image sequences are either cerebral vein image sequences or cerebral artery image sequences. Since the contrast agent reaches the arteries earlier than it reaches the veins, a preset time threshold can be used to distinguish between cerebral vein images and cerebral artery images within cerebral vascular images.

[0118] S330. Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, perform a segmentation operation on the cerebral vascular image sequence to obtain the image sequence of each vessel of interest.

[0119] In an optional embodiment, the brain vascular image sequence is segmented according to the zero-order time decay curves corresponding to each brain vascular voxel in the brain vascular image sequence to obtain the image sequence of the vessel of interest corresponding to each vessel of interest. This includes: determining the peak time corresponding to each brain vascular voxel according to the zero-order time decay curves corresponding to each brain vascular voxel in the brain vascular image sequence; obtaining the difference time between each peak time and a preset time threshold; and for each vessel of interest, using the image sequence composed of at least one brain vascular voxel whose difference time satisfies the difference time range corresponding to the vessel of interest as the image sequence of the vessel of interest.

[0120] Specifically, the method for determining the preset time threshold is the same as or similar to the method for determining the preset time threshold in the process of determining the preset time range in the above embodiments, and will not be repeated here.

[0121] Specifically, the time range of the difference corresponding to each blood vessel of interest can be the same or different. For example, the time range of the difference can be [0 5]. This embodiment does not limit this, and the specific setting can be customized according to actual needs.

[0122] S340. Add the first brain image to be detected and the sequence of images of each blood vessel of interest to the image dataset corresponding to the type of stroke to be detected.

[0123] S350. Based on the first brain image to be detected, determine the baseline lesion volume corresponding to the baseline acquisition time.

[0124] Specifically, the first brain image to be detected can be any second brain image in the second brain image sequence, or it can be other brain images. The image type of the first brain image to be detected can be the same as or different from the image type of the second brain image to be detected.

[0125] S360. For each vessel of interest, determine the spatiotemporal characteristics of the vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest.

[0126] S370. Input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0127] S350-S370 in this embodiment are the same as those in the above embodiment. Figure 1 The S120-S140 shown are the same as or similar to those in the above embodiments. Figure 4 The S220-S250 shown are the same or similar, and will not be described again in this embodiment.

[0128] Figure 7 This is a flowchart illustrating a specific example of a method for monitoring the volume of a stroke lesion provided in an embodiment of the present invention. Specifically, taking the first brain image to be detected as an NCCT image (Non-comparative Computed Tomography), the second brain image sequence to be detected as a TRCT image sequence (Temporal resolution Computed Tomography), and the stroke type to be detected as hemorrhagic stroke as an example, the baseline NCCT image represents the NCCT image acquired at the baseline acquisition time, and the baseline TRCT image sequence represents the TRCT image sequence acquired at the baseline acquisition time.

[0129] Specifically, based on the first brain image in the baseline TRCT image sequence, a rigid registration operation is performed on the baseline TRCT image sequence to obtain a registered TRCT image sequence. The image composed of image voxels with a voxel value less than 155 HU in the first brain image is used as a skull mask image. Based on the skull mask image, a deletion operation is performed on the registered TRCT image sequence to obtain the TRCT image sequence to be detected.

[0130] Specifically, based on the TRCT image sequence to be detected, the first-order time decay curve and the second-order time decay curve are determined, and the sum of the absolute values ​​of the curves corresponding to the first-order time decay curve (FG) and the sum of the absolute values ​​of the curves corresponding to the second-order time decay curve (SG) are obtained. Based on the preset sequence conditions constructed by the sum of the absolute values ​​of curves (FG) being less than 21 and the sum of the absolute values ​​of curves (SG), the TRCT image sequence to be detected is segmented to obtain the cerebral vascular image sequence.

[0131] Specifically, based on the zero-order time decay curves corresponding to the cerebral vascular image sequences, a cerebral vascular voxel-peak time histogram is determined. The minimum peak time between the peak times of the two peaks in the cerebral vascular voxel-peak time histogram is taken as the preset time threshold Vttp. The image sequence composed of cerebral vascular voxels whose peak time TTP is greater than the preset time threshold Vttp is taken as the cerebral vein image sequence. The image sequence composed of cerebral vein voxels whose difference between peak time TTP and preset time threshold Vttp is less than 5 is taken as the vessel of interest image sequence.

[0132] Specifically, the VMTK modeling tool is used to obtain the vessel centerlines from each image of the vessel of interest. Based on the vessel centerlines, the straightened vessel images corresponding to each image of the vessel of interest are determined. The straightened vessel images are then input into a pre-trained encoder to obtain the spatiotemporal features of the vessels. The MITK segmentation tool is used to segment the baseline NCCT images to obtain stroke lesion images, and the baseline lesion volume corresponding to the stroke lesion images is obtained. The fused features obtained based on the spatiotemporal features of the vessels, the baseline lesion volume, and the volume monitoring time are input into a pre-trained lesion volume prediction model to obtain the monitored lesion volume corresponding to the target monitoring time.

[0133] The technical solution of this embodiment obtains the blood vessel type of at least one blood vessel of interest corresponding to the type of stroke to be detected. Based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, a segmentation operation is performed on each second brain image to be detected to obtain a cerebral vascular image sequence corresponding to the blood vessel type. Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, a segmentation operation is performed on at least two cerebral vascular images to obtain a blood vessel of interest image sequence corresponding to each blood vessel of interest. The first brain image to be detected and the blood vessel of interest image sequences are added to the image dataset to be detected corresponding to the type of stroke to be detected. This solves the problem of high acquisition difficulty caused by directly acquiring blood vessel of interest image sequences through image imaging equipment, improves the image quality of blood vessel of interest image sequences, and thus improves the accuracy of monitoring the lesion volume of stroke.

[0134] The following are embodiments of the stroke lesion volume monitoring device provided in this invention. This device and the stroke lesion volume monitoring method described in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the stroke lesion volume monitoring device, please refer to the content of the stroke lesion volume monitoring method described in the above embodiments.

[0135] Figure 8 This is a schematic diagram of a device for monitoring the volume of a stroke lesion, provided in one embodiment of the present invention. Figure 8 As shown, the device includes: a target image dataset acquisition module 410, a baseline lesion volume determination module 420, a vascular spatiotemporal feature determination module 430, and a lesion volume monitoring output module 440.

[0136] The image dataset acquisition module 410 is used to acquire an image dataset corresponding to the type of stroke to be detected; wherein the image dataset to be detected includes a first brain image to be detected corresponding to the baseline acquisition time and an image sequence of at least one vessel of interest.

[0137] The baseline lesion volume determination module 420 is used to determine the baseline lesion volume corresponding to the baseline acquisition time based on the first brain image to be detected.

[0138] The vascular spatiotemporal feature determination module 430 is used to determine the vascular spatiotemporal features of each vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest.

[0139] The lesion volume monitoring output module 440 is used to input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time.

[0140] Among them, the volume monitoring duration represents the length of time between the baseline acquisition time and the target monitoring time.

[0141] The technical solution of this embodiment utilizes the correlation between the spatiotemporal characteristics of blood vessels and the changes in the volume of stroke lesions, which solves the problem that the volume of stroke lesions needs to be monitored using medical imaging equipment, thereby reducing the medical cost and radiation dose of stroke monitoring.

[0142] In an optional embodiment, the vascular spatiotemporal feature determination module 430 includes:

[0143] The straightened vessel image determination unit is used to perform a straightening operation on at least two vessels of interest in the image sequence corresponding to the vessel of interest, so as to obtain at least two straightened vessel images.

[0144] The vascular spatiotemporal feature output unit is used to input each straightened vascular image into a pre-trained feature extraction model to obtain the vascular spatiotemporal features of the vascular of interest.

[0145] In one optional embodiment, the straightened blood vessel image determination unit is specifically used for:

[0146] For each image of a vessel of interest, obtain the coordinates of the center points of at least two vessels on the vessel centerline corresponding to the image of interest.

[0147] Based on the coordinates of each center point, determine the tangent vector corresponding to each blood vessel center point;

[0148] Based on the coordinates of each center point and each tangent vector, determine the normal plane coordinates corresponding to at least two vessel center points respectively;

[0149] Based on the coordinates of each normal plane, determine the straightened blood vessel image corresponding to the image of the blood vessel of interest.

[0150] In an optional embodiment, the image dataset acquisition module 410 includes:

[0151] The vessel type acquisition unit is used to acquire the vessel type of at least one vessel of interest corresponding to the type of stroke to be detected.

[0152] The cerebral vascular image sequence determination unit is used to perform a segmentation operation on the second brain image sequence to be detected based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, to obtain a cerebral vascular image sequence corresponding to the blood vessel type.

[0153] The unit for determining the image sequence of vessels of interest is used to perform a segmentation operation on the brain blood vessel image sequence based on the time decay zero-order curves corresponding to each brain blood vessel voxel in the brain blood vessel image sequence to obtain the image sequence of the vessel of interest corresponding to each vessel of interest.

[0154] In one optional embodiment, the cerebrovascular image sequence determination unit is specifically used for:

[0155] Based on at least one time decay higher-order curve corresponding to each brain voxel in the second brain image sequence to be detected, determine the curve sum sequence corresponding to each brain voxel; and take the image sequence composed of at least one brain voxel in the second brain image sequence to be detected that satisfies the preset sequence conditions as the cerebral vascular image sequence.

[0156] Based on the zero-order time decay curves of each cerebral vascular voxel in the cerebral vascular image sequence, the peak time of each cerebral vascular voxel is determined.

[0157] A brain vascular image sequence consisting of at least one brain vascular voxel whose peak time meets the preset time range corresponding to the blood vessel type is regarded as a brain vascular image sequence corresponding to the blood vessel type.

[0158] In an optional embodiment, the device further includes:

[0159] The preset time range determination module is used to generate a cerebral vascular voxel-peak time histogram based on the peak time corresponding to each cerebral vascular voxel.

[0160] The minimum peak time between the peak times of two peaks in the cerebral vascular voxel-peak time histogram is used as the preset time threshold.

[0161] Based on the blood vessel type and preset time threshold, a preset time range is constructed;

[0162] Specifically, when the blood vessel type is a vein, the minimum boundary value in the preset time range is the preset time threshold value, and when the blood vessel type is an artery, the maximum boundary value in the preset time range is the preset time threshold value.

[0163] In one optional embodiment, the unit for determining the sequence of blood vessels of interest is specifically used for:

[0164] Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, the peak time corresponding to each cerebral vascular voxel is determined.

[0165] Obtain the time difference between each peak time and the preset time threshold;

[0166] For each vessel of interest, the image sequence consisting of at least one brain vessel voxel whose time difference satisfies the time difference range corresponding to the vessel of interest is taken as the image sequence of the vessel of interest.

[0167] In an optional embodiment, the device further includes:

[0168] The second brain image sequence determination module is used to obtain the skull coordinates in the first frame of the original brain image sequence, and perform rigid registration operation on the original brain images other than the first frame of the original brain image sequence according to the skull coordinates to obtain the registered brain image sequence.

[0169] Based on a preset skull threshold, a segmentation operation is performed on the first frame of the registered brain image in the registered brain image sequence to obtain a skull mask image;

[0170] Based on the skull mask image, a deletion operation is performed on each registered brain image in the registered brain image sequence to obtain a second brain image sequence to be detected.

[0171] The stroke lesion volume monitoring device provided in this embodiment of the invention can execute the stroke lesion volume monitoring method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0172] Figure 9 This is a schematic diagram of an electronic device provided according to one embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0173] like Figure 9 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor 11. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0174] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information or data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0175] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the method for monitoring the volume of stroke lesions provided in the above embodiments.

[0176] In some embodiments, the method for monitoring the volume of a stroke lesion provided in the above embodiments can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for monitoring the volume of a stroke lesion described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the method for monitoring the volume of a stroke lesion by any other suitable means (e.g., by means of firmware).

[0177] Various embodiments of the systems and techniques described above herein can be implemented in the following systems or combinations thereof: digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0178] Computer programs for implementing the method for monitoring the volume of stroke lesions according to the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0179] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable storage medium. Examples of machine-readable storage media include, based on an electrical connection of at least one wire, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0180] To provide interaction with a user, the systems and techniques described herein can be implemented on a terminal device having: a display device for displaying information to the user (e.g., a cathode-ray tube (CRT) or liquid crystal display (LCD) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the terminal device. Other types of devices can also provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0181] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0182] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system. It addresses the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0183] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0184] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for monitoring the volume of a stroke lesion, characterized in that, include: Acquire a dataset of images to be detected corresponding to the type of stroke to be detected; wherein, the dataset of images to be detected contains a first brain image to be detected corresponding to the baseline acquisition time and a sequence of images of at least one vessel of interest; Based on the first brain image to be detected, determine the baseline lesion volume corresponding to the baseline acquisition time; For each vessel of interest, the spatiotemporal features of the vessel of interest are determined based on the image sequence of the vessel of interest. The spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration are input into a pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time. Wherein, the volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time; The acquisition of the image dataset corresponding to the type of stroke to be detected includes: Obtain the vessel type of at least one vessel of interest corresponding to the type of stroke to be detected; Based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, a segmentation operation is performed on the second brain image sequence to be detected to obtain a cerebral vascular image sequence corresponding to the blood vessel type. Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, a segmentation operation is performed on the cerebral vascular image sequence to obtain the image sequence of each vessel of interest. The first brain image to be detected and the sequence of images of each of the blood vessels of interest are added to the image dataset to be detected corresponding to the type of stroke to be detected.

2. The method according to claim 1, characterized in that, The step of determining the spatiotemporal features of the blood vessel of interest based on the image sequence corresponding to the blood vessel of interest includes: Perform a straightening operation on at least two images of the vessel of interest in the image sequence corresponding to the vessel of interest to obtain at least two straightened vessel images; Each of the straightened blood vessel images is input into a pre-trained feature extraction model to obtain the spatiotemporal features of the blood vessel of interest.

3. The method according to claim 2, characterized in that, The step of performing a straightening operation on at least two blood vessel images in the image sequence corresponding to the blood vessel of interest to obtain at least two straightened blood vessel images includes: For each image of a vessel of interest, obtain the coordinates of the center points corresponding to at least two vessel center points on the vessel centerline of the image of interest. Based on the coordinates of each of the aforementioned center points, determine the tangent vector corresponding to each of the aforementioned blood vessel center points; Based on the coordinates of each of the center points and the tangent vectors, determine the normal plane coordinates corresponding to at least two blood vessel center points respectively; Based on the coordinates of each normal plane, the straightened blood vessel image corresponding to the blood vessel image of interest is determined.

4. The method according to claim 1, characterized in that, The step of performing a segmentation operation on the second brain image sequence to obtain a cerebral vascular image sequence corresponding to the blood vessel type based on the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected includes: Based on at least one time decay high-order curve corresponding to each brain voxel in the second brain image sequence to be detected, determine the sum sequence of curves corresponding to each brain voxel. The image sequence consisting of at least one brain voxel in the second brain image sequence to be detected, whose curve sum sequence satisfies the preset sequence conditions, is taken as the cerebral vascular image sequence. Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, the peak time corresponding to each cerebral vascular voxel is determined. The image sequence consisting of at least one cerebral vascular voxel whose peak time satisfies a preset time range corresponding to the vascular type is taken as the cerebral vascular image sequence corresponding to the vascular type.

5. The method according to claim 4, characterized in that, The method further includes: Based on the peak time of each of the described cerebrovascular voxels, a cerebrovascular voxel-peak time histogram is generated. The minimum peak time between the peak times of the two peaks in the cerebral vascular voxel-peak time histogram is used as a preset time threshold. Based on the blood vessel type and the preset time threshold, a preset time range is constructed; Wherein, when the blood vessel type is a vein, the minimum boundary value in the preset time range is the preset time threshold value, and when the blood vessel type is an artery, the maximum boundary value in the preset time range is the preset time threshold value.

6. The method according to claim 1, characterized in that, The step of performing a segmentation operation on the cerebral vascular image sequence based on the time decay zero-order curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence to obtain image sequences of each vessel of interest, includes: Based on the zero-order time decay curves corresponding to each cerebral vascular voxel in the cerebral vascular image sequence, the peak time corresponding to each cerebral vascular voxel is determined. Obtain the time difference between each peak time and the preset time threshold; For each vessel of interest, an image sequence consisting of at least one brain vessel voxel whose time difference satisfies the time difference range corresponding to the vessel of interest is taken as the image sequence of the vessel of interest.

7. The method according to claim 1, characterized in that, The method further includes: Obtain the skull coordinates in the first frame of the original brain image sequence, and perform rigid registration on the original brain images other than the first frame of the original brain image sequence based on the skull coordinates to obtain the registered brain image sequence. Based on a preset skull threshold, a segmentation operation is performed on the first frame of the registered brain image in the registered brain image sequence to obtain a skull mask image; Based on the skull mask image, a deletion operation is performed on each registered brain image in the registered brain image sequence to obtain a second brain image sequence to be detected.

8. A device for monitoring the volume of a stroke lesion, characterized in that, include: The image dataset acquisition module is used to acquire an image dataset corresponding to the type of stroke to be detected; wherein, the image dataset to be detected includes a first brain image to be detected corresponding to the baseline acquisition time and an image sequence of at least one vessel of interest; The baseline lesion volume determination module is used to determine the baseline lesion volume corresponding to the baseline acquisition time based on the first brain image to be detected. The vascular spatiotemporal feature determination module is used to determine the vascular spatiotemporal features of each vessel of interest based on the image sequence of the vessel of interest corresponding to the vessel of interest. The lesion volume monitoring output module is used to input the spatiotemporal characteristics of each blood vessel, the baseline lesion volume, and the volume monitoring duration into the pre-trained lesion volume prediction model to obtain the output lesion volume corresponding to the target monitoring time. Wherein, the volume monitoring duration represents the time length between the baseline acquisition time and the target monitoring time; The module for acquiring the image dataset to be detected includes: The vessel type acquisition unit is used to acquire the vessel type of at least one vessel of interest corresponding to the type of stroke to be detected. The cerebral vascular image sequence determination unit is used to perform a segmentation operation on the second brain image sequence to be detected according to the time decay curves corresponding to each brain voxel in the second brain image sequence to be detected, to obtain a cerebral vascular image sequence corresponding to the blood vessel type. The unit for determining the image sequence of vessels of interest is used to perform a segmentation operation on the brain vessel image sequence according to the time decay zero-order curves corresponding to each brain vessel voxel in the brain vessel image sequence to obtain the image sequence of vessels of interest corresponding to each vessel of interest; and to add the first brain image to be detected and each image sequence of vessels of interest to the image dataset to be detected corresponding to the type of stroke to be detected.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory that is communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the method for monitoring the volume of a stroke lesion as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for monitoring the volume of a stroke lesion as described in any one of claims 1-7.

Images