A method, device and apparatus for detecting intracranial aneurysms based on magnetic resonance imaging

Abstract

The embodiments of this specification disclose a method, device and apparatus for detecting intracranial aneurysms based on nuclear magnetic resonance imaging. The method comprises: obtaining image data to be processed, wherein the image data to be processed is nuclear magnetic resonance imaging data containing intracranial aneurysms; using sliding window processing to extract the first image block of the image data to be processed; inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image; using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, and obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model; performing connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result.

Description

Technical Field

[0001] This specification relates to the fields of artificial intelligence and medical imaging, and in particular to a method, device, and equipment for detecting intracranial aneurysms based on magnetic resonance imaging. Background Art

[0002] Intracranial aneurysms are tumor-like protrusions formed due to local congenital defects or damage to the walls of intracranial arteries, resulting in abnormal expansion of the local lumen. Intracranial aneurysms vary in size, from small aneurysms less than 5mm to giant aneurysms greater than 25mm; their location is unstable and may appear anywhere in the entire cerebral blood vessels, and the probability of aneurysms appearing in different locations is uncertain; their shapes vary, including sac-shaped, fusiform, and aneurysms with sub-sacs, which are very irregular. Some fusiform aneurysms are particularly close to blood vessels and difficult to identify, making them very easy to miss. In the existing technology, the identification of intracranial aneurysms often uses a "manual" method, that is, based on the doctor's experience, the patient's medical images are read to identify the aneurysm. This manual method is labor-intensive and requires manual reaction recognition, which takes a long time and has a low accuracy rate.

[0003] Magnetic resonance imaging (MRI) is currently a routine screening method for aneurysms and other conditions, serving as a necessary scanning method for initial consultations, physical examinations, and other procedures. Time-of-flight magnetic resonance angiography (TOF-MRA) is a key imaging technique used in clinical diagnosis of vascular diseases. This technique produces angiographic effects by suppressing the signal from stationary tissue and enhancing the signal from flowing blood, offering advantages such as high contrast, high spatial resolution, and wide coverage. Summary of the Invention

[0004] The embodiments of this specification provide a method, device and equipment for detecting intracranial aneurysms based on magnetic resonance imaging, which are used to solve the following technical problems: being able to quickly and accurately detect potential intracranial aneurysms as an auxiliary means for rapid screening of intracranial aneurysms.

[0005] To solve the above technical problems, the embodiments of this specification are implemented as follows:

[0006] The embodiments of this specification provide a method for detecting intracranial aneurysms based on magnetic resonance imaging, comprising:

[0007] Acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing an intracranial aneurysm;

[0008] Using sliding window processing to extract the first image block of the image data to be processed;

[0009] Inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image;

[0010] Using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model;

[0011] Performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result;

[0012] The acquiring of the second image block of the image data to be processed includes:

[0013] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0014] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0015] The embodiments of this specification provide a device for detecting intracranial aneurysms based on magnetic resonance imaging, comprising:

[0016] an acquisition module for acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing intracranial aneurysms;

[0017] an extraction module, which extracts a first image block of the image data to be processed by using a sliding window process;

[0018] a blood vessel segmentation module, inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image;

[0019] an aneurysm prediction module, inputting a second image block of the image data to be processed, obtained based on the blood vessel segmentation mask image, into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, and obtaining an aneurysm prediction result corresponding to the second image block, wherein the blood vessel segmentation model is cascaded with the aneurysm segmentation model;

[0020] an aneurysm segmentation module, performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result;

[0021] The acquiring of the second image block of the image data to be processed includes:

[0022] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0023] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0024] An embodiment of this specification further provides an electronic device, including:

[0025] at least one processor; and,

[0026] a memory communicatively connected to the at least one processor; wherein,

[0027] The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0028] Acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing an intracranial aneurysm;

[0029] Using sliding window processing to extract the first image block of the image data to be processed;

[0030] Inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image;

[0031] Using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model;

[0032] Performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result;

[0033] The acquiring of the second image block of the image data to be processed includes:

[0034] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0035] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0036] The embodiment of this specification obtains image data to be processed, which is magnetic resonance imaging data containing intracranial aneurysms; uses sliding window processing to extract the first image block of the image data to be processed; inputs the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image; uses the blood vessel segmentation mask image as sampling position reference data, inputs the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, and obtains an aneurysm prediction result corresponding to the second image block, and the blood vessel segmentation model is cascaded with the aneurysm segmentation model; performs connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result, which can realize automatic segmentation of intracranial aneurysms based on magnetic resonance images, with fast processing speed and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0038] Figure 1 A framework diagram of a method for detecting intracranial aneurysms based on magnetic resonance imaging provided in an embodiment of this specification;

[0039] Figure 2 A schematic diagram of a method for detecting intracranial aneurysms based on magnetic resonance imaging provided in an embodiment of this specification;

[0040] Figure 3 A schematic diagram of a 3D-Unet structure network provided in an embodiment of this specification;

[0041] Figure 4 A schematic diagram of a device for detecting intracranial aneurysms based on magnetic resonance imaging provided in an embodiment of this specification;

[0042] Figure 5 It is a schematic structural diagram of the detection device of the present invention. DETAILED DESCRIPTION

[0043] In order to help those skilled in the art better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0044] Segmentation of intracranial aneurysms based on whole-brain 3D imaging data often involves using technical means to identify intracranial aneurysms from whole-brain 3D images, thereby achieving segmentation of intracranial aneurysms. This method has a large search range, slow processing speed, and a high false positive rate. Another method for segmenting intracranial aneurysms from whole-brain 3D imaging data is to convert the whole-brain 3D image into a 2D-MIP (maximum intensity projection image) for processing and identification, thereby achieving segmentation of intracranial aneurysms. However, due to occlusion, this method makes the 2D-MIP image inaccurate, resulting in unclear morphology of small aneurysms. Based on this, a new method for detecting intracranial aneurysms is needed.

[0045] Figure 1 This is a framework diagram of a method for detecting intracranial aneurysms based on magnetic resonance imaging provided in an embodiment of this specification, such as Figure 1 As shown, the MRI data to be processed is input into a pre-trained vascular segmentation model to obtain a vascular segmentation mask image. The obtained vascular segmentation mask image is used as sampling position reference data. After extracting image blocks from the MRI data to be processed, the pre-trained aneurysm segmentation model is input to obtain an aneurysm segmentation mask image, thereby achieving aneurysm segmentation. To achieve better segmentation results, reduce the amount of calculation, and improve segmentation accuracy, the vascular segmentation mask image is further multiplied with the MRI data to be processed, and tissue in the unmasked area is removed to obtain a dot product calculation result. The obtained vascular segmentation mask image is then used as sampling position reference data. Image blocks are extracted from the dot product calculation result and input into the pre-trained aneurysm segmentation model to obtain an aneurysm segmentation mask image, thereby achieving aneurysm segmentation.

[0046] Figure 2 A schematic diagram of a method for detecting intracranial aneurysms based on magnetic resonance imaging provided in an embodiment of this specification is shown in FIG. Figure 2 As shown, the detection method includes:

[0047] Step S201: Acquire image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing intracranial aneurysms.

[0048] In the embodiments of this specification, the image data to be processed is magnetic resonance imaging data, specifically, 3D TOF-MRA imaging data or CTA imaging data. This embodiment of the specification uses 3D TOF-MRA imaging data as an example to illustrate the detection method provided in the embodiments of this specification.

[0049] 3D TOF-MRA is a non-invasive MRA bright blood imaging technique based on the inflow enhancement effect that does not require the injection of contrast agents. It is the most widely used MRA technique in clinical practice. The image data to be processed in the embodiments of this specification is 3D TOF-MRA image data containing intracranial aneurysms.

[0050] To achieve better aneurysm segmentation results, the image data to be processed must undergo further preprocessing. Preprocessing methods include at least grayscale normalization and vessel enhancement. In a specific embodiment, grayscale normalization is an image conversion method that reduces or even eliminates grayscale inconsistencies within the image while preserving diagnostically valuable grayscale differences. Common algorithms, depending on the basis of grayscale conversion, include histogram-based grayscale normalization and image content-based grayscale normalization. Angiographic image enhancement enhances vascular structures while suppressing background and non-vascular structures, ensuring that the enhanced results are as close to the actual vascular structure as possible. Methods include: Hessian matrix-based methods; diffusion equation-based methods; strain energy-based methods; and polar coordinate intensity profile-based methods. The vessel enhancement diffusion algorithm combines a Hessian matrix-based Frangi filter with a nonlinear anisotropic diffusion equation. The Frangi vessel measurement function is modified, with a smoothing constraint added to ensure continuity, allowing it to be directly used to construct the diffusion tensor.

[0051] It should be noted that the specific method of preprocessing the image data to be processed does not constitute a limitation to the present application.

[0052] Step S203: adopting sliding window processing to extract the first image block of the image data to be processed.

[0053] In the embodiment of this specification, the extracting the first image block of the image data to be processed by using sliding window processing specifically includes:

[0054] Based on the sliding window method, a sliding window with a size of first preset pixels is used, and adjacent sliding windows have a 50% overlap. A pixel block of the first preset pixels is generated during the sliding process as the first image block of the image data to be processed.

[0055] In this embodiment of the present disclosure, the first preset pixel size can be selected as 56*224*224. Through sliding window processing, a 56*224*224 pixel block can be obtained, that is, the first image block is a 56*224*224 pixel block. Furthermore, in this embodiment of the present disclosure, the resolution of the pixel block of the first preset pixel size can be 0.5*0.4*0.4.

[0056] Step S205: inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image.

[0057] The pre-trained vascular segmentation model can output a vascular segmentation mask image after inputting the first pixel block of the image data to be processed, that is, the vascular segmentation mask image is used as the position information of the aneurysm image block, so that global processing is not required, and the aneurysm segmentation result is then obtained through the aneurysm segmentation model.

[0058] In the embodiment of this specification, the blood vessel segmentation model is a model obtained by training a 3D-Unet structure network, specifically including:

[0059] Extracting image blocks from the blood vessel-annotated clinical image data to obtain a third image block for training a blood vessel segmentation model;

[0060] Performing data enhancement on the third image block used for blood vessel segmentation model training to obtain an enhanced third image block;

[0061] The third image block is input into the 3D-Unet structure network for model training, and the model is optimized using a loss function combining Dice and CE to obtain a blood vessel segmentation model.

[0062] In the embodiment of this specification, extracting image blocks from the blood vessel-annotated clinical image data to obtain a third image block for training a blood vessel segmentation model specifically includes:

[0063] After preprocessing the clinical image data, blood vessels are labeled to obtain blood vessel labeled clinical image data;

[0064] Random sampling is adopted to sample from the clinical image data with blood vessel annotations to obtain a pixel block of a second preset pixel as a third image block for training the blood vessel segmentation model.

[0065] In the embodiment of the present specification, the second preset pixel can be selected as 96*96*96, and a 96*96*96 pixel block can be obtained by random sampling, that is, the second image block is a 96*96*96 pixel block.

[0066] In the embodiments of this specification, vascular annotation of clinical imaging data is achieved by generating initial vascular labels based on technical means, and then obtaining accurate vascular labels through manual inspection. Among them, the method of generating initial vascular labels based on technical means can adopt threshold methods, region growing methods, etc., which are all existing technologies and will not be described in detail here. It should be noted that vascular annotation includes foreground and background, and the area marked with vascular labels is regarded as foreground, and the other areas are regarded as background. In principle, for the accuracy of the training results, the sampling ratio of foreground to background is as close to 1:1 as possible.

[0067] In order to achieve diversity in training samples and expand the data of training samples, the third image block used for vascular segmentation model training needs to be further enhanced. In the embodiment of this specification, the data enhancement of the third image block used for vascular segmentation model training to obtain the enhanced third image block specifically includes:

[0068] The third image block used for blood vessel segmentation model training is subjected to one or more operations of rotation, scaling, flipping, blurring, and gama enhancement to perform data enhancement to obtain enhanced third image data.

[0069] In the embodiments of this specification, the loss function of the combination of Dice and CE is expressed as:

[0070] Loss1=1-Dice-CE

[0071]

[0072]

[0073] therefore,

[0074] Among them, Loss1 is the loss function combining Dice and CE;

[0075] g i Indicates whether the classification of the pixel as a blood vessel is correct, 1 if correct, 0 otherwise;

[0076] s i Indicates the probability value of a pixel being a blood vessel pixel;

[0077] N is the number of all pixels in all image blocks.

[0078] In the embodiment of this specification, the DICE of the validation set of the blood vessel segmentation model can reach at least 0.95, so accurate segmentation of the blood vessels can be achieved.

[0079] In order to further understand the sequence of the blood vessel segmentation model, it will be explained in conjunction with the 3D-Unet structure network. Figure 3 A schematic diagram of a 3D-Unet structure network provided in an embodiment of this specification.

[0080] Step S207: Using the blood vessel segmentation mask image as sampling position reference data, the second image block of the image data to be processed is input into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, and an aneurysm prediction result corresponding to the second image block is obtained. The blood vessel segmentation model is cascaded with the aneurysm segmentation model.

[0081] In the embodiment of the present specification, the aneurysm segmentation model can obtain a probability value of aneurysm prediction based on the input second image block.

[0082] In this embodiment of the present specification, obtaining the second image block of the image data to be processed includes:

[0083] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0084] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0085] It should be noted that in the embodiments of this specification, the second image block used as input for the aneurysm segmentation model can be extracted from the image data to be processed, using the vascular segmentation mask image as sampling position reference data. Alternatively, the input for the aneurysm segmentation model can be extracted from a dot product image obtained by performing a dot product calculation between the vascular segmentation mask image and the image data to be processed, using the vascular segmentation mask image as sampling position reference data. The dot product image is an image that filters out tissue in the non-masked area.

[0086] In an embodiment of the present specification, based on the vascular segmentation mask image, a dot product calculation is performed on the volume data corresponding to the image data to be processed to obtain a dot product calculation image. That is, based on the mask position of the vascular segmentation mask image, a dot product calculation is performed on the volume data corresponding to the image data to be processed to filter out tissue in the non-masked area, and a value of 0 is assigned to obtain a dot product calculation image. This method can ensure that the mask position remains unchanged and can filter tissue in the non-masked area, thereby reducing the amount of calculation in subsequent calculations and improving the accuracy of the results.

[0087] In the embodiment of this specification, the aneurysm segmentation model is a model obtained by training a 3D-Unet structure network, specifically including:

[0088] Extracting image blocks from volume data corresponding to the clinical image data to obtain a fourth image block for training an aneurysm segmentation model;

[0089] Performing data enhancement on the fourth image block used for aneurysm segmentation model training to obtain an enhanced fourth image block;

[0090] The fourth image block is input into the 3D-Unet structure network for model training, and the model is optimized using a loss function combining Dice and TopK to obtain an aneurysm segmentation model.

[0091] Volume data is composed of voxels, which are basic volume elements and can also be understood as points or small areas with arrangement and color in three-dimensional space. Usually voxels belong to a fixed grid, so volume data can be stored as a table. Volume data can be treated as a csv file stored locally. Commonly, the data set is divided into several slices, and each slice is stored as a bitmap image. Due to the complex compression algorithms that can be applied to images, the model size is significantly reduced. In the embodiments of this specification, the methods for obtaining volume data of clinical imaging data are all existing technologies and will not be repeated here.

[0092] In the embodiment of this specification, extracting image blocks from volume data corresponding to the clinical image data to obtain a fourth image block for training the aneurysm segmentation model specifically includes:

[0093] Based on random sampling, image blocks are extracted from the data belonging to the blood vessel labels in the volume data corresponding to the clinical image data to generate a 96*96*96 pixel block as the third image block for aneurysm segmentation model training.

[0094] In the embodiments of this specification, the loss function of the combination of Dice and TopK is expressed as:

[0095] Loss2 = 1-Dice-TopK

[0096]

[0097]

[0098] therefore,

[0099] Among them, Loss2 is the loss function that combines Dice and TopK;

[0100] h i Indicates whether the classification of the pixel point as an intracranial aneurysm is correct, 1 if correct, 0 otherwise;

[0101] j iIt represents the probability value that the pixel point is a pixel point of an intracranial aneurysm;

[0102] N is the number of all pixel points in all image patches;

[0103] t is a preset threshold;

[0104] 1{j i <t} represents that if the probability of predicting an intracranial aneurysm is less than t, the calculated value is 1, otherwise it is 0.

[0105] In the embodiments of this specification, the preset threshold t ∈ (0, 1]. In the embodiments of this specification, the preset threshold t is preferably 0.1. It should be particularly noted that it is precisely because in the training of the aneurysm segmentation model, sampling is performed on the data belonging to the blood vessel labels, and in the specific application of the model, the blood vessel segmentation mask image is used as the sampling position reference data that the cascade of the blood vessel segmentation model and the aneurysm segmentation model can be realized.

[0106] Due to the small size of aneurysms and uneven distribution, etc., the segmentation of aneurysms is relatively difficult. In the embodiments of this specification, the DICE of the validation set of the aneurysm segmentation model can reach 0.7, achieving a good segmentation result.

[0107] Step S209: Perform connected component analysis on the aneurysm prediction result corresponding to the second image patch to obtain an aneurysm segmentation result. [[ID=XXX]]

[0108] In the embodiments of this specification, the performing connected component analysis on the aneurysm prediction result corresponding to the second image patch to obtain an aneurysm segmentation result specifically includes:

[0109] Perform connected component analysis on the aneurysm prediction result corresponding to the second image patch, and use the connected regions in the connected component analysis result that are not less than the preset number of pixel points as candidate aneurysm regions;

[0110] Or [[ID=XXX]] [[ID=XXX]]

[0111] Perform connected component analysis on the aneurysm prediction result corresponding to the second image patch, sort the connected component analysis result in descending order of the connected components, and use the preset number of connected components as candidate aneurysm regions;

[0112] Use the aneurysm regions in the candidate aneurysm regions whose probability values are greater than or equal to the preset probability value as the aneurysm segmentation result.

[0113] In the embodiment of the present specification, the number of preset pixels may be 11. That is, a connected region with a number of pixels not less than 11 in the connected domain analysis result is used as a candidate aneurysm region. The candidate aneurysm region and its corresponding probability value are further used, based on the candidate aneurysm region and its corresponding probability value, to select an aneurysm region with a probability value greater than or equal to the preset probability value as the aneurysm segmentation result. If a connected region with a number of pixels less than 11 in the connected domain analysis result is present, the result of the connected region is considered to be an incorrect result, and the connected region with a number of pixels less than 11 cannot be used as a candidate aneurysm region.

[0114] In the embodiment of the present specification, the preset number is preferably 5, that is, the connected domain analysis results are sorted in order of connected domains from large to small, and the first 5 connected domains are used as candidate aneurysm regions. Further, based on the candidate aneurysm regions and their corresponding probability values, the aneurysm region whose probability value corresponding to the candidate aneurysm region is greater than or equal to the preset probability value is used as the aneurysm segmentation result.

[0115] In a specific embodiment, the probability value corresponding to the candidate aneurysm region can be determined based on the probability value of each pixel belonging to an aneurysm, or other methods can be used. The method for determining the probability value corresponding to the candidate aneurysm region does not constitute a limitation of this application. The value of the preset probability value can be determined based on the preset business scenario. In one embodiment of this specification, the preset probability value is preferably 0.5.

[0116] In the embodiment of the present specification, the aneurysm segmentation result includes the aneurysm location of the image data to be processed and the probability of the aneurysm of the image data to be processed.

[0117] The method provided in the embodiments of this specification can realize automatic segmentation of intracranial aneurysms based on magnetic resonance imaging, with fast processing speed and high accuracy.

[0118] The above content introduces in detail a method for detecting intracranial aneurysms based on nuclear magnetic resonance imaging. Correspondingly, this specification also provides a device for detecting intracranial aneurysms based on nuclear magnetic resonance imaging. Figure 4 A schematic diagram of a device for detecting intracranial aneurysms based on magnetic resonance imaging is provided in an embodiment of this specification, as shown in FIG. Figure 4 As shown, the detection device includes:

[0119] An acquisition module 401 acquires image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing intracranial aneurysms;

[0120] An extraction module 403 extracts a first image block of the image data to be processed using a sliding window process;

[0121] The blood vessel segmentation module 405 inputs the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image;

[0122] an aneurysm prediction module 407, using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model;

[0123] an aneurysm segmentation module 409 , performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result;

[0124] The acquiring of the second image block of the image data to be processed includes:

[0125] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0126] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0127] An embodiment of this specification further provides an electronic device, including:

[0128] at least one processor; and,

[0129] a memory communicatively connected to the at least one processor; wherein,

[0130] The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0131] Acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing an intracranial aneurysm;

[0132] Using sliding window processing to extract the first image block of the image data to be processed;

[0133] Inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image;

[0134] Using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model;

[0135] Performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result;

[0136] The acquiring of the second image block of the image data to be processed includes:

[0137] performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image;

[0138] Based on the volume data corresponding to the image data to be processed or the dot product calculated image, sliding window processing is used to obtain the second image block.

[0139] Those skilled in the art will appreciate that various aspects of the present invention may be implemented as systems, methods, or program products. Accordingly, various aspects of the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, microcode, etc.), or in a combination of hardware and software, collectively referred to herein as "circuits," "modules," or "platforms."

[0140] Figure 5 This is a schematic diagram of the structure of the detection device of the present invention. Figure 5 An electronic device 500 according to this embodiment of the present invention will be described. Figure 5 The electronic device 500 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0141] like Figure 5 As shown, electronic device 500 is implemented as a general-purpose computing device. Components of electronic device 500 may include, but are not limited to, at least one processing unit 510, at least one storage unit 520, a bus 530 connecting different platform components (including storage unit 520 and processing unit 510), a display unit 540, and the like.

[0142] The storage unit stores program code, which can be executed by the processing unit 510, so that the processing unit 510 performs the steps of various exemplary embodiments of the present invention described in the above detection method section of this specification. For example, the processing unit 510 can perform the steps of the detection method.

[0143] The storage unit 520 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 521 and / or a cache memory unit 522 , and may further include a read-only memory unit (ROM) 523 .

[0144] The storage unit 520 may also include a program / utility 524 having a set (at least one) of program modules 525, such program modules 525 including but not limited to: a processing system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0145] Bus 530 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0146] The electronic device 500 may also communicate with one or more external devices 570 (e.g., a keyboard, a pointing device, a Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 500, and / or any device that enables the electronic device 500 to communicate with one or more other computing devices (e.g., a router, a modem, etc.). Such communication may occur through an input / output (I / O) interface 550.

[0147] Furthermore, the electronic device 500 can also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 560. The network adapter 560 can communicate with other modules of the electronic device 500 via the bus 530. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with the electronic device 500, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0148] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0149] The various embodiments in this specification are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences between the other embodiments. In particular, the device, electronic device, and non-volatile computer storage medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simplified. For relevant details, refer to the descriptions of the method embodiments.

[0150] The apparatus, electronic device, and non-volatile computer storage medium provided in the embodiments of this specification correspond to the method. Therefore, the apparatus, electronic device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device, and non-volatile computer storage medium will not be repeated here.

[0151] In the 1990s, technological improvements could be clearly distinguished as either hardware improvements (for example, improvements to circuit structures such as diodes, transistors, and switches) or software improvements (improvements to process flows). However, with the advancement of technology, many process flow improvements today can now be considered direct improvements to hardware circuit structures. Designers almost always create the corresponding hardware circuit structure by programming the improved process flow into the hardware circuit. Therefore, it cannot be said that a process flow improvement cannot be implemented using hardware modules. For example, a programmable logic device (PLD), such as a field programmable gate array (FPGA), is an integrated circuit whose logical function is determined by user programming. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to hire a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly done using "logic compiler" software. This is similar to the software compiler used when developing programs. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There is not just one HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art will also understand that by simply programming the method flow in one of these hardware description languages ​​and then programming it into an integrated circuit, a hardware circuit that implements the logic method flow can be easily obtained.

[0152] The controller can be implemented in any suitable manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91 SAM, Microchip PIC1 8F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that in addition to implementing the controller in a purely computer-readable program code manner, it is entirely possible to implement the same function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules that implement the method and structures within the hardware component.

[0153] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0154] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0155] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0156] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data optimization device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data optimization device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data optimization device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions may also be loaded onto a computer or other programmable data optimization device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide for implementing the process described in the flow. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0159] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0160] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0161] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0162] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0163] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0164] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0165] The foregoing is merely an embodiment of the present invention and is not intended to limit the present application. For those skilled in the art, various modifications and variations may be made to the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting intracranial aneurysms based on magnetic resonance imaging, characterized in that: The method comprises: Acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing an intracranial aneurysm; Using sliding window processing to extract the first image block of the image data to be processed; Inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image; Using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model; Performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result; The acquiring of the second image block of the image data to be processed includes: performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image; Based on the dot product calculation image, sliding window processing is used to obtain the second image block.

2. The detection method according to claim 1, wherein The step of extracting the first image block of the image data to be processed by using a sliding window process specifically includes: Based on the sliding window method, a sliding window with a size of first preset pixels is used, and adjacent sliding windows have a 50% overlap. A pixel block of the first preset pixels is generated during the sliding process as the first image block of the image data to be processed.

3. The detection method according to claim 1, wherein The blood vessel segmentation model is a model obtained through 3D-Unet structural network training, specifically including: Extracting image blocks from the blood vessel-annotated clinical image data to obtain a third image block for training a blood vessel segmentation model; Performing data enhancement on the third image block used for blood vessel segmentation model training to obtain an enhanced third image block; The third image block is input into the 3D-Unet structure network for model training, and the model is optimized using a loss function combining Dice and CE to obtain a blood vessel segmentation model.

4. The detection method according to claim 3, wherein The step of extracting image blocks from the blood vessel labeled clinical image data to obtain a third image block for training the blood vessel segmentation model specifically includes: After preprocessing the clinical image data, blood vessels are labeled to obtain blood vessel labeled clinical image data; Random sampling is adopted to sample from the clinical image data with blood vessel annotations to obtain a pixel block of a second preset pixel as a third image block for training the blood vessel segmentation model.

5. The method according to claim 3, wherein The step of performing data enhancement on the third image block used for training the blood vessel segmentation model to obtain an enhanced third image block specifically includes: The third image block used for blood vessel segmentation model training is subjected to one or more operations of rotation, scaling, flipping, blurring, and gama enhancement to perform data enhancement to obtain enhanced third image data.

6. The detection method according to claim 3, wherein The loss function of the combination of Dice and CE is expressed as: Among them, Loss1 is the loss function combining Dice and CE; gi indicates whether the classification of the pixel as a blood vessel is correct, 1 if correct, 0 otherwise; si represents the probability value of a pixel being a blood vessel pixel; N is the number of all pixel points in all image blocks.

7. The detection method according to claim 1, wherein The aneurysm segmentation model is a model obtained by training based on a 3D-Unet structure network, specifically including: Performing image block extraction on the volume data corresponding to the clinical image data to obtain a fourth image block for training the aneurysm segmentation model; Performing data enhancement on the fourth image block for training the aneurysm segmentation model to obtain an enhanced fourth image block; Inputting the fourth image block into a 3D-Unet structure network for model training, and optimizing the model with a loss function combining Dice and TopK to obtain an aneurysm segmentation model.

8. The detection method according to claim 7, wherein The performing image block extraction on the volume data corresponding to the clinical image data to obtain a fourth image block for training the aneurysm segmentation model specifically includes: Based on a random sampling method, performing image block extraction on the data belonging to the blood vessel label in the volume data corresponding to the clinical image data to generate a pixel block of 96*96*96 pixels as the fourth image block for training the aneurysm segmentation model.

9. The detection method according to claim 7, wherein The expression of the loss function combining Dice and TopK is: where Loss2 is the loss function combining Dice and TopK; hi represents whether the classification of the pixel point belonging to the intracranial aneurysm is correct. If it is correct, it is 1, otherwise it is 0; ji represents the probability value of the pixel point being an intracranial aneurysm pixel point; N is the number of all pixel points in all image blocks; t is a preset threshold; 1{ji<t} means that if the predicted probability of belonging to an intracranial aneurysm is less than t, the calculated value is 1, otherwise it is 0.

10. The detection method according to claim 1, wherein The performing connected component analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result specifically includes: ​ ​ ​ ​ 11. The detection method according to claim 1, wherein ​ ​ 12. A device for detecting intracranial aneurysms based on nuclear magnetic resonance imaging, characterized in that: ​ ​ ​ ​ an aneurysm prediction module, using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model; an aneurysm segmentation module, performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result; The acquiring of the second image block of the image data to be processed includes: performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image; Based on the dot product calculation image, sliding window processing is used to obtain the second image block.

13. An electronic device comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor. processors to execute, so that the at least one processor is capable of: Acquiring image data to be processed, wherein the image data to be processed is magnetic resonance imaging data containing an intracranial aneurysm; Using sliding window processing to extract the first image block of the image data to be processed; Inputting the first image block of the image data to be processed into a pre-trained blood vessel segmentation model to obtain a blood vessel segmentation mask image; Using the blood vessel segmentation mask image as sampling position reference data, inputting the second image block of the image data to be processed into a pre-trained aneurysm segmentation model to predict the probability of an aneurysm, obtaining an aneurysm prediction result corresponding to the second image block, and cascading the blood vessel segmentation model with the aneurysm segmentation model; Performing a connected domain analysis on the aneurysm prediction result corresponding to the second image block to obtain an aneurysm segmentation result; The acquiring of the second image block of the image data to be processed includes: performing a dot product calculation based on the blood vessel segmentation mask image and the image data to be processed to obtain a dot product calculation image; Based on the dot product calculation image, sliding window processing is used to obtain the second image block.

Images