Nucleus region determination method, device, computer equipment and readable storage medium

Abstract

The present application relates to a method, device, computer equipment and readable storage medium for determining a nucleus region. The method obtains a registered deformation field by inputting a brain medical image and a brain map of a subject to be examined into a deformation field determination model; deforms the brain map according to the registered deformation field to obtain a deformation map, and determines the nucleus region in the brain medical image according to the deformation map. The method for determining a nucleus region provided by the present application does not need to rely on the experience of doctors and can improve the accuracy of determining the nucleus region.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device, computer equipment and readable storage medium for determining a nucleus region. Background Art

[0002] For patients with neurological diseases such as Parkinson's disease, dystonia, depression, and epilepsy, long-term drug treatment can cause drug resistance in patients. After stopping the drug, they are unable to walk and their limbs move, which seriously affects their quality of life. Therefore, in clinical practice, deep brain stimulation (DBS) is generally used for treatment. DBS uses a small pacemaker-like device placed under the skin of the chest to transmit electrical signals to the brain area that controls movement through extremely fine wires to help control the patient's movement. Related studies have shown that the nuclei of the human brain are closely related to Parkinson's disease and other related neurological diseases, and lesions in the nuclei may cause a variety of movement and cognitive disorders. When using DBS for treatment, electrical signals are transmitted to the nucleus area through wires to control the patient's movement. Therefore, it is necessary to locate the nucleus area in the brain medical image.

[0003] In traditional technology, the nucleus area in the brain medical image is usually determined based on the doctor's experience, which results in low accuracy in determining the nucleus area. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment and readable storage medium for determining a nucleus region in order to solve the above technical problems.

[0005] In a first aspect, an embodiment of the present application provides a method for determining a nucleus region, comprising:

[0006] Inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas set and the brain medical image samples;

[0007] Deform the brain atlas according to the registered deformation field to obtain a deformation atlas;

[0008] According to the deformation atlas, a nucleus region in the brain medical image is determined.

[0009] In one embodiment, before inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field, the nucleus region determination method further includes:

[0010] Acquire an original brain medical image of a subject to be examined;

[0011] Inputting the original brain medical image into the image preprocessing model to obtain the skull separation mask image and the bias field image;

[0012] A brain medical image is determined according to the skull separation mask image and the bias field image.

[0013] In one embodiment, the image preprocessing model includes an encoder, a first decoder, and a second decoder; the original brain medical image is input into the image preprocessing model to obtain a skull separation mask image and a bias field image, including:

[0014] Inputting the original brain medical image into the encoder to obtain L-layer features, where L is an integer greater than or equal to zero;

[0015] Input the L-layer features into the first decoder to obtain a skull separation mask image;

[0016] The L-layer features are input into the second decoder to obtain the bias field image.

[0017] In one embodiment, before inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field, the nucleus region determination method further includes:

[0018] Acquire a brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups;

[0019] According to the age of the subject to be examined, a brain map corresponding to the subject to be examined is determined in the brain map set.

[0020] In one embodiment, obtaining a brain atlas set includes:

[0021] Acquire multiple brain medical image sample sets, wherein each brain medical image sample set includes multiple brain medical image samples corresponding to examination subjects of the same age group, and each brain medical image sample includes a marked nucleus region;

[0022] For each brain medical image sample set, determining a brain atlas corresponding to the brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set;

[0023] A plurality of brain atlases corresponding to a plurality of brain medical image sample sets are used as a brain atlas set.

[0024] In one embodiment, determining a brain atlas corresponding to the brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set includes:

[0025] Based on the affine transformation method, each brain medical image sample in the brain medical image sample set is registered to the standard space;

[0026] According to the distribution of the nucleus area in each brain medical image sample in the standard space, a probability map of the nucleus area is determined;

[0027] According to the acquired brain medical image template and probability map of the nucleus area, the brain atlas corresponding to the brain medical image sample set is determined to obtain the brain atlas set.

[0028] In one embodiment, the method for determining the nucleus region further includes:

[0029] The nucleus region is corrected using the region growing method or the boundary correction method.

[0030] In a second aspect, an embodiment of the present application provides a nucleus region determination device, comprising:

[0031] A registration deformation field determination module is used to input the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas and the brain medical image samples;

[0032] A deformation atlas determination module is used to deform the brain atlas according to the registered deformation field to obtain a deformation atlas;

[0033] The nucleus region determination module is used to determine the nucleus region in the brain medical image according to the deformation map.

[0034] In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method provided in the above embodiment when executing the computer program.

[0035] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method provided in the above embodiment when the computer program is executed by a processor.

[0036] The embodiments of the present application provide a method, device, computer equipment and readable storage medium for determining a nucleus region. The method obtains a registered deformation field by inputting a brain medical image and a brain map of an object to be examined into a deformation field determination model; deforms the brain map according to the registered deformation field to obtain a deformation map, and determines the nucleus region in the brain medical image according to the deformation map. The method for determining a nucleus region provided in the embodiments of the present application deforms the brain map according to the registered deformation field determined by a pre-trained deformation field determination model to obtain a deformation map, so that the nucleus region in the brain medical image is determined according to the deformation map. The nucleus region determined in this way does not need to rely on the doctor's experience, can avoid the influence of the doctor's subjective judgment, and thus can improve the accuracy of the determination of the nucleus region. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the conventional technology, the drawings required for use in the embodiments or the conventional technology descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For different technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 A schematic diagram of the steps of a method for determining a nucleus region provided in one embodiment of the present application;

[0039] Figure 2 A schematic diagram of a deformation field determination model training process provided by an embodiment of the present application;

[0040] Figure 3 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0041] Figure 4 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0042] Figure 5 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0043] Figure 6 A schematic diagram of an image preprocessing model training process provided by one embodiment of the present application;

[0044] Figure 7 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0045] Figure 8 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0046] Fig. 9 A schematic diagram of the steps of another method for determining a nucleus region provided in one embodiment of the present application;

[0047] Fig.10 A schematic diagram of another step flow of determining and applying a nucleus region provided in one embodiment of the present application;

[0048] Fig.11 A schematic diagram of the structure of a nucleus region determination device provided in one embodiment of the present application;

[0049] Fig.12 A schematic diagram of the structure of a computer device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.

[0051] The serial numbers assigned to the components in this article, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning.

[0052] The method for determining the nucleus region provided in the present application can be implemented by a computer device. The computer device includes but is not limited to a control chip, a personal computer, a laptop computer, a smart phone, a tablet computer, and a portable wearable device. The method provided in the present application can be implemented by JAVA software, and can also be applied to other software.

[0053] The technical solution of the present application and how the technical solution of the present application solves the technical problem are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0054] Figure 1 A method for determining a nucleus region is provided for an embodiment of the present application. The present application embodiment uses a computer device as an execution subject to specifically describe the method for determining a nucleus region. The steps of the method include:

[0055] Step 100: Input the brain medical image and brain atlas of the object to be examined into the deformation field determination model to obtain a registered deformation field; wherein the brain atlas includes a probability map of the nucleus area and a brain medical image template, and the deformation field determination model is obtained by training a deep learning network based on a brain atlas set and brain medical image samples.

[0056] In order to execute the method for determining the nucleus region, the computer device needs to obtain a brain medical image of the object to be examined. The brain medical image includes an imaging area corresponding to the nucleus that needs to be located, that is, the nucleus region. The brain medical image corresponding to the object to be examined can be pre-stored in the memory of the computer device, and the computer device can directly obtain it from the memory when needed. This embodiment does not limit the specific method for obtaining the brain medical image, as long as its function can be achieved. In an optional embodiment, the brain medical image can be a magnetic resonance medical image, specifically a magnetic resonance T1 image and / or a magnetic resonance T2 image. This embodiment does not limit the type and quantity of the brain medical image, as long as its function can be achieved.

[0057] The brain map can be pre-stored in the memory of the computer device, and there can be multiple types of brain maps. Different inspection objects correspond to different brain maps, and the brain map includes probability maps of nucleus areas and brain medical image templates. The brain map in this embodiment refers to the brain map corresponding to the object to be inspected. The brain medical image template refers to the brain medical image corresponding to the probability map of the nucleus area, and the probability map of a nucleus area corresponds to a brain medical image template. The probability map of the nucleus area carries information related to the probability value of each area in the corresponding brain medical image template being determined as a nucleus area, and the position of the nucleus area in the brain medical image can be determined based on the probability map of the nucleus area.

[0058] After obtaining the brain medical image, the computer device obtains the brain atlas corresponding to the object to be examined from the memory, and inputs both the brain medical image and the brain atlas of the object to be examined into the pre-trained deformation field determination model to obtain the registration deformation field. The registration deformation field is used to align the brain atlas with the brain medical image. The deformation field determination model is obtained by the computer device training the deep learning neural network according to the brain atlas set and the brain medical image samples. The brain atlas set includes multiple brain maps, and the brain medical image samples can be the brain medical images of the examination object obtained in advance. This embodiment does not limit the specific process of the deformation field determination model, as long as its function can be achieved.

[0059] See also Figure 2In an optional embodiment, the training process of the deformation field determination model is as follows: inputting the brain medical image sample and the brain map corresponding to the brain medical image sample in the brain map set into the deep learning network to obtain a trained registration deformation field; deforming the brain map according to the trained registration deformation field, and comparing the deformed brain map with the brain medical image sample, and adjusting the parameters of the deep learning network according to the comparison result; obtaining a new brain medical image sample, returning to execute the step of inputting the brain medical image sample and the brain map corresponding to the brain medical image sample in the brain map set into the deep learning network to obtain a trained registration deformation field, until the comparison result between the deformed brain map and the brain medical image sample meets the preset condition (the similarity between the brain map and the brain medical image sample is greater than a preset threshold or the difference between the brain map and the brain medical image sample is less than another preset threshold), a trained deep learning network, i.e., the deformation field determination model, can be obtained.

[0060] Step 110: deform the brain atlas according to the registered deformation field to obtain a deformation atlas.

[0061] Step 120: Determine the nucleus region in the brain medical image according to the deformation map.

[0062] After obtaining the registration deformation field, the computer device deforms the brain map corresponding to the object to be inspected according to the registration deformation field to obtain a deformation map. The brain map includes a probability map of the nucleus area and a brain medical image template, and the deformation map includes a probability map of the deformed nucleus area and a deformed brain medical image template. The deformed brain medical image template is mapped to the brain medical image, and the nucleus area in the brain medical image can be determined according to the probability map of the deformed nucleus area corresponding to the deformed brain medical image template. This embodiment does not limit the specific process of deforming the brain map and the specific method of determining the nucleus area, as long as its function can be achieved. There are many types of nuclei in the brain medical image, such as the subthalamic nucleus and the medial globus pallidus. In this embodiment, there is no restriction on the type and number of the nucleus areas finally determined, and the user can choose according to actual needs.

[0063] The method for determining the nucleus region provided in the embodiment of the present application obtains a registered deformation field by inputting the brain medical image of the object to be examined and the brain map matching the object to be examined into the deformation field determination model; the brain map is deformed according to the registered deformation field to obtain a deformation map, and the nucleus region in the brain medical image is determined according to the deformation map. The method for determining the nucleus region provided in the embodiment of the present application obtains a deformation map by deforming the brain map through the registered deformation field determined by the pre-trained deformation field determination model, so that the nucleus region in the brain medical image can be determined according to the deformation map. The nucleus region determined in this way does not need to rely on the doctor's experience, can avoid the influence of the doctor's subjective judgment, and thus can improve the accuracy of the determination of the nucleus region.

[0064] See also Figure 3 In one embodiment, a possible implementation method for obtaining a brain medical image of a subject to be examined by inputting a brain medical image and a brain map of the subject to be examined into a deformation field determination model and obtaining a registration deformation field is proposed, and the steps include:

[0065] Step 300: Acquire an original brain medical image of the subject to be examined.

[0066] The original brain medical image refers to an image obtained by directly scanning the object to be examined. The original brain medical image can be pre-stored in the memory of the computer device, and when the computer device obtains the brain medical image of the object to be examined, it directly obtains the original brain medical image in the memory.

[0067] Step 310: Perform skull separation processing and bias field removal processing on the original brain medical image to obtain a brain medical image.

[0068] After obtaining the original brain medical image, the computer device processes it to obtain the final required brain medical image. Specifically, the computer device performs skull separation processing on the original brain medical image, and can remove the scalp imaging area and skull imaging area in the original brain medical image, so as to avoid interference of structures such as the skull and scalp in the subsequent process of determining the nucleus area, thereby improving the accuracy of determining the nucleus area in the brain medical image. The bias field is an uneven field caused by factors such as the scanner itself and the deviation during the scanning process, resulting in a large brightness difference in the original brain medical image even between the same tissues. The computer device performs bias field removal processing on the original brain medical image, which can ensure that the brightness of the same tissue in the obtained brain medical image is consistent, thereby improving the accuracy of determining the nucleus area in the brain medical image. This embodiment does not limit the specific method of skull separation processing and bias field removal processing on the brain medical image, as long as its function can be achieved.

[0069] See also Figure 4 In one embodiment, a possible implementation method for performing skull separation processing and bias field removal processing on an original brain medical image to obtain a brain medical image is proposed, and the steps include:

[0070] Step 400: Input the original brain medical image into the image preprocessing model to obtain a skull separation mask image and a bias field image; wherein the image preprocessing model is obtained by training a deep learning network based on the original brain medical image samples.

[0071] After obtaining the brain medical image, the computer device inputs it into a pre-trained image preprocessing model, and obtains the skull separation mask image and bias field image corresponding to the original brain medical image through the image preprocessing model. The image preprocessing model is obtained by the computer device training the deep learning network according to the original brain medical image sample. The original brain medical image sample can be the same as the original brain medical image. The description of the original brain medical image sample can refer to the above-mentioned specific description of the original brain medical image, which will not be repeated here. This embodiment does not limit the specific training method of the image preprocessing model, as long as its function can be achieved.

[0072] Step 410: Determine a brain medical image based on the skull separation mask image and the bias field image.

[0073] After the computer device obtains the skull separation mask image and the bias field image according to the image preprocessing model, the computer device can first separate the skull and scalp imaging areas in the original brain medical image according to the skull separation mask image, and then remove the bias field in the brain medical image after the skull separation according to the bias field image to obtain the final brain medical image; the computer device can also first remove the bias field in the original brain medical image according to the bias field image, and then separate the skull in the brain medical image after the bias field is removed according to the skull separation mask image to obtain the final brain medical image. This embodiment does not limit the specific method of determining the brain medical image according to the skull separation mask image and the bias field image, as long as its function can be achieved.

[0074] Specifically, the computer device can separate the skull from the original brain medical image according to preset logical operations and arithmetic operations, and remove the bias field to obtain the brain medical image.

[0075] In an optional embodiment, the computer device performs supervised training on the deep learning network, that is, the computer device stores a standard skull separation mask image, i.e., the first gold standard image, and a standard bias field image, i.e., the second gold standard image. The computer device inputs the original brain medical image sample into the deep learning network to obtain the initial preprocessed skull separation mask image and the bias field image, compares the initial preprocessed skull separation mask image with the first gold standard image, and compares the initial preprocessed bias field image with the second gold standard image, and adjusts the parameters in the deep learning network according to the comparison results; obtains a new original brain medical image sample, returns to execute the step of inputting the original brain medical image sample into the deep learning network, and obtains the initial preprocessed skull separation mask image and the bias field image, until the comparison results of the preprocessed skull separation mask image and the first gold standard image, and the comparison results of the preprocessed bias field image and the second gold standard image meet the preset conditions (the degree of similarity is greater than a preset threshold or the degree of difference is less than another preset threshold), then a trained deep learning network, i.e., an image preprocessing model, can be obtained.

[0076] In this embodiment, the original brain medical image is directly processed using the pre-trained image preprocessing model to obtain the skull separation mask image and the bias field image, which can improve the efficiency of determining the brain medical image, thereby improving the efficiency of determining the nucleus area.

[0077] See also Figure 5 In one embodiment, the image preprocessing model includes an encoder, a first decoder, and a second decoder; the encoder includes two output terminals, one output terminal is connected to the first decoder, and the other output terminal is connected to the second decoder. A possible implementation method of inputting the original brain medical image into the image preprocessing model to obtain the skull separation mask image and the bias field image includes the following steps:

[0078] Step 500: Input the original brain medical image into the encoder to obtain L-layer features, where the L-layer features are used to characterize the features of the original brain medical image at L resolutions, where L is an integer greater than or equal to zero;

[0079] After the computer device inputs the original brain medical image into the encoder in the image preprocessing model, the encoder can extract L layers of features of the original brain medical image at L resolutions, one resolution corresponds to one layer of features, and L is an integer greater than or equal to zero. In an optional embodiment, when L is zero, the encoder extracts the features of the original brain medical image at full resolution, and the larger the value of L, the smaller the resolution corresponding to the L layer features extracted by the encoder. The number of resolutions can be set by the user, and this embodiment does not limit this.

[0080] Step 510: Input the L-layer features into the first decoder to obtain a skull separation mask image.

[0081] After obtaining the L-layer features, the computer device inputs them into the first decoder, and the L-layer features are processed by the first decoder to obtain a skull separation mask image.

[0082] In an optional embodiment, the first decoder uses a skip connection to retain more information, and the loss function corresponding to the first decoder uses a dice loss function.

[0083] Step 520: Input the L-layer features into the second decoder to obtain a bias field image.

[0084] After obtaining the L-layer features, the computer device inputs them into the second decoder. The second decoder processes the L-layer features to obtain a bias field image.

[0085] In an optional embodiment, the second decoder can predict multiple bias fields by processing each layer of features in the L layers of features. The greater the resolution of the feature, the finer the bias field image predicted by the feature. The second decoder also uses skip connections to retain more information.

[0086] In an optional embodiment, the image preprocessing model is trained based on the original brain medical image samples. The specific training process of the image preprocessing model is as follows: Figure 6 As shown, it is assumed that after the original brain medical image sample passes through the encoder, three layers of features are obtained, that is, L=3, and these three layers of features are respectively called L0 layer features, L1 layer features and L2 layer features. Among them, when L0=0, the L0 layer features represent the features of the original brain medical image samples at full resolution, when L1=1, the L1 layer features represent the features of the original brain medical image samples at the first resolution, and when L2=2, the L2 layer features represent the features of the original brain medical image at the second resolution. Among them, the resolutions of the full resolution, the first resolution and the second resolution decrease in sequence. The L0 layer features, the L1 layer features and the L2 layer features can be processed by the first decoder to obtain a predicted skull separation mask image, and the image is compared with the first gold standard image, and the parameters of the image preprocessing model are adjusted until the comparison result between the image and the first gold standard image meets the preset conditions. The second decoder can process the L0 layer features, L1 layer features and L2 layer features to obtain a predicted bias field image, compare the image with the second gold standard image, and adjust the parameters of the image preprocessing model until the comparison result between the image and the second gold standard image meets the preset conditions. Specifically, the predicted bias field image obtained by the second decoder from processing the L0 layer features is more refined and accurate.

[0087] The loss functions used by the second decoder include histogram loss, regularization term loss and mean square error loss.

[0088] The histogram loss function is used to calculate the difference in the voxel grayscale value histogram between the brain medical image after removing the predicted bias field image obtained by the second decoder and the third gold standard image, where the third gold standard image refers to the brain medical image after removing the standard bias field image. The calculation formula of the histogram loss function is: L HC =(1-d(H g ,H p )), where L HC represents the histogram loss function, d(Hg, Hp) is expressed by the formula Calculate, where Hg represents the histogram distribution of the voxel grayscale value of the second gold standard image, and Hp represents the histogram distribution of the voxel grayscale value of the brain medical image after removing the predicted bias field image. Hg(M) represents the number of voxels in the second gold standard image whose voxel grayscale value falls within the grayscale level M, It represents the average number of voxels in the second gold standard image when each voxel falls evenly within each small interval of the histogram. Hp(M) represents the number of voxels whose voxel grayscale values ​​of the brain medical image fall within the grayscale level M after removing the predicted bias field image. It represents the average number of voxels in the brain medical image after removing the predicted bias field image, when each voxel falls evenly within each small interval of the histogram.

[0089] The regularization loss function is mainly used to ensure the continuity and smoothness of the bias field. The specific calculation formula is: in, represents the bias field, represents the regularization loss function of the bias field, p represents the bias field voxel, Represents the gradient of the predicted bias field. The gradient of the bias field calculated according to the above formula can control the smoothness of the bias field.

[0090] The calculation formula of the mean square error loss function is: Among them, MSE(I 1 ,I 2 ) indicates I 1 and I 2 The mean square error loss function between the second gold standard image I 1 and the bias field image I output by the second decoder 2 The mean square error function between g (x, y) represents the pixel value of the second gold standard image at (x, y), I m (x, y) represents the pixel value of the bias field image at (x, y) output by the second decoder, and N represents the number of pixels of the bias field image.

[0091] In this embodiment, since the image preprocessing model needs to perform two tasks, namely skull separation and bias field removal, a dual-branch structure is adopted, i.e., a first decoder and a second decoder are adopted. Since some parameters used by the first decoder and the second decoder in processing L-layer features can be shared, parameters can be reduced, the amount of calculation can be reduced, and complementary information can be provided, so that the final brain medical image is more accurate.

[0092] See also Figure 7 In one embodiment, it is proposed that the brain medical image and brain atlas of the subject to be examined are input into the deformation field determination model, and before obtaining the registration deformation field, the steps of the nucleus region determination method include:

[0093] Step 700: Obtain a brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups.

[0094] Step 710: Determine a brain map corresponding to the subject to be examined in the brain map set according to the age of the subject to be examined.

[0095] The brain map collection includes brain maps corresponding to the subjects of different age groups. The computer device obtains the age of the subject to be examined, determines the age group of the subject, and determines the brain map corresponding to the subject of the age group as the brain map corresponding to the subject to be examined.

[0096] In this embodiment, the brain map corresponding to the object to be examined is determined in the brain map set according to the age of the object to be examined. This can avoid the problem of inaccurate nucleus areas finally determined due to different brain structures of objects of different age groups. In other words, this can improve the accuracy of the final determination of the nucleus areas.

[0097] See also Figure 8 In one embodiment, a possible implementation method for obtaining a brain map set is proposed, and the steps include:

[0098] Step 800: Acquire multiple brain medical image sample sets, wherein each brain medical image sample set includes multiple brain medical image samples corresponding to examination subjects in the same age group, and each brain medical image sample includes a marked nucleus area.

[0099] The computer device can obtain multiple brain medical image sample sets based on multiple brain medical image samples corresponding to the inspection objects of each age group. A brain medical image sample set includes multiple brain medical image samples corresponding to the inspection objects of the same age group, and each brain medical image sample includes a marked nucleus area, that is, the position of the nucleus area in the brain medical image sample is known. One inspection can correspond to one brain medical image sample or multiple brain medical image samples, and there can be multiple inspection objects of the same age group. The computer device can eventually obtain multiple brain medical image sample sets corresponding to inspection objects of different age groups. The brain medical image samples can be stored in the memory of the computer device, and the computer device can directly obtain them from the memory when needed. For the description of the brain medical image samples, reference can be made to the specific description of the above-mentioned brain medical images, which will not be repeated here.

[0100] Step 810: for each brain medical image sample set, determine a brain atlas corresponding to the brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set;

[0101] Step 820: taking the multiple brain atlases corresponding to the multiple brain medical image sample sets as a brain atlas set.

[0102] After obtaining multiple brain medical image sample sets, the computer device can determine the brain map corresponding to each brain medical image sample set according to multiple brain medical image samples in the brain medical image sample set, and finally obtain the brain map corresponding to each brain medical image sample set, that is, the brain map set. This embodiment does not limit the specific method of determining the brain map corresponding to each brain medical image sample set, as long as its function can be achieved.

[0103] In one embodiment, a possible implementation method of determining a brain atlas corresponding to a brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set is proposed, such as Fig. 9 The steps shown include:

[0104] Step 900: Based on an affine transformation method, each brain medical image sample in the brain medical image sample set is registered to a standard space.

[0105] After obtaining multiple brain medical image sample sets, the computer device aligns multiple brain medical image samples in each brain medical image sample set to a standard space, such as the Montreal Neurological Institute (MNI) space, according to an affine transformation method. In other words, the pixel values ​​of the multiple brain medical image samples are mapped to the standard space. This embodiment does not limit the specific alignment method, as long as its function can be achieved.

[0106] Step 910: Determine a probability map of the nucleus region according to the distribution of the nucleus region in each brain medical image sample in the standard space.

[0107] After the computer equipment registers multiple brain medical image samples in each brain medical image sample set into the standard space, it can determine the probability map of the nucleus area corresponding to each brain medical image sample set by calculating the distribution number of nucleus areas in each brain medical image sample in the standard space, that is, the probability map of the nucleus area corresponding to the examination subjects of each age group.

[0108] Step 920: Determine a brain atlas corresponding to the brain medical image sample set according to the acquired brain medical image template and probability map of the nucleus region.

[0109] The computer device can determine the brain atlas corresponding to the brain medical image sample set, that is, the brain atlas of the examination subject in the age group corresponding to the brain medical image sample set, based on the brain medical image template and the probability map of the nucleus area corresponding to each acquired brain medical image sample set.

[0110] In this embodiment, brain maps of different age groups are determined for the examination subjects of different age groups, so as to facilitate the subsequent determination of more accurate nucleus areas for the brain medical image samples of the examination subjects of different age groups.

[0111] In one embodiment, the method for determining the nucleus region further comprises:

[0112] The nucleus region is corrected using the region growing method or the boundary correction method.

[0113] The computer device can use any one of the region growing method and the boundary correction method to realize the correction of the initial nucleus region. Among them, the region growing correction method is to quickly fill and remove the deviation area in the nucleus region by responding to the user's click operation. Among them, the deviation area in the nucleus region refers to the area that is obviously different from the accurate nucleus region. The boundary correction method is to represent the outline of the nucleus region in the form of boundary points, and correct the outline of the nucleus region in response to the user's operation of dragging the boundary points to obtain a more accurate nucleus region after correction. When performing the boundary correction method, the boundary points near the dragging point can be adaptively adjusted, and the user does not need to check all the boundary points of the nucleus region, which can not only reduce the user's workload, but also improve the efficiency of determining the correction.

[0114] In an optional embodiment, after determining the nucleus region, the computer device can also plan the electrode implantation path according to the nucleus region. Specifically, the center position of the nucleus region is used as the initial target position, and the voxel points of the skin area in the brain medical image are used as the set of needle entry points for electrode implantation. According to the vascular area and brain tissue area in the brain medical image, the safe distance from each needle entry point to the implantation path formed by the initial target point to the brain tissue and vascular area is calculated. The optimal implantation path is determined by comprehensively considering maximizing the safe distance, minimizing the implantation path length, and maximizing the nucleus area that can be covered by the electrode contacts.

[0115] In another optional implementation, the initial target point and implantation path can be adjusted according to the actual application. Specifically, the initial target point can be extended 5 mm inward, with the needle entry point as the vertex, and the space within the cone range within an angle of 3° with the implantation path at this time is used as the adjustment space of the implantation path. The minimum safety distance in the adjustment space is calculated, and the minimum safety distance is marked to serve as a prompt. The user judges the planned implantation path. If the judgment result is not satisfactory, the initial target point and the needle entry point position can be readjusted, and then a new minimum safety distance mark can be reset.

[0116] In an optional embodiment, in actual applications, a computed tomography (CT) image of the object to be examined is acquired before surgery, and the CT image is registered with a brain medical image of a determined nucleus region, and then the pixels of the image are changed by a weighted average method of pixels to achieve fusion of the CT image and the brain medical image for navigation of subsequent surgery.

[0117] In one embodiment, a whole process of nucleus region determination and application is provided as follows: Fig.10 As shown, the steps include:

[0118] Step 101, obtaining an original brain medical image of the subject to be examined, specifically, performing a magnetic resonance scan on the head of the subject to be examined before surgery, scanning a T1 sequence and / or a T2 sequence, and obtaining a magnetic resonance image of the subject to be examined, which is referred to as an original brain medical image;

[0119] Step 102, preprocessing the original brain medical image, specifically inputting the original brain medical image into an image preprocessing model to obtain a skull separation mask image and a bias field image, and determining the brain medical image according to the skull separation mask image and the bias field image;

[0120] Step 103, selecting a brain map corresponding to the subject to be examined, specifically determining a brain map corresponding to the subject to be examined in a brain map set according to the age of the subject to be examined;

[0121] Step 104: using a registration method based on deep learning, the nucleus region in the brain medical image is determined according to the brain medical image and the brain map of the subject to be examined, specifically, the brain medical image and the brain map of the subject to be examined are input into a deformation field determination model to obtain a registration deformation field, the brain map is deformed according to the registration deformation field to obtain a deformation map, and the nucleus region in the brain medical image is determined according to the deformation map;

[0122] Step 105, correcting the nucleus region using a region growing method or a boundary correction method;

[0123] Step 106, planning the electrode implantation path according to the nucleus area;

[0124] Step 107, obtaining a three-dimensional brain medical image with marked points of the object to be examined, specifically, performing a computer tomography scan or a magnetic resonance scan on the object to be examined in the morning of the operation, obtaining a computer tomography image or a magnetic resonance image with bone marked points, and using it as a three-dimensional brain medical image with marked points;

[0125] Step 108, fusing the three-dimensional brain medical image with the marked points with the brain medical image with the nucleus area determined, specifically registering the two, and then changing the pixels of the image by a weighted average method of pixels to achieve image fusion, thereby determining the surgical path and navigating for subsequent surgery;

[0126] Step 109: Save the determined surgical path as a surgical plan.

[0127] It should be understood that, although the various steps in the flowchart in the figure are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the figure may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0128] See also Fig.11 An embodiment of the present application provides a nucleus region determination device 10, which includes a registration deformation field determination module 11, a deformation spectrum determination module 12 and a nucleus region determination module 13. Among them,

[0129] The registration deformation field determination module 11 is used to input the brain medical image and brain atlas of the object to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas and the brain medical image samples;

[0130] The deformation atlas determination module 12 is used to perform deformation processing on the brain atlas according to the registration deformation field to obtain a deformation atlas;

[0131] The nucleus region determination module 13 is used to determine the nucleus region in the brain atlas medical image according to the deformation atlas.

[0132] In one embodiment, the nucleus region determination device 10 also includes a first acquisition unit and a processing unit. The first acquisition unit is used to acquire the original brain medical image of the object to be examined; the processing unit is used to perform skull separation processing and bias field removal processing on the original brain medical image to obtain a brain medical image.

[0133] In one embodiment, the processing unit is specifically used to input the original brain medical image into an image preprocessing model to obtain a skull separation mask image and a bias field image; wherein the image preprocessing model is obtained by training a deep learning network based on the original brain medical image samples; and the brain medical image is determined based on the skull separation mask image and the bias field image.

[0134] In one embodiment, the processing unit is specifically used to input the original brain medical image into an encoder to obtain L-layer features, where the L-layer features are used to characterize the features of the original brain medical image at L resolutions, where L is an integer greater than or equal to zero; input the L-layer features into a first decoder to obtain a skull separation mask image; and input the L-layer features into a second decoder to obtain a bias field image.

[0135] In one embodiment, the nucleus region determination device 10 also includes a second acquisition unit, which is used to acquire a brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups; according to the age of the examination subject, the brain map corresponding to the examination subject is determined in the brain map set.

[0136] In one embodiment, the nucleus region determination device 10 also includes a determination unit, which is used to obtain multiple brain medical image sample sets; wherein each brain medical image sample set contains multiple brain medical image samples corresponding to the examination subjects in the same age group, and each brain medical image sample includes a labeled nucleus region; for each brain medical image sample set, a brain map corresponding to the brain medical image sample set is determined based on the multiple brain medical image samples in the brain medical image sample set; and the multiple brain maps corresponding to the multiple brain medical image sample sets are used as a brain map set.

[0137] In one embodiment, the determination unit is specifically used to align each brain medical image sample in the brain medical image sample set to the standard space based on the affine transformation method; determine the probability map of the nucleus area according to the distribution of the nucleus area in each brain medical image sample in the standard space; determine the brain atlas corresponding to the brain medical image sample set according to the acquired brain medical image template and the probability map of the nucleus area, and obtain the brain atlas set.

[0138] In one embodiment, the nucleus region determining device 10 further includes a correction unit, which is used to correct the nucleus region by using a region growing method or a boundary correction method.

[0139] For the specific definition of the above-mentioned nucleus region determination device 10, please refer to the definition of the nucleus region determination method above, which will not be repeated here. Each module in the nucleus region determination device 10 can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned devices, modules or units can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned devices or modules.

[0140] See also Fig.12In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.12 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store a brain map collection, a deformation field determination model, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer device is executed by the processor, a method for determining a nucleus region is implemented.

[0141] Those skilled in the art will understand that Fig.12 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0142] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0143] Inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas set and the brain medical image samples;

[0144] Deform the brain atlas according to the registered deformation field to obtain a deformation atlas;

[0145] Based on the deformation atlas, the nucleus area in the brain medical image is determined.

[0146] In one embodiment, when the processor executes the computer program, the following steps are also implemented: acquiring an original brain medical image of the object to be examined; performing skull separation processing and bias field removal processing on the original brain medical image to obtain a brain medical image.

[0147] In one embodiment, when the processor executes the computer program, the following steps are also implemented: the original brain medical image is input into the image preprocessing model to obtain a skull separation mask image and a bias field image; the image preprocessing model is obtained by training a deep learning network based on the original brain medical image samples; and the brain medical image is determined based on the skull separation mask image and the bias field image.

[0148] In one embodiment, when the processor executes the computer program, the following steps are also implemented: inputting the original brain medical image into the encoder to obtain L-layer features, where the L-layer features are used to characterize the features of the original brain medical image at L resolutions, where L is an integer greater than or equal to zero; inputting the L-layer features into the first decoder to obtain a skull separation mask image; and inputting the L-layer features into the second decoder to obtain a bias field image.

[0149] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining a brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups; and determining the brain map corresponding to the examination subject in the brain map set according to the age of the examination subject.

[0150] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining multiple brain medical image sample sets, wherein each brain medical image sample set contains multiple brain medical image samples corresponding to the examination subjects in the same age group, and each brain medical image sample includes a labeled nucleus area; for each brain medical image sample set, determining a brain map corresponding to the brain medical image sample set based on the multiple brain medical image samples in the brain medical image sample set; and using the multiple brain maps corresponding to the multiple brain medical image sample sets as a brain map set.

[0151] In one embodiment, when the processor executes the computer program, the following steps are also implemented: based on the affine transformation method, each brain medical image sample in the brain medical image sample set is registered to the standard space; according to the distribution of the nucleus area in each brain medical image sample in the standard space, the probability map of the nucleus area is determined; according to the acquired brain medical image template and the probability map of the nucleus area, the brain atlas corresponding to the brain medical image sample set is determined to obtain the brain atlas set.

[0152] In one embodiment, when the processor executes the computer program, the following steps are further implemented: correcting the nucleus region using a region growing method or a boundary correction method.

[0153] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0154] Inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas set and the brain medical image samples;

[0155] Deform the brain atlas according to the registered deformation field to obtain a deformation atlas;

[0156] Based on the deformation atlas, the nucleus area in the brain medical image is determined.

[0157] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: acquiring an original brain medical image of the object to be examined; performing skull separation processing and bias field removal processing on the original brain medical image to obtain a brain medical image.

[0158] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: the original brain medical image is input into the image preprocessing model to obtain a skull separation mask image and a bias field image; the image preprocessing model is obtained by training a deep learning network based on the original brain medical image samples; and the brain medical image is determined based on the skull separation mask image and the bias field image.

[0159] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: inputting the original brain medical image into the encoder to obtain L-layer features, where the L-layer features are used to characterize the features of the original brain medical image at L resolutions, where L is an integer greater than or equal to zero; inputting the L-layer features into the first decoder to obtain a skull separation mask image; and inputting the L-layer features into the second decoder to obtain a bias field image.

[0160] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining a brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups; and determining the brain map corresponding to the examination subject in the brain map set according to the age of the examination subject.

[0161] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining multiple brain medical image sample sets, wherein each brain medical image sample set contains multiple brain medical image samples corresponding to the examination subjects in the same age group, and each brain medical image sample includes a labeled nucleus area; for each brain medical image sample set, determining a brain map corresponding to the brain medical image sample set based on the multiple brain medical image samples in the brain medical image sample set; and using the multiple brain maps corresponding to the multiple brain medical image sample sets as a brain map set.

[0162] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: based on the affine transformation method, each brain medical image sample in the brain medical image sample set is registered to the standard space; according to the distribution of the nucleus area in each brain medical image sample in the standard space, the probability map of the nucleus area is determined; according to the acquired brain medical image template and the probability map of the nucleus area, the brain atlas corresponding to the brain medical image sample set is determined to obtain the brain atlas set.

[0163] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: correcting the nucleus region using a region growing method or a boundary correction method.

[0164] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0165] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0166] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for determining a nucleus region, characterized in that: include: Inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field; wherein the brain atlas includes the probability map of the nucleus area and the brain medical image template, and the deformation field determination model is obtained by training the deep learning network according to the brain atlas set and the brain medical image samples; Deforming the brain atlas according to the registered deformation field to obtain a deformation atlas; the deformation atlas includes a probability map of the deformed nucleus area and a deformed brain medical image template; Determining a nucleus region in the brain medical image according to the deformation atlas; Wherein, determining the nucleus region in the brain medical image according to the deformation atlas includes: The deformed brain medical image template is mapped onto the brain medical image, and the nucleus region in the brain medical image is determined according to a probability map of the deformed nucleus region corresponding to the deformed brain medical image template.

2. The method for determining a nucleus region according to claim 1, characterized in that: Before inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field, the method further includes: Acquiring an original brain medical image of the subject to be examined; Inputting the original brain medical image into an image preprocessing model to obtain a skull separation mask image and a bias field image; The brain medical image is determined according to the skull separation mask image and the bias field image.

3. The method for determining a nucleus region according to claim 2, characterized in that: The image preprocessing model includes an encoder, a first decoder and a second decoder; the step of inputting the original brain medical image into the image preprocessing model to obtain a skull separation mask image and a bias field image includes: Inputting the original brain medical image into the encoder to obtain L-layer features, where L is an integer greater than or equal to zero; Inputting the L-layer features into the first decoder to obtain the skull separation mask image; The L-layer features are input into the second decoder to obtain the bias field image.

4. The method for determining a nucleus region according to claim 1, characterized in that: Before inputting the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain the registration deformation field, the method further includes: Acquire the brain map set; wherein the brain map set includes brain maps corresponding to examination subjects of different age groups; According to the age of the object to be examined, a brain map corresponding to the object to be examined is determined in the brain map set.

5. The method for determining a nucleus region according to claim 4, characterized in that: The step of obtaining the brain map set includes: Acquire multiple brain medical image sample sets; wherein each of the brain medical image sample sets includes multiple brain medical image samples corresponding to examination subjects of the same age group, and each of the brain medical image samples includes a marked nucleus region; For each of the brain medical image sample sets, determining a brain atlas corresponding to the brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set; The multiple brain atlases corresponding to the multiple brain medical image sample sets are used as the brain atlas set.

6. The method for determining a nucleus region according to claim 5, characterized in that: The step of determining a brain atlas corresponding to the brain medical image sample set according to a plurality of brain medical image samples in the brain medical image sample set comprises: Based on an affine transformation method, each of the brain medical image samples in the brain medical image sample set is registered to a standard space; Determining a probability map of the nucleus region according to the distribution of the nucleus region in each of the brain medical image samples in the standard space; A brain atlas corresponding to the brain medical image sample set is determined according to the acquired brain medical image template and the probability map of the nucleus area.

7. The method for determining a nucleus region according to claim 1, characterized in that: The method further comprises: The nucleus region is corrected by adopting a region growing method or a boundary correction method.

8. A device for determining a nucleus region, characterized in that: include: A registration deformation field determination module is used to input the brain medical image and brain atlas of the subject to be examined into the deformation field determination model to obtain a registration deformation field; wherein the brain atlas includes a probability map of the nucleus area and a brain medical image template, and the deformation field determination model is obtained by training a deep learning network based on the brain atlas and brain medical image samples; A deformation atlas determination module, used for deforming the brain atlas according to the registered deformation field to obtain a deformation atlas; the deformation atlas includes a probability map of the deformed nucleus area and a deformed brain medical image template; A nucleus region determination module, used to determine the nucleus region in the brain medical image according to the deformation atlas; The nucleus region determination module is specifically used to map the deformed brain medical image template onto the brain medical image, and determine the nucleus region in the brain medical image according to the probability map of the deformed nucleus region corresponding to the deformed brain medical image template.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images