A method, device and equipment for positioning a target region in a brain CT image

Abstract

This application discloses a method, apparatus, and device for locating a target region in a brain CT image. The method involves acquiring a standard brain CT image and its associated brain anatomical structure partitioning labels. The brain CT image is then registered with the standard brain CT image to generate a target deformation field. This target deformation field is applied to the brain anatomical structure partitioning labels of the standard brain CT image to obtain the brain anatomical structure partitioning labels for the brain CT image. Then, by determining the segmentation threshold corresponding to each slice image, the target hemorrhage region is identified from the brain CT image. Matching the target hemorrhage region with the brain anatomical structure partitioning labels of the brain CT image yields the brain anatomical structure partition where the target hemorrhage region is located, thereby locating the site of the brain hemorrhage.

Description

Technical Field

[0001] This application relates to the field of image processing, specifically to a method, apparatus, and device for locating a target region in a brain CT image. Background Technology

[0002] Cerebral hemorrhage is intracranial bleeding caused by the rupture of blood vessels. It is characterized by rapid onset, critical condition, severe sequelae, high mortality and disability rates, and multiple complications. Computed tomography (CT) scans are typically used to perform continuous scans of the patient's brain at different levels, thereby obtaining corresponding brain CT images.

[0003] Currently, methods exist to automatically identify the type of brain hemorrhage using brain CT images, aiding doctors in diagnosis. Types of brain hemorrhage include parenchymal hemorrhage, subarachnoid hemorrhage, intraventricular hemorrhage, subdural hemorrhage, and epidural hemorrhage. However, it is not yet possible to automatically pinpoint the exact location of a brain hemorrhage using brain CT images. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method, apparatus, and device for locating a target region in a brain CT image, so as to automatically locate the specific location of a brain hemorrhage through brain CT images.

[0005] To address the above problems, the technical solutions provided in this application are as follows:

[0006] A method for locating a target region in a brain CT image, the method comprising:

[0007] Obtain standard brain CT images and brain anatomical structure partition labels of the standard brain CT images;

[0008] The brain CT image and the standard brain CT image are input into the registration network to obtain the target deformation field, which identifies the deformation pattern between the brain CT image and the standard brain CT image.

[0009] The target deformation field is applied to the brain anatomy section label of the standard brain CT image to obtain the brain anatomy section label of the brain CT image.

[0010] Identify brain parenchymal regions from the brain CT images;

[0011] Based on the histogram of CT values ​​of the brain parenchyma region in each slice of the brain CT image, the segmentation threshold corresponding to each slice of the brain CT image is determined;

[0012] Identify regions larger than the stated segmentation threshold from each slice of a brain CT image as target hemorrhage areas;

[0013] The target hemorrhage area is matched with the brain anatomical structure partition labels of the brain CT image to obtain the brain anatomical structure partition where the target hemorrhage area is located.

[0014] In one possible implementation, determining the segmentation threshold for each slice of the brain CT image based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image includes:

[0015] The target CT value is obtained by taking the CT value with the most pixels from the CT value histogram of the brain parenchyma region in the target slice image; the target slice image is each slice image of the brain CT image.

[0016] The target CT value is multiplied by a preset weight to obtain the segmentation threshold corresponding to the target slice image.

[0017] In one possible implementation, after identifying regions larger than the segmentation threshold as target hemorrhage regions from each slice of the brain CT image, and before matching the target hemorrhage regions with brain anatomical structure partition labels of the brain CT image, the method further includes:

[0018] An opening operation of a preset scale is performed on the target bleeding area to generate a processed target bleeding area;

[0019] The target value is obtained by subtracting the number of three-dimensional connected components of the processed target bleeding region from the number of three-dimensional connected components of the target bleeding region, and then dividing by the number of three-dimensional connected components of the target bleeding region.

[0020] If the target value is greater than the target threshold, the target bleeding area is retained;

[0021] If the target value is less than or equal to the target threshold, the segmentation threshold is increased, and the process of identifying regions with values ​​greater than the segmentation threshold from each slice of the brain CT image as target hemorrhage regions is repeated. An opening operation of a preset scale is performed on the target hemorrhage regions to generate the processed target hemorrhage regions and subsequent steps.

[0022] In one possible implementation, increasing the segmentation threshold includes:

[0023] Divide the segmentation threshold by a preset value and add the segmentation threshold back to obtain the new segmentation threshold.

[0024] In one possible implementation, identifying brain parenchyma regions from the brain CT image includes:

[0025] Identify the extracerebral space region, falx cerebri region, and ventricle region from the brain CT images;

[0026] The brain parenchyma region is obtained by removing the extracerebral space region, the falx cerebri region, and the ventricle region from the brain CT image.

[0027] In one possible implementation, the method further includes, before matching the target hemorrhage region with brain anatomical region labels of the brain CT image:

[0028] The side containing the target bleeding area is defined as the side to be corrected, and the opposite side of the target bleeding area is defined as the reference side.

[0029] The brain anatomical structure partition labels of the brain CT image are symmetrically mapped along the center line to the side to be corrected, replacing the partition labels on the side to be corrected, in order to regenerate the brain anatomical structure partition labels of the brain CT image.

[0030] In one possible implementation, the method further includes:

[0031] The volume of the target bleed region is obtained based on the number of pixels in the target bleed region and the volume of a unit pixel.

[0032] A device for locating a target region in a brain CT image, the device comprising:

[0033] The acquisition unit is used to acquire a standard brain CT image and the brain anatomical structure partition labels of the standard brain CT image;

[0034] The registration unit is used to input the brain CT image and the standard brain CT image into the registration network to obtain the target deformation field, which identifies the deformation pattern between the brain CT image and the standard brain CT image.

[0035] A deformation unit is used to apply the target deformation field to the brain anatomy partition label of the standard brain CT image to obtain the brain anatomy partition label of the brain CT image.

[0036] The first identification unit is used to identify brain parenchyma regions from the brain CT image;

[0037] The first determining unit is used to determine the segmentation threshold corresponding to each slice of the brain CT image based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image.

[0038] The second identification unit is used to identify regions larger than the segmentation threshold as target hemorrhage areas from each slice of the brain CT image;

[0039] A matching unit is used to match the target hemorrhage area with the brain anatomical structure partition labels of the brain CT image to obtain the brain anatomical structure partition label where the target hemorrhage area is located.

[0040] A device for locating a target region in a brain CT image includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method for locating a target region in a brain CT image as described above.

[0041] A computer-readable storage medium storing instructions that, when executed on a terminal device, cause the terminal device to perform the method for locating a target region in a brain CT image as described above.

[0042] Therefore, the embodiments of this application have the following beneficial effects:

[0043] This application embodiment acquires a standard brain CT image and its corresponding brain anatomical structure partition labels. The brain CT image is then registered with the standard brain CT image to generate a target deformation field. This target deformation field is applied to the brain anatomical structure partition labels of the standard brain CT image, thereby obtaining the brain anatomical structure partition labels for the brain CT image. Then, by determining the segmentation threshold corresponding to each slice image, the target hemorrhage region is identified from the brain CT image. Matching the target hemorrhage region with the brain anatomical structure partition labels of the brain CT image allows for the determination of the brain anatomical structure partition where the target hemorrhage region is located, thus achieving the localization of the brain hemorrhage site. Attached Figure Description

[0044] Figure 1 A schematic diagram illustrating an exemplary application scenario provided in this application embodiment;

[0045] Figure 2 A flowchart illustrating a method for locating a target region in a brain CT image, provided as an embodiment of this application;

[0046] Figure 3 This is a schematic diagram of the brain anatomy section labels in a standard brain CT image as described in this application embodiment;

[0047] Figure 4 This is a schematic diagram of the identification results of the extracerebral space region in an embodiment of this application;

[0048] Figure 5 This is a schematic diagram of the recognition results of the falx cerebri region in an embodiment of this application;

[0049] Figure 6 This is a schematic diagram of the recognition results of brain parenchyma regions in the embodiments of this application;

[0050] Figure 7 This is a schematic diagram of the CT value histogram of the brain parenchyma region of the target slice image in the embodiments of this application;

[0051] Figure 8 This is a schematic diagram illustrating the identification results of the target bleeding area in an embodiment of this application;

[0052] Figure 9 This is a schematic diagram illustrating the correction of brain anatomy section labels in a standard brain CT image in an embodiment of this application;

[0053] Figure 10 This is a schematic diagram of a target region localization device in a brain CT image provided in an embodiment of this application. Detailed Implementation

[0054] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the embodiments of this application will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0055] To facilitate understanding and explanation of the technical solutions provided in the embodiments of this application, the background technology of the embodiments of this application will be described first below.

[0056] Computed tomography (CT) scans are typically used to perform continuous scans of the patient's brain at different levels, resulting in corresponding brain CT images. Currently, methods exist to automatically identify the type of brain hemorrhage using brain CT images, including parenchymal hemorrhage, subarachnoid hemorrhage, intraventricular hemorrhage, subdural hemorrhage, and epidural hemorrhage. For parenchymal hemorrhage, the location can include deep hemisphere hemorrhage (the deep hemispheres include the basal ganglia, thalamus, internal capsule, and corpus callosum) and lobar hemorrhage (the lobes include the frontal, temporal, parietal, and occipital lobes, as well as multiple lobes). However, it is currently not possible to automatically pinpoint the exact location of a brain hemorrhage using brain CT images.

[0057] Based on this, embodiments of this application provide a method for locating a target region in a brain CT image. This method involves acquiring a standard brain CT image and its corresponding brain anatomical structure partition labels. The brain CT image is then registered with the standard brain CT image to generate a target deformation field. This target deformation field is applied to the brain anatomical structure partition labels of the standard brain CT image to obtain the brain anatomical structure partition labels for the brain CT image. Then, by determining the segmentation threshold corresponding to each slice image, the target hemorrhage region is identified from the brain CT image. Matching the target hemorrhage region with the brain anatomical structure partition labels of the brain CT image yields the brain anatomical structure partition where the target hemorrhage region is located, thereby locating the site of the brain hemorrhage.

[0058] To facilitate understanding of the method for locating target regions in brain CT images provided in the embodiments of this application, the following is combined with... Figure 1 The example scenario is shown below. See also... Figure 1 As shown in the figure, this figure is a schematic diagram of an exemplary application scenario provided in the embodiments of this application.

[0059] First, standard brain CT images and manually labeled brain anatomical structures are acquired. The brain CT image to be identified is registered with the standard brain CT image to generate a target deformation field characterizing the deformation pattern between them. This target deformation field is then applied to the brain anatomical structure labels of the standard brain CT image to obtain the brain anatomical structure labels for the brain CT image. Next, brain parenchymal regions are identified from the brain CT image, and a segmentation threshold is determined for each slice image based on the CT value histogram of the brain parenchymal region. Since the CT value of the brain hemorrhage region is high, regions with values ​​greater than the segmentation threshold can be identified as target hemorrhage regions from each slice image. Matching the target hemorrhage region with the brain anatomical structure labels of the brain CT image yields the brain anatomical structure region where the target hemorrhage region is located. This achieves batch, rapid, automated, and accurate identification of brain hemorrhage locations in brain CT images.

[0060] Those skilled in the art will understand that Figure 1 The schematic diagram shown is merely one example in which embodiments of this application can be implemented. The scope of application of the embodiments of this application is not limited by any aspect of this framework.

[0061] To facilitate understanding of this application, the following description, in conjunction with the accompanying drawings, illustrates a method for locating a target region in a brain CT image provided by an embodiment of this application.

[0062] See Figure 2 As shown, this figure is a flowchart of a method for locating a target region in a brain CT image according to an embodiment of this application. Figure 2 As shown, the method may include S201-S207:

[0063] S201: Obtain standard brain CT images and brain anatomical structure partition labels of standard brain CT images.

[0064] Obtain standard brain CT images and manually annotated labels for the brain anatomical structures on those images. In practice, a set of standard brain CT images (e.g., images without missing structures, with a symmetrical, centered head shape, and no lesions) can be selected and labeled layer by layer to obtain the brain anatomical structure labels for the standard brain CT images. Brain anatomical structures include, for example, the basal ganglia, thalamus, internal capsule, corpus callosum, frontal lobe, temporal lobe, parietal lobe, occipital lobe, and multiple lobes. The labeling results need to be quality-controlled by a professional radiologist. See also... Figure 3 The diagram illustrates a schematic of brain anatomical structure partitioning labels on a standard brain CT image. In the brain anatomical structure partitioning labels for a standard brain CT image, each slice of the standard brain CT image is labeled with a brain anatomical structure partition. Different brain anatomical structure partitions can be labeled with different colors. This application does not limit the method of labeling brain anatomical structure partitions.

[0065] S202: Input the brain CT image and the standard brain CT image into the registration network to obtain the target deformation field. The target deformation field identifies the deformation pattern between the brain CT image and the standard brain CT image.

[0066] This application describes the acquisition of brain CT images to be identified. The embodiments of this application are applicable to various brain CT images obtained after tomographic scanning of the brain under CT plain scan or CT perfusion imaging. Brain CT images generally include multi-slice images.

[0067] Because brain structures vary between individuals, there will be differences between brain CT images and standard brain CT images. To locate the site of brain parenchymal hemorrhage using brain CT images, it is necessary to register the brain CT images with standard brain CT images. Specifically, the brain CT images and standard brain CT images can be input into a registration network to obtain a target deformation field. The target deformation field identifies the deformation patterns between the brain CT images and standard brain CT images; specifically, it identifies the movement of pixels between the two images.

[0068] The registration network can be a two-level cascaded VTN (Volume Tweening Network) network, or other networks that achieve image registration. The embodiments of this application do not limit the structure of the registration network.

[0069] S203: Apply the target deformation field to the brain anatomy section labels of the standard brain CT image to obtain the brain anatomy section labels of the brain CT image.

[0070] Since the target deformation field identifies the deformation pattern between the brain CT image and the standard brain CT image, the target deformation field can be used to deform the brain anatomical structure partition labels of the standard brain CT image to obtain the brain anatomical structure partition labels of the brain CT image.

[0071] S204: Identify brain parenchymal regions from brain CT images.

[0072] The brain tissue area in brain CT images mainly includes brain parenchyma, extracerebral space, falx cerebri, and ventricles. Since the embodiments of this application mainly aim to locate the hemorrhage in the brain parenchyma, it is necessary to identify the brain parenchyma region from brain CT images.

[0073] In brain CT images, hemorrhage areas appear as high-density features, the falx cerebri and ventricles appear as strip-shaped high-density features, the extracerebral space appears as low-density features, and the brain parenchyma shows a uniform density distribution of medium gray. Since the extracerebral space, falx cerebri, and ventricles are relatively easy to identify, the brain parenchyma can be obtained by removing these regions from the brain tissue.

[0074] In one possible implementation, the specific implementation of S204 in identifying brain parenchyma regions from brain CT images may include:

[0075] A1: Identify the extracerebral space region, falx cerebri region, and ventricle region from brain CT images.

[0076] See Figure 4 As shown, extracerebral spaces can be identified layer by layer from brain CT images, with the darker areas representing extracerebral spaces. See also Figure 5 As shown, the falx cerebri region can be identified layer by layer from a brain CT image, with the darker area representing the falx cerebri region. Similarly, the ventricular region can be identified layer by layer from a brain CT image. This application does not limit the specific implementation method for identifying the extracerebral space region, falx cerebri region, and ventricular region from brain CT images.

[0077] A2: The brain parenchyma is obtained by removing the extracerebral space region, falx cerebri region, and ventricle region from brain CT images.

[0078] By removing the extracerebral space, falx cerebri, and ventricles from each slice of a brain CT image, the brain parenchyma regions included in the brain CT image can be obtained. See also Figure 6 The image shown is a schematic diagram of the brain parenchyma region in a brain CT image.

[0079] S205: Determine the segmentation threshold for each slice of brain CT image based on the CT value histogram of the brain parenchyma region in each slice of brain CT image.

[0080] In brain CT images, hemorrhage areas exhibit high-density characteristics. These areas can be identified by pinpointing CT values ​​exceeding a segmentation threshold; therefore, a segmentation threshold must be determined first. However, limitations of the CT scanning equipment itself can lead to abnormally high overall CT values ​​in certain slices, for example, the skull may be more prominent in sections closer to the top of the head, resulting in a more pronounced volumetric effect. Therefore, the segmentation threshold needs to be calculated separately for each slice, as a global threshold lacks robustness.

[0081] Specifically, the segmentation threshold for each slice of a brain CT image can be determined based on the CT value histogram of the brain parenchyma region within each slice. The CT value histogram is a statistical representation of the number of pixels for each CT value in the brain parenchyma region of a slice image. By analyzing the CT value histogram, the segmentation threshold for each slice image can be determined.

[0082] In one possible implementation, S205, based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image, determines the segmentation threshold corresponding to each slice of the brain CT image. The specific implementation of this can include:

[0083] B1: The CT value with the most pixels is obtained from the CT value histogram of the brain parenchyma region in the target slice image as the target CT value; the target slice images are each slice image of the brain CT image.

[0084] This explanation uses a specific slice image as the target slice image as an example. (See also...) Figure 7 The image shown is a histogram of CT values ​​for the brain parenchyma region in the target slice image. The horizontal axis represents the CT value (HU), and the vertical axis represents the number of pixels. Regardless of the presence of hemorrhage, the region with the highest CT value distribution in a single slice image represents normal brain parenchyma. The peak CT value of 41.5 HU in the image represents the region with the highest concentration of CT values ​​(HU) in the brain parenchyma region of this slice image. max This CT value is then determined as the target CT value. Similarly, each slice of a brain CT image can be used as a target slice image to obtain the target CT value in each slice image.

[0085] B2: Multiply the target CT value by a preset weight to obtain the segmentation threshold corresponding to the target slice image.

[0086] Multiplying the target CT value in each slice image by a preset weight yields the segmentation threshold for each slice image. The preset weight can be set based on actual conditions and experience; for example, a preset weight of 1.2 is possible. This embodiment does not limit the value of the preset weight.

[0087] S206: Identify regions larger than the segmentation threshold from each slice of brain CT images as target hemorrhage areas.

[0088] After determining the segmentation threshold for each slice image, regions larger than the corresponding segmentation threshold are identified from each slice image of the brain CT image. This allows for the identification of the target hemorrhage regions included in each slice image, thus forming the three-dimensional target hemorrhage region of the brain CT image. See also... Figure 8 As shown, the target bleeding area is included in each slice image, where the dark area represents the target bleeding area.

[0089] S207: Match the target hemorrhage area with the brain anatomical structure partition labels of the brain CT image to obtain the brain anatomical structure partition where the target hemorrhage area is located.

[0090] Matching the target hemorrhage area with the brain anatomical structure partition labels on brain CT images, that is, determining which brain anatomical structure partitions the target hemorrhage area intersects with, can obtain the brain anatomical structure partition where the target hemorrhage area is located, thereby achieving the localization of the hemorrhage area.

[0091] Based on the descriptions in S201-S207, this embodiment of the application acquires a standard brain CT image and its corresponding brain anatomical structure partition labels. The brain CT image is then registered with the standard brain CT image to generate a target deformation field. This target deformation field is applied to the brain anatomical structure partition labels of the standard brain CT image to obtain the brain anatomical structure partition labels for the brain CT image. Then, by determining the segmentation threshold corresponding to each slice image, the target hemorrhage region is identified from the brain CT image. Matching the target hemorrhage region with the brain anatomical structure partition labels of the brain CT image yields the brain anatomical structure partition where the target hemorrhage region is located, thereby locating the site of the brain hemorrhage.

[0092] Since the imaging quality of brain CT images is generally poor and there will be noise, and the noise is highlighted, which can be easily confused with the hemorrhage area, the target hemorrhage area can be corrected.

[0093] In some possible implementations, after S206 identifying regions larger than a segmentation threshold as target hemorrhage regions from each slice of the brain CT image, and before S207 matching the target hemorrhage regions with brain anatomical structure partition labels of the brain CT image, the following steps may also be included:

[0094] C1: Perform an opening operation on the target bleeding area at a preset scale to generate the processed target bleeding area.

[0095] An opening operation is performed on the currently identified target bleed region, for example, a one-scale opening operation, to obtain the processed target bleed region. Performing the opening operation is equivalent to removing noise and abnormally isolated highlight areas that were mistakenly segmented into the target bleed region according to the threshold segmentation.

[0096] C2: Calculate the number of three-dimensional connected components in the target bleeding region minus the number of three-dimensional connected components in the processed target bleeding region, and then divide by the number of three-dimensional connected components in the target bleeding region to obtain the target value.

[0097] Since noise, abnormally bright areas, and artifacts have obvious discrete distribution and poor three-dimensional connectivity, the target bleeding area and segmentation threshold can be iteratively adjusted through three-dimensional connectivity.

[0098] Specifically, the target value can be obtained by subtracting the number of three-dimensional connected components in the processed target bleeding region from the number of three-dimensional connected components in the target bleeding region, and then dividing by the number of three-dimensional connected components in the target bleeding region.

[0099] C3: If the target value is greater than the target threshold, retain the target bleeding area.

[0100] If the target value is greater than the target threshold, it indicates that the number of three-dimensional connected components in the target bleed region is similar to the number of three-dimensional connected components in the processed target bleed region, meaning that not too much noise has been removed, and the target bleed region can be retained. The target threshold can be set according to actual conditions; this application embodiment does not limit the value of the target threshold.

[0101] C4: If the target value is less than or equal to the target threshold, increase the segmentation threshold, re-execute the process of identifying regions with values ​​greater than the segmentation threshold from each slice of the brain CT image as target hemorrhage regions, perform an opening operation on the target hemorrhage region at a preset scale, generate the processed target hemorrhage region and subsequent steps.

[0102] If the target value is less than or equal to the target threshold, it indicates that the number of noise points removed is too large, and the currently identified target hemorrhage area is inaccurate. It is necessary to further increase the segmentation threshold and re-execute the steps of C1-C4 in this embodiment to identify areas greater than the segmentation threshold from each slice of the brain CT image as target hemorrhage areas until the target hemorrhage area is retained.

[0103] In one possible implementation, the specific implementation of increasing the segmentation threshold may include:

[0104] Divide the segmentation threshold by the preset value and add it back to the original segmentation threshold.

[0105] The preset value can be set according to the actual situation. In this embodiment of the application, the value of the preset value is not limited.

[0106] In one example, the final target bleeding area vBloodmask_end can be obtained according to the following formula.

[0107]

[0108] Where NumBefore represents the number of 3D connected components in the target bleeding region, and NumAfter represents the number of 3D connected components in the processed target bleeding region. The target threshold is 0.9. If the calculated target value is greater than 0.9, the target bleeding region is retained as the final target bleeding region vBloodmask_end; otherwise, the segmentation threshold TH is increased, where Δth = TH / 20, i.e., the preset value is 20.

[0109] This embodiment can remove noise in the target bleeding area, resulting in a more accurate target bleeding area.

[0110] Furthermore, because the compression of normal brain tissue by hemorrhage causes deformation of the brain tissue distribution, the anatomical region labels on brain CT images will also be deformed. Therefore, the anatomical region labels on brain CT images can be corrected.

[0111] In one possible implementation, before matching the target hemorrhage area with the brain anatomical structure partition labels of the brain CT image in S207, the following steps may also be included:

[0112] D1: The side containing the target bleeding area is designated as the side to be corrected, and the opposite side of the target bleeding area is designated as the reference side.

[0113] D2: The brain anatomical structure partition labels in the brain CT image are symmetrically mapped along the center line to the side to be corrected, replacing the partition labels on the side to be corrected, in order to regenerate the brain anatomical structure partition labels of the brain CT image.

[0114] Brain hemorrhage typically occurs on only one side of the brain, and the two sides of the brain tissue are symmetrically distributed. Therefore, all the partition labels on the side without the target hemorrhage area are symmetrically mapped to the other side along the central line to regenerate the brain anatomical structure partition labels of the brain CT image.

[0115] See Figure 9As shown, in one example, the midline symmetry method is used to symmetrically fit region 1 on the normal side (reference side) to region 2 on the abnormal side (side to be corrected), and symmetrically fit to generate region 3. Similarly, all regions on the affected side are symmetrically fitted to generate new regions.

[0116] The embodiments of this application modify the brain anatomical structure partition labels on the side where the target hemorrhage area is located, thereby making the localization of the target hemorrhage area more accurate.

[0117] In some possible implementations, the volume of the target bleed region can also be obtained based on the number of pixels in the target bleed region and the volume of a unit pixel.

[0118] In practical implementation, the bleed volume can also be quantified by multiplying the number of pixels in the target bleed region by the volume per unit pixel. Here, the volume per unit pixel Δu = spacing. x ×spacing y ×spacing z / 1000, spacing is the spatial resolution.

[0119] The embodiments of this application have high detection accuracy for dense hemorrhages such as brain parenchymal hemorrhage and diffuse hemorrhages such as subarachnoid hemorrhage. They can accurately locate the anatomical location and volume of the hemorrhage. The adaptive threshold segmentation method based on physiological characteristics makes the detection of the target hemorrhage area robust.

[0120] Based on the method embodiment provided above for locating a target region in a brain CT image, this application embodiment also provides a device for locating a target region in a brain CT image, which will be described below with reference to the accompanying drawings.

[0121] See Figure 10 As shown in the figure, this is a schematic diagram of the structure of a target region localization device in a brain CT image provided in an embodiment of this application. Figure 10 As shown, the device for locating the target region in the brain CT image includes:

[0122] The acquisition unit 1001 is used to acquire a standard brain CT image and the brain anatomical structure partition labels of the standard brain CT image;

[0123] The registration unit 1002 is used to input the brain CT image and the standard brain CT image into the registration network to obtain a target deformation field, wherein the target deformation field identifies the deformation pattern between the brain CT image and the standard brain CT image.

[0124] Deformation unit 1003 is used to apply the target deformation field to the brain anatomy partition label of the standard brain CT image to obtain the brain anatomy partition label of the brain CT image.

[0125] The first identification unit 1004 is used to identify brain parenchyma regions from the brain CT image;

[0126] The first determining unit 1005 is used to determine the segmentation threshold corresponding to each slice of the brain CT image based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image.

[0127] The second identification unit 1006 is used to identify regions larger than the segmentation threshold as target hemorrhage regions from each slice of a brain CT image;

[0128] The matching unit 1007 is used to match the target hemorrhage area with the brain anatomical structure partition label of the brain CT image to obtain the brain anatomical structure partition label where the target hemorrhage area is located.

[0129] In one possible implementation, the determining unit includes:

[0130] The sub-unit is used to obtain the CT value with the most pixels from the CT value histogram of the brain parenchyma region in the target slice image as the target CT value; the target slice image is each slice image of the brain CT image;

[0131] The calculation subunit is used to multiply the target CT value by a preset weight to obtain the segmentation threshold corresponding to the target slice image.

[0132] In one possible implementation, the device further includes:

[0133] The first calculation unit is used to perform an opening operation on the target bleeding area at a preset scale to generate the processed target bleeding area.

[0134] The second calculation unit is used to calculate the number of three-dimensional connected components in the target bleeding region minus the number of three-dimensional connected components in the processed target bleeding region, and then divide by the number of three-dimensional connected components in the target bleeding region to obtain the target value;

[0135] A retention unit is used to retain the target bleeding area if the target value is greater than the target threshold.

[0136] The processing unit is configured to increase the segmentation threshold if the target value is less than or equal to the target threshold.

[0137] The triggering unit is used to re-execute the second recognition unit to identify regions larger than the segmentation threshold from each slice of the brain CT image as target hemorrhage regions. The first calculation unit performs an opening operation on the target hemorrhage region at a preset scale to generate the processed target hemorrhage region and subsequent steps.

[0138] In one possible implementation, the processing unit is specifically used for:

[0139] If the target value is less than or equal to the target threshold, the segmentation threshold is divided by a preset value and then added back to the segmentation threshold to obtain a new segmentation threshold.

[0140] In one possible implementation, the first identification unit includes:

[0141] The identification subunit is used to identify the extracerebral space region, falx cerebri region, and ventricle region from the brain CT image;

[0142] The removal subunit is used to remove the extracerebral space region, the falx cerebri region, and the ventricular region from the brain CT image to obtain the brain parenchyma region.

[0143] In one possible implementation, the device further includes:

[0144] The second determining unit is used to determine the side where the target bleeding area is located as the side to be corrected, and the other side where the target bleeding area is located as the reference side;

[0145] The mapping unit is used to symmetrically map the brain anatomical structure partition labels of the brain CT image on the reference side to the side to be corrected along the center line, and replace the partition labels on the side to be corrected, so as to regenerate the brain anatomical structure partition labels of the brain CT image.

[0146] In one possible implementation, the device further includes:

[0147] The third determining unit is used to obtain the volume of the target bleeding region based on the number of pixels in the target bleeding region and the volume of a unit pixel.

[0148] In addition, this application embodiment also provides a device for locating a target region in a brain CT image, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method for locating a target region in a brain CT image as described in any of the above claims.

[0149] In addition, embodiments of this application also provide a computer-readable storage medium storing instructions that, when executed on a terminal device, cause the terminal device to perform the method for locating a target region in a brain CT image as described in any of the preceding claims.

[0150] This application embodiment acquires a standard brain CT image and its corresponding brain anatomical structure partition labels. The brain CT image is then registered with the standard brain CT image to generate a target deformation field. This target deformation field is applied to the brain anatomical structure partition labels of the standard brain CT image, thereby obtaining the brain anatomical structure partition labels for the brain CT image. Then, by determining the segmentation threshold corresponding to each slice image, the target hemorrhage region is identified from the brain CT image. Matching the target hemorrhage region with the brain anatomical structure partition labels of the brain CT image allows for the determination of the brain anatomical structure partition where the target hemorrhage region is located, thus achieving the localization of the brain hemorrhage site.

[0151] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems or apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and relevant parts can be referred to the method section.

[0152] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0153] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0154] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0155] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of positioning a target region in a brain CT image, characterized by, The method includes: Obtain standard brain CT images and brain anatomical structure partition labels of the standard brain CT images; The brain CT image and the standard brain CT image are input into the registration network to obtain the target deformation field, which identifies the deformation pattern between the brain CT image and the standard brain CT image. The target deformation field is applied to the brain anatomy section label of the standard brain CT image to obtain the brain anatomy section label of the brain CT image. Identify brain parenchymal regions from the brain CT images; Based on the histogram of CT values ​​of the brain parenchyma region in each slice of the brain CT image, the segmentation threshold corresponding to each slice of the brain CT image is determined; Identify regions larger than the stated segmentation threshold from each slice of a brain CT image as target hemorrhage areas; The target hemorrhage area is matched with the brain anatomical structure partition labels of the brain CT image to obtain the brain anatomical structure partition where the target hemorrhage area is located. The step of determining the segmentation threshold for each slice of the brain CT image based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image includes: The target CT value is obtained by taking the CT value with the most pixels from the CT value histogram of the brain parenchyma region in the target slice image; the target slice image is each slice image of the brain CT image. The target CT value is multiplied by a preset weight to obtain the segmentation threshold corresponding to the target slice image.

2. The method according to claim 1, characterized in that, After identifying regions larger than the segmentation threshold as target hemorrhage regions from each slice of a brain CT image, and before matching the target hemorrhage regions with brain anatomical structure partition labels of the brain CT images, the method further includes: An opening operation of a preset scale is performed on the target bleeding area to generate a processed target bleeding area; The target value is obtained by subtracting the number of three-dimensional connected components of the processed target bleeding region from the number of three-dimensional connected components of the target bleeding region, and then dividing by the number of three-dimensional connected components of the target bleeding region. If the target value is greater than the target threshold, the target bleeding area is retained; If the target value is less than or equal to the target threshold, the segmentation threshold is increased, and the process of identifying regions with values ​​greater than the segmentation threshold from each slice of the brain CT image as target hemorrhage regions is repeated. An opening operation of a preset scale is performed on the target hemorrhage regions to generate the processed target hemorrhage regions and subsequent steps.

3. The method according to claim 2, characterized in that, Increasing the segmentation threshold includes: Divide the segmentation threshold by a preset value and add the segmentation threshold back to obtain the new segmentation threshold.

4. The method according to claim 1, characterized in that, The process of identifying brain parenchyma regions from the brain CT images includes: Identify the extracerebral space region, falx cerebri region, and ventricle region from the brain CT images; The brain parenchyma region is obtained by removing the extracerebral space region, the falx cerebri region, and the ventricle region from the brain CT image.

5. The method according to claim 1 or 3, characterized in that, Before matching the target hemorrhage region with the brain anatomical structure partition labels of the brain CT image, the method further includes: The side containing the target bleeding area is defined as the side to be corrected, and the opposite side of the target bleeding area is defined as the reference side. The brain anatomical structure partition labels of the brain CT image are symmetrically mapped along the center line to the side to be corrected, replacing the partition labels on the side to be corrected, in order to regenerate the brain anatomical structure partition labels of the brain CT image.

6. The method according to claim 1 or 2, characterized in that, The method further includes: The volume of the target bleed region is obtained based on the number of pixels in the target bleed region and the volume of a unit pixel.

7. A device for locating a target region in a brain CT image, characterized in that, The device includes: The acquisition unit is used to acquire a standard brain CT image and the brain anatomical structure partition labels of the standard brain CT image; The registration unit is used to input the brain CT image and the standard brain CT image into the registration network to obtain the target deformation field, which identifies the deformation pattern between the brain CT image and the standard brain CT image. A deformation unit is used to apply the target deformation field to the brain anatomy partition label of the standard brain CT image to obtain the brain anatomy partition label of the brain CT image. The first identification unit is used to identify brain parenchyma regions from the brain CT image; The first determining unit is used to determine the segmentation threshold corresponding to each slice of the brain CT image based on the CT value histogram of the brain parenchyma region in each slice of the brain CT image. The second identification unit is used to identify regions larger than the segmentation threshold as target hemorrhage areas from each slice of the brain CT image; A matching unit is used to match the target hemorrhage area with the brain anatomical structure partition labels of the brain CT image to obtain the brain anatomical structure partition label where the target hemorrhage area is located. The determining unit includes: The sub-unit is used to obtain the CT value with the most pixels from the CT value histogram of the brain parenchyma region in the target slice image as the target CT value; the target slice image is each slice image of the brain CT image; The calculation subunit is used to multiply the target CT value by a preset weight to obtain the segmentation threshold corresponding to the target slice image.

8. A device for locating a target region in a brain CT image, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for locating a target region in a brain CT image as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a terminal device, cause the terminal device to perform the method for locating a target region in a brain CT image as described in any one of claims 1-6.

Images