WO2021057148A1

WO2021057148A1 - Brain tissue layering method and device based on neural network, and computer device

Info

Publication number: WO2021057148A1
Application number: PCT/CN2020/098936
Authority: WO
Inventors: 卓柏全; 周鑫; 吕传峰
Original assignee: 平安科技（深圳）有限公司
Priority date: 2019-09-25
Filing date: 2020-06-29
Publication date: 2021-04-01
Also published as: CN110827236A; CN110827236B

Abstract

A brain tissue layering method and device based on a neural network, a computer device, and a storage medium, relating to the field of artificial intelligence. The method comprises: obtaining brain CT image information (201); extracting features of the brain CT image information by means of a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image (202); performing a candidate box alignment operation on the feature map of the brain CT image to obtain semantic information of an instance category and location information of an instance pixel level (203); and inputting the semantic information of the instance category and the location information of the instance pixel level into a pre-trained layering neural network, and outputting a layered result of the brain CT image (204). By fusing the brain cutting convolutional neural network and the brain layering neural network, the results of brain cutting and brain layering are obtained simultaneously by using one model, so that the operation time and the consumption of operation resources are reduced, and the tasks of brain cutting and brain layering can share feature information, thereby improving the accuracy of brain layering.

Description

Brain tissue layering method, device and computer equipment based on neural network

This application is based on the Chinese invention patent application filed on September 25, 2019 with the application number 2019109090928, titled "Neural network-based brain tissue layering method, device, and computer equipment", and claims its priority.

Technical field

This application relates to the field of artificial intelligence technology, and in particular to a brain tissue layering method, device, computer equipment, and storage medium based on neural networks.

Background technique

In recent years, deep learning technology has been widely used in various fields, especially computer vision, which is used to realize face recognition, target detection, image segmentation, etc. In the medical field, it is usually necessary to analyze brain CT images, which contain a lot of information. In addition to cerebral hemorrhage detection and brain tissue segmentation, even brain stratification is also a very important piece of information to be provided. At present, neural networks in deep learning technology are commonly used to construct models to achieve tasks such as detection, classification, and prediction. However, in the process of implementing this application, the inventor realized that multiple models are usually used to handle multiple tasks respectively. For example, a detection model for cerebral hemorrhage, a segmentation model for brain tissue, and a classification model for brain stratification. Such a strategy of multiple models to solve multiple tasks will double the computing time and computing resource consumption. And because the information of a single classification model is not comprehensive, the model misclassifies unseen brain CT images, thereby reducing the accuracy of brain stratification.

technical problem

Using multiple models to handle multiple tasks respectively will double the computing time and computing resource consumption, which reduces the accuracy of brain layering.

Technical solutions

In order to solve the above technical problems, an embodiment of the present application provides a brain tissue layering method based on a neural network, which includes the following steps:

Obtain imaging information of brain CT;

Extracting features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

After performing a candidate frame alignment operation on the feature map of the brain CT image, the semantic information of the instance category and the pixel-level position information of the instance are obtained;

The semantic information of the instance category and the pixel-level position information of the instance are input to the pre-trained layered neural network, and the layered result of the brain CT image is output.

In order to solve the above technical problems, an embodiment of the present application also provides a brain tissue layering device based on a neural network, which adopts the following technical solutions:

Brain tissue layering device based on neural network, including:

The first acquisition module is used to acquire brain CT image information;

An extraction module for extracting the features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

The second acquiring module is used to acquire the semantic information of the instance category and the pixel-level position information of the instance after performing the candidate frame alignment operation on the feature map of the brain CT image;

The output module is used to input the semantic information of the instance category and the pixel-level position information of the instance into the pre-trained layered neural network, and output the layered result of the brain CT image.

In order to solve the above technical problems, the embodiments of the present application also provide a computer device, which adopts the following technical solutions:

The computer device includes a memory and a processor, and a computer process is stored in the memory. When the processor executes the computer process, the steps of the neural network-based brain tissue layering method are implemented:

Obtain imaging information of brain CT;

In order to solve the above technical problems, the embodiments of the present application also provide a computer-readable storage medium, which adopts the following technical solutions:

The computer-readable storage medium stores a computer process, and when the computer process is executed by a processor, the steps of the neural network-based brain tissue layering method are realized:

Obtain imaging information of brain CT;

Beneficial effect

In this embodiment, the image information of the brain CT is acquired; the features of the brain CT image information are extracted through the pre-trained brain cutting convolutional neural network to obtain the feature map of the brain CT image; the brain CT image After the candidate frame alignment operation is performed on the feature map, the semantic information of the instance category and the pixel-level position information of the instance are obtained; the semantic information of the instance category and the pixel-level position information of the instance are input to the pre-trained hierarchical neural network, and the output is The stratification result of the brain CT image. The brain CT image information is extracted from the pre-trained brain cutting convolutional neural network to obtain the feature map. After the feature map is subjected to the candidate frame alignment operation, the semantic information of the instance category and the pixel-level position information of the instance are obtained as the result of brain segmentation. The semantic information and pixel-level position information of the instance are obtained through the pre-trained layered neural network to obtain the brain layered results of the brain CT image. By fusing the brain cutting convolutional neural network and the brain layered neural network, one model can be obtained at the same time The result of brain segmentation and brain layering reduces the computing time and the consumption of computing resources, and enables the two tasks of brain segmentation and brain layering to share feature information, thereby improving the accuracy of brain layering.

Description of the drawings

In order to explain the solution in this application more clearly, the following will briefly introduce the drawings used in the description of the embodiments of the application. Obviously, the drawings in the following description are some embodiments of the application. Ordinary technicians can obtain other drawings based on these drawings without creative work.

Figure 1 is an exemplary system architecture diagram to which the present application can be applied;

Fig. 2 is a flowchart of an embodiment of a method for brain tissue layering based on neural network according to the present application;

FIG. 3 is a flowchart of a specific implementation of step 202 in FIG. 2;

FIG. 4 is a flowchart of a specific implementation of step 203 in FIG. 2;

FIG. 5 is a flowchart of a specific implementation of step 204 in FIG. 2;

Fig. 6 is a schematic structural diagram of an embodiment of an apparatus for layering brain tissue based on a neural network according to the present application;

Fig. 7 is a schematic structural diagram of an embodiment of a computer device according to the present application.

The best mode of the present invention

Referring to FIG. 2, there is shown a flowchart of an embodiment of a method for layering brain tissue based on a neural network according to the present application. The brain tissue layering method based on neural network includes the following steps:

Step 201: Obtain brain CT image information.

In this embodiment, the electronic device (such as the server/terminal device shown in FIG. 1) on which the neural network-based brain tissue layering method runs can obtain brain CT image information through a wired connection or a wireless connection. It should be pointed out that the above-mentioned wireless connection methods can include but are not limited to 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, as well as other wireless connection methods that are currently known or developed in the future.

The above-mentioned brain CT image information can be obtained by scanning the target object with a CT machine, and then obtained from the CT machine through the above-mentioned wired connection or wireless connection, or from the data package derived from the CT machine, or by extracting DICOM The data format is obtained from brain CT image data stored in the database. Through wired connection or wireless connection, the brain CT image information of multiple CT machines can be obtained at the same time, and the data transmission capacity can be improved.

Step 202: Extract the features of the brain CT image information through the pre-trained brain cutting convolutional neural network to obtain the feature map of the brain CT image.

In this embodiment, Convolutional Neural Networks (CNN) is a type of Feedforward Neural Networks (Feedforward Neural Networks) that includes convolution calculations and has a deep structure, and is one of the representative algorithms of deep learning. The convolutional neural network includes: data input layer, convolutional calculation layer, ReLU excitation layer, pooling layer, and fully connected layer. The purpose is to extract features from a certain model, and then classify, identify, and identify things based on the features. Forecast or decision-making, etc.

The above-mentioned brain cutting convolutional neural network obtains the corresponding cutting feature map by performing convolution calculation on the input brain CT image data. It should be noted that the above feature map is the intermediate image data in the brain cutting convolutional neural network, not the final brain cutting image data.

Step 203: After performing a candidate frame alignment operation on the feature map of the brain CT image, the semantic information of the instance category and the pixel-level position information of the instance are obtained.

In this embodiment, after the feature map of the brain CT image is obtained, a candidate frame is generated for each pixel and the candidate frame alignment operation is performed. The candidate frame is the position of the pixel on the feature map, and the network can be generated directly through a region (Region Proposal Network (RPN) is generated; then, an instance mask is generated for each candidate frame after the alignment operation of the candidate frame through a full convolutional neural network (FCN network), so as to segment the instance; finally, the instance segmented by the mask is input to The fully connected layer network obtains the semantic information of the instance category of the feature map and the pixel-level location information of the instance as a result of brain cutting; the semantic information of the instance category includes the category label of the above-mentioned instance, and the location information includes the coordinate of the instance on the feature map .

Step 204: Input the semantic information of the instance category and the pixel-level position information of the instance into the pre-trained layered neural network, and output the layered result of the brain CT image.

In this embodiment, the above-mentioned layered neural network includes a convolutional layer and a fully connected layer. The convolutional layer is used to perform secondary feature extraction on the segmented instances to obtain highly separable local features of the instance, and then combine the local features It is input to the fully connected layer for feature combination and then classified by the logistic regression of the layer, and the category label is obtained and output as the layered result of the above-mentioned brain CT image. By sharing the instance feature information extracted by the brain cutting convolutional neural network, the accuracy of brain layering can be improved.

It should be noted that the above-mentioned brain cutting convolutional neural network and hierarchical neural network need to be pre-trained, that is, after the neural network model is constructed, the training data set is input to the model to make the output of the model meet the expectations or the error is as small as possible. Among them, the acquired sample data set needs to be pre-processed, such as positive and negative examples sampling, filtering, etc. In this embodiment, cross-checking is applied after the brain CT image sample data is obtained, and the sample data is divided into four points Using three of them to train and one test, the samples with obvious differences in the test set can be shaved off, and the filtered sample data is the real training sample and the verification sample.

Embodiments of the present invention

Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the technical field of the application; the terms used in the specification of the application herein are only for describing specific embodiments. The purpose is not to limit the application; the terms "including" and "having" in the specification and claims of the application and the above-mentioned description of the drawings and any variations thereof are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the application or the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific sequence.

Reference to "embodiments" herein means that a specific feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art clearly and implicitly understand that the embodiments described herein can be combined with other embodiments.

In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the accompanying drawings.

As shown in FIG. 1, FIG. 1 is a system architecture diagram that may be used in this application. The system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, and so on.

The user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, data, and so on. The terminal devices 101, 102, 103 may be installed with various communication client application APPs, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

The terminal devices 101, 102, 103 can be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, dynamic Video experts compress standard audio level 3), MP4 (Moving Picture Experts Group Audio Layer IV, Motion Picture Experts compress standard audio layer 4) Players, laptops and desktop computers, etc.

The server 105 may be a server that provides various services, for example, a background server that provides support for pages displayed on the terminal devices 101, 102, and 103.

It should be noted that the neural network-based brain tissue layering method provided in the embodiments of the present application is generally executed by a server/terminal device. Correspondingly, the neural network-based brain tissue layering device is generally set in the server/terminal device.

It should be understood that the numbers of terminal devices, networks, and servers in FIG. 1 are merely illustrative. According to implementation needs, there can be any number of terminal devices, networks, and servers.

Continuing to refer to FIG. 2, a flowchart of an embodiment of a method for layering brain tissue based on a neural network according to the present application is shown. The brain tissue layering method based on neural network includes the following steps:

Step 201: Obtain brain CT image information.

Further, before the step 201 of acquiring brain CT image information, the method further includes:

Step 200: Perform channel preprocessing on the brain CT image to obtain brain CT image information.

Among them, CT (Computed Tomography imaging is a scanning method that uses computer technology to reconstruct the tomographic image of the measured object to obtain a three-dimensional tomographic image. This scanning method is to penetrate the object under test through the rays of a single axis. According to the different absorption and transmittance of each part of the object under test, the transmitted rays are collected by the computer and imaged through three-dimensional reconstruction. Brain CT images include information on the structure and shape of brain tissue, which can be used to clearly show the number, location, size, contour, density, intratumoral hemorrhage, calcification, and degree of spread of intracranial tumors.

The above-mentioned brain CT images can be obtained by scanning the target object by a CT machine and are mostly stored in DICOM data format. To obtain brain CT image data, it is necessary to parse the DICOM file. The structured information of DICOM can be divided into four layers: Patient, Study, Series and Image. Each information (IE, Information entity) is stored by the combination of "key-value" (Tag, Value), and the brain information can be converted from DICOM file to HU value image by analyzing the Tag location, and then the above-mentioned brain CT image information is obtained through channel preprocessing. The CT image of HU value is converted into a three-channel gray image value according to the three window widths and window levels of full window (range of 0-4096), brain window (40,120), and bone window (450, 2000).

Further, as shown in FIG. 3, the above step 202 specifically includes:

Step 2021: Input the acquired brain CT image information into the trained ResNet convolutional neural network, and extract the feature map of the brain CT image.

In this embodiment, you can choose a standard convolutional neural network (usually ResNet50 and ResNet101) is trained as a feature extractor. The bottom layer of the convolutional neural network detects low-level features (edges and corners, etc.), and the higher layer detects more advanced features (cars, people, sky, etc.). The brain CT image data preprocessed by the above channels are input to the input layer of the above-mentioned standard trained convolutional neural network, and the feature map of the image is obtained after convolution calculation, pooling dimensionality reduction, and fully connected classification. The feature map will be used as the input data for the next step.

Further, as shown in FIG. 4, the above step 203 specifically includes:

Step 2031: Obtain a candidate frame of each pixel on the feature map of the brain CT image and perform a candidate frame alignment operation on the obtained candidate frame.

In this embodiment, firstly, 9 kinds of anchor points are generated for each pixel on the above feature map through the region generation network (RPN). These 9 kinds of initial anchors can contain three kinds of areas (128×128, 256×256, 512×512), each area can contain three aspect ratios (1:1, 1:2, 2:1). For the generated anchor, the RPN first judges whether the anchor is the foreground or the background, that is, whether the anchor covers the target or not, and the second is the first coordinate correction for the anchor belonging to the foreground, so as to obtain the candidate frame of each pixel. . Then return to the feature map to perform feature selection based on the obtained candidate box, and mark the instance through the candidate box.

Step 2032: Input the feature map after the candidate frame alignment operation is performed into the fully connected layer network to obtain the semantic information of the instance category of the feature map and the position information of the instance pixel level.

Among them, fully connected layers (FC) play the role of a "classifier" in the neural network, and can integrate local information with category discrimination in the convolutional layer or the pooling layer. For example, in convolutional neural networks, such as in CNN, the fully connected layer often appears in the last few layers. Operations such as the convolutional layer, pooling layer, and activation function layer map the original data to the hidden layer feature space, and the fully connected layer It plays the role of mapping the learned "distributed feature representation" to the sample label space. In actual use, the fully connected layer can be realized by the convolution operation: the fully connected layer that is fully connected to the previous layer can be transformed into a convolution with a 1x1 convolution kernel; and the fully connected layer that is the convolutional layer in the previous layer can be transformed into The convolution kernel is the global convolution of hxw, and h and w are the height and width of the previous convolution result.

In this embodiment, the above-mentioned examples labeled with candidate boxes obtained after candidate box alignment are input into the fully connected layer network, and the semantic information of the instance (ie, the category label of the instance) is obtained after the candidate box classification is performed, and then the candidate box regression is performed (Further fine-tune the position and size of the candidate frame) Obtain the pixel-level position information of the instance (including the pixel points of the instance and the coordinates of the pixel points on the feature map).

Further, after step 2031 and before step 2032, it may further include:

In step 20311, a mask is generated for each pixel after the candidate frame alignment operation through the full convolutional neural network, and the instances are segmented.

In this embodiment, a mask neural network branch containing a full convolutional neural network (FCN) is used to generate an instance mask for each pixel after the candidate frame alignment operation, so that the features can be segmented. Different examples on the map. Among them, FCN can accept input images of any size, and use the deconvolution layer to upsample the feature map of the last convolution layer to restore it to the same size of the input image, so that a prediction can be generated for each pixel At the same time, the spatial information in the original input image is retained, and finally pixel-by-pixel classification is performed on the up-sampled feature map.

Further, as shown in FIG. 5, the above step 204 specifically includes:

Step 2041: Jointly input the semantic information of the instance category and the pixel-level position information of the instance into the convolutional layer of the hierarchical neural network to extract features again.

Among them, fully connected layers (FC) play the role of a "classifier" in the neural network, and can integrate local information with category discrimination in the convolutional layer or the pooling layer. In this embodiment, the convolutional layer of the hierarchical network can be used to perform convolution and pooling operations on the joint information of the instances of the input network (that is, the semantic information of the instance category and the location information of the instance pixel level), and perform secondary operations. Feature extraction, so as to obtain highly separable local features of the instance, which facilitates accurate stratification of brain CT images.

In step 2042, the features extracted from the convolutional layer of the hierarchical neural network are input to the fully connected layer of the hierarchical neural network for classification, and the hierarchical result of the brain CT image is obtained and output.

In the embodiment of the present application, the local features obtained by the above-mentioned secondary feature extraction are input to the fully connected layer for feature combination, and then classified and predicted through the softmax logistic regression of the layer, that is, several categories and corresponding probabilities, where The category with the highest probability is the result of brain CT image layering and then output the result. Specifically, for brain CT image, you can get which layer of the brain the image shows, such as the sella at the base of the skull and the saddle. Upper cistern layer, top layer of lateral ventricle, etc.

It should be noted that the above-mentioned brain cutting convolutional neural network and hierarchical neural network need to be pre-trained, that is, after the neural network model is constructed, the training data set is input to the model to make the output of the model meet the expectations or the error is as small as possible. Among them, the acquired brain CT image sample data set needs to be preprocessed, such as positive and negative examples sampling, filtering, etc. In this embodiment, after the brain CT image data is obtained, cross-checking is applied to divide the sample data into Four points, using three of them to train a test, you can shave the samples with obvious differences in the test set, and the filtered sample data set is the real training data, and then input to the above-mentioned brain cutting convolutional neural network and layered neural network Perform pre-training and verification.

A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through a computer process. The computer process can be stored in a computer readable storage medium. When executed, it may include the procedures of the above-mentioned method embodiments. Among them, the aforementioned storage medium may be a magnetic disk, an optical disk, a read-only storage memory (Read-Only Memory, ROM) and other non-volatile storage media, or random storage memory (Random Access Memory, RAM) etc.

It should be understood that although the various steps in the flowchart of the drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless explicitly stated in this article, the execution of these steps is not strictly limited in order, and they can be executed in other orders. Moreover, at least part of the steps in the flowchart of the drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the order of execution is also It is not necessarily performed sequentially, but may be performed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

With further reference to FIG. 6, as an implementation of the neural network-based brain tissue layering method shown in FIG. 2, this application provides an embodiment of a neural network-based brain tissue layering device. The device embodiment and the diagram Corresponding to the method embodiment shown in 2, the device can be specifically applied to various electronic devices.

As shown in FIG. 6, the neural network-based brain tissue layering apparatus 300 in this embodiment includes: a first acquisition module 301, an extraction module 302, a second acquisition module 303, and an output module 304. among them:

The first obtaining module 301 is used to obtain brain CT image information;

The extraction module 302 is configured to extract the features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

The second obtaining module 303 is configured to obtain the semantic information of the instance category and the pixel-level position information of the instance after performing the candidate frame alignment operation on the feature map of the brain CT image;

The output module 304 is configured to input the semantic information of the instance category and the pixel-level position information of the instance into the pre-trained layered neural network, and output the layered result of the brain CT image.

In some optional implementation manners of this embodiment, the foregoing apparatus 300 further includes:

The preprocessing module 305 is configured to perform channel preprocessing on the brain CT image to obtain brain CT image information.

The neural network-based brain tissue layering device provided in the embodiments of the present application can realize the various implementation modes in the method embodiments in FIGS. 2 to 5 and the corresponding beneficial effects. To avoid repetition, details are not described herein again.

In order to solve the above technical problems, the embodiments of the present application also provide computer equipment. Please refer to FIG. 7 for details. FIG. 7 is a block diagram of the basic structure of the computer device in this embodiment.

The computer device 7 includes a memory 71, a processor 72, and a network interface 73 that are connected to each other in communication via a system bus. It should be noted that the figure only shows the computer device 7 with components 71-73, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculation and/or information processing in accordance with pre-set or stored instructions. Its hardware includes, but is not limited to, a microprocessor, a dedicated Integrated Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA), Digital Signal Processor (DSP), embedded devices, etc.

The computer device may be a computing device such as a desktop computer, a notebook, a palmtop computer, and a cloud server. The computer device can interact with the user through a keyboard, a mouse, a remote control, a touch panel, or a voice control device.

The memory 71 includes at least one type of readable storage medium, the readable storage medium includes flash memory, hard disk, multimedia card, card type memory (for example, SD or DX memory, etc.), random access memory (RAM), static Random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disks, optical disks, etc., the computer readable storage The medium can be non-volatile or volatile. In some embodiments, the memory 71 may be an internal storage unit of the computer device 7, such as a hard disk or memory of the computer device 7. In other embodiments, the memory 71 may also be an external storage device of the computer device 7, such as a plug-in hard disk equipped on the computer device 7, a smart memory card (Smart Memory Card). Media Card, SMC), Secure Digital (Secure Digital, SD) card, flash card (Flash Card), etc. Of course, the memory 71 may also include both the internal storage unit of the computer device 7 and the external storage device thereof. In this embodiment, the memory 71 is generally used to store the operating system and various application software installed in the computer device 7, such as the process code of a neural network-based brain tissue layering method. In addition, the memory 71 can also be used to temporarily store various types of data that have been output or will be output.

The processor 72 may be a central processing unit (Central Processing Unit) in some embodiments. Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 72 is generally used to control the overall operation of the computer device 7. In this embodiment, the processor 72 is configured to run the process code or process data stored in the memory 71, for example, run the process code of the neural network-based brain tissue layering method.

The network interface 73 may include a wireless network interface or a wired network interface, and the network interface 73 is generally used to establish a communication connection between the computer device 7 and other electronic devices.

This application also provides another implementation manner, that is, a computer-readable storage medium storing a neural network-based brain tissue layering process, and the neural network-based brain tissue layering process The process may be executed by at least one processor, so that the at least one processor executes the steps of the above-mentioned neural network-based brain tissue layering method.

Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。 Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, The optical disc) includes several instructions to enable a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the method described in each embodiment of the present application.

Obviously, the embodiments described above are only a part of the embodiments of the present application, rather than all of the embodiments. The drawings show preferred embodiments of the present application, but do not limit the patent scope of the present application. This application can be implemented in many different forms. On the contrary, the purpose of providing these examples is to make the understanding of the disclosure of this application more thorough and comprehensive. Although this application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, it is still possible for those skilled in the art to modify the technical solutions described in each of the foregoing specific embodiments, or equivalently replace some of the technical features. . All equivalent structures made using the contents of the description and drawings of this application, directly or indirectly used in other related technical fields, are similarly within the scope of patent protection of this application.

Claims

A brain tissue layering method based on neural network, which includes:

Obtain imaging information of brain CT;

Extracting features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

After performing a candidate frame alignment operation on the feature map of the brain CT image, the semantic information of the instance category and the pixel-level position information of the instance are obtained;

The semantic information of the instance category and the pixel-level position information of the instance are input to the pre-trained layered neural network, and the layered result of the brain CT image is output.
The method according to claim 1, wherein before the step of acquiring brain CT image information, the method further comprises the step of:

The brain CT image is preprocessed to obtain the brain CT image information.
The method of claim 1, wherein the step of extracting the features of the brain CT image information through a pre-trained brain cutting convolutional neural network, and obtaining the feature map of the brain CT image specifically comprises:

Input the acquired brain CT image information into the trained ResNet convolutional neural network, and extract the feature map of the brain CT image.
The method according to claim 3, wherein the step of obtaining the semantic information of the instance category and the position information of the instance pixel level after performing the candidate frame alignment operation on the feature map of the brain CT image specifically comprises:

Obtain the candidate frame of each pixel on the feature map of the brain CT image and perform the candidate frame alignment operation on the obtained candidate frame;

The feature map after the candidate frame alignment operation is input into the fully connected layer network, and the semantic information of the instance category of the feature map and the position information of the instance pixel level are obtained.
The method according to claim 4, wherein, after the step of obtaining the candidate frame of each pixel on the feature map of the brain CT image and performing the candidate frame alignment operation on the obtained candidate frame, the method further comprises:

Through the full convolutional neural network, a mask is generated for each pixel point after the alignment operation of the candidate frame, and the instance is segmented.
The method according to claim 4, wherein the semantic information of the instance category and the pixel-level position information of the instance are input to a pre-trained layered neural network, and the layered result of the brain CT image is output The specific steps include:

The semantic information of the instance category and the pixel-level position information of the instance are jointly input to the convolutional layer of the hierarchical neural network to extract the features again;

The features extracted from the convolutional layer of the hierarchical neural network are input to the fully connected layer of the hierarchical neural network for classification, and the hierarchical result of the brain CT image is obtained and output.
A brain tissue layering device based on neural network, which includes:

The first acquisition module is used to acquire brain CT image information;

An extraction module for extracting the features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

The second acquiring module is used to acquire the semantic information of the instance category and the pixel-level position information of the instance after performing the candidate frame alignment operation on the feature map of the brain CT image;

The output module is used to input the semantic information of the instance category and the pixel-level position information of the instance into the pre-trained layered neural network, and output the layered result of the brain CT image.
8. The device according to claim 7, wherein before the first obtaining module, the method further comprises:

The preprocessing module is used for channel preprocessing the brain CT image to obtain brain CT image information.
8. The neural network-based brain tissue layering device according to claim 7, wherein the extraction module comprises:

The extraction sub-module is used to input the acquired brain CT image information into the trained ResNet convolutional neural network, and extract the feature map of the brain CT image.
9. The neural network-based brain tissue layering device according to claim 9, wherein the second acquisition module comprises:

The selection frame alignment sub-module is used to obtain the candidate frame of each pixel on the feature map of the brain CT image and perform the candidate frame alignment operation on the obtained candidate frame;

The second acquisition sub-module is used to input the feature map after the candidate frame alignment operation is performed into the fully connected layer network to obtain the semantic information of the instance category of the feature map and the pixel-level position information of the instance.
A computer device including a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor, wherein the processor executes the computer-readable instructions as follows The steps of the neural network-based brain tissue layering method:

Obtain imaging information of brain CT;

Extracting features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

After performing a candidate frame alignment operation on the feature map of the brain CT image, the semantic information of the instance category and the pixel-level position information of the instance are obtained;

The semantic information of the instance category and the pixel-level position information of the instance are input to the pre-trained layered neural network, and the layered result of the brain CT image is output.
The computer device according to claim 11, wherein, before the step of obtaining brain CT image information, it further comprises the step of:

The brain CT image is preprocessed to obtain the brain CT image information.
11. The computer device of claim 11, wherein the step of extracting features of the brain CT image information through a pre-trained brain cutting convolutional neural network, and obtaining the feature map of the brain CT image specifically comprises:

Input the acquired brain CT image information into the trained ResNet convolutional neural network, and extract the feature map of the brain CT image.
The computer device according to claim 13, wherein the step of obtaining the semantic information of the instance category and the position information of the instance pixel level after performing the candidate frame alignment operation on the feature map of the brain CT image specifically comprises:

Obtain the candidate frame of each pixel on the feature map of the brain CT image and perform the candidate frame alignment operation on the obtained candidate frame;

The feature map after the candidate frame alignment operation is input to the fully connected layer network to obtain the semantic information of the instance category of the feature map and the location information of the instance pixel level.
The computer device according to claim 14, wherein after the step of obtaining the candidate frame of each pixel on the feature map of the brain CT image and performing the candidate frame alignment operation on the obtained candidate frame, the method further comprises:

Through the full convolutional neural network, a mask is generated for each pixel point after the alignment operation of the candidate frame, and the instance is segmented.
A computer-readable storage medium, wherein, when the computer-readable instruction is executed by a processor, the one processing is caused to perform the following steps:

Obtain imaging information of brain CT;

Extracting features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain a feature map of the brain CT image;

After performing a candidate frame alignment operation on the feature map of the brain CT image, the semantic information of the instance category and the pixel-level position information of the instance are obtained;

The semantic information of the instance category and the pixel-level position information of the instance are input to the pre-trained layered neural network, and the layered result of the brain CT image is output.
15. The computer-readable storage medium of claim 16, wherein before the step of acquiring brain CT image information, the method further comprises:

The brain CT image is preprocessed to obtain the brain CT image information.
The computer-readable storage medium of claim 16, wherein the step of extracting the features of the brain CT image information through a pre-trained brain cutting convolutional neural network to obtain the feature map of the brain CT image is specifically include:

Input the acquired brain CT image information into the trained ResNet convolutional neural network, and extract the feature map of the brain CT image.
18. The computer-readable storage medium according to claim 18, wherein the step of obtaining the semantic information of the instance category and the position information of the instance pixel after performing the candidate frame alignment operation on the feature map of the brain CT image specifically comprises:

Obtain the candidate frame of each pixel on the feature map of the brain CT image and perform the candidate frame alignment operation on the obtained candidate frame;

The feature map after the candidate frame alignment operation is input to the fully connected layer network to obtain the semantic information of the instance category of the feature map and the location information of the instance pixel level.
The computer-readable storage medium of claim 19, wherein after the step of obtaining the candidate frame of each pixel on the feature map of the brain CT image and performing the candidate frame alignment operation on the obtained candidate frame, further Including steps:

Through the full convolutional neural network, a mask is generated for each pixel point after the alignment operation of the candidate frame, and the instance is segmented.