CN114494239A

CN114494239A - Focus identification method and device, electronic equipment and computer storage medium

Info

Publication number: CN114494239A
Application number: CN202210146192.1A
Authority: CN
Inventors: 王关政; 吴海萍; 王立龙
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2022-02-17
Filing date: 2022-02-17
Publication date: 2022-05-13

Abstract

The invention relates to an intelligent medical technology, and discloses a focus identification method, which comprises the following steps: acquiring a training focus image and a focus label corresponding to the training focus image, and performing image enhancement on the training focus image to obtain an enhanced image; extracting a characteristic region of the enhanced image, and performing image expansion on the enhanced image based on the characteristic region to obtain an expanded image set; extracting features in the extended image set to obtain image features; performing iterative training on a pre-constructed focus identification model by using image characteristics to obtain a trained focus identification model; and acquiring a focus image to be recognized, and performing focus recognition on the focus image to be recognized by using the trained focus recognition model to obtain a focus type corresponding to the focus image to be recognized. In addition, the invention also relates to a block chain technology, and the training focus image can be stored in the node of the block chain. The invention also provides a focus recognition device, equipment and a medium. The invention can improve the accuracy of the identification of the focus type.

Description

Focus identification method and device, electronic equipment and computer storage medium

Technical Field

The invention relates to the technical field of intelligent medical treatment, in particular to a focus identification method, a focus identification device, electronic equipment and a computer-readable storage medium.

Background

In the medical field, a doctor observes and analyzes a medical image, and then detects information of a focus from the medical image is a common medical means, which greatly helps the doctor to know and analyze the state of an illness of a patient. For example, the analysis of the focus image of the patient can help to understand the type and stage of the focus as early as possible, so as to assist the treatment in time.

At present, for identifying the type of a focus in a medical image, observation and marking are carried out manually by a doctor in most cases. The method is too dependent on the experience of doctors, so that missed detection can occur to the focus which is difficult to observe; and for many different types of lesions with similar forms, the method cannot identify the accurate information of the lesions, and is not beneficial to the subsequent analysis of the disease condition.

Disclosure of Invention

The invention provides a method, a device and a computer readable storage medium for identifying a focus, and mainly aims to solve the problem of low accuracy in identifying a focus type.

In order to achieve the above object, the present invention provides a lesion identification method, including:

acquiring a training focus image and a focus label corresponding to the training focus image, and performing image enhancement on the training focus image to obtain an enhanced image;

extracting a characteristic region of the enhanced image, and performing image expansion on the enhanced image based on the characteristic region to obtain an expanded image set;

performing feature extraction on each expanded image in the expanded image set by using multi-dimensional cross convolution to obtain image features corresponding to each expanded image;

performing iterative training on a pre-constructed focus recognition model by using the image characteristics to obtain a trained focus recognition model;

and acquiring a focus image to be recognized, and performing focus recognition on the focus image to be recognized by using the trained focus recognition model to obtain a focus type corresponding to the focus image to be recognized.

Optionally, the extracting the feature region of the enhanced image includes:

dividing the enhanced image into a plurality of enhanced image blocks according to a preset proportion;

selecting one enhanced image block from the plurality of enhanced image blocks one by one as a target enhanced image block;

generating global features of the target enhanced image blocks according to the pixel gradients in the target enhanced image blocks;

performing frame selection on the areas in the target enhanced image block one by using a preset sliding window to obtain a pixel window;

generating local features of the target enhanced image block according to the pixel value in each pixel window;

and collecting the global features and the local features as the image features of the target enhanced image block.

Optionally, the performing image expansion on the enhanced image based on the feature region to obtain an expanded image set includes:

acquiring a random focus image set, and dividing each image in the random focus image set into image blocks to be expanded corresponding to the plurality of enhanced image blocks according to the preset proportion;

identifying the feature importance of each image feature by using a feature visualization technology, and selecting the image feature with the feature importance greater than a preset threshold value as a target feature;

and mapping the enhanced image block corresponding to each target feature to the position of the corresponding image block to be expanded in each image in the random focus image set to obtain an expanded image set.

Optionally, the generating the global feature of the target enhanced image block according to the pixel gradient in the target enhanced image block includes:

counting the pixel value of each pixel point in the target enhanced image block;

taking the maximum pixel value and the minimum pixel value in the pixel values as parameters of a preset mapping function, and mapping the pixel value of each pixel point in the target enhanced image block to a preset range by using the preset function;

calculating the pixel gradient of each line of pixels in the mapped target enhanced image block, converting the pixel gradient of each line of pixels into a line vector, and splicing the line vector into the global feature of the target enhanced image block.

Optionally, the performing, by using multi-dimensional cross convolution, feature extraction on each expanded image in the expanded image set to obtain an image feature corresponding to each expanded image includes:

selecting one of the extended images from the extended image set one by one as a target image;

performing convolution on the target image by using a preset first-dimension convolution kernel to obtain a first convolution image;

pooling the first convolution image to obtain an intermediate feature;

performing convolution on the intermediate features by using a preset second dimension convolution kernel to obtain a second convolution image;

and performing pooling processing on the second convolution image to obtain image characteristics corresponding to the target image.

Optionally, the performing image enhancement on the training focus image to obtain an enhanced image includes:

uniformly cutting the training focus image to obtain a plurality of focus image blocks;

performing pixel convolution on the plurality of focus image blocks respectively to obtain a plurality of convolution focus image blocks;

respectively carrying out Gaussian smoothing on the plurality of convolution focus image blocks to obtain a plurality of smooth focus image blocks;

and splicing the smooth focus image blocks to obtain a de-noised image of the training focus image.

sequentially selecting regions in the training focus image by using an n x n image window to obtain a plurality of image regions, wherein n is a positive integer;

calculating a binary code element of the central pixel of each image area by using a preset algorithm according to the central pixel of each image area and the neighborhood pixels of the central pixel;

and enhancing the central pixel according to the binary code element to obtain an enhanced image.

In order to solve the above problems, the present invention also provides a lesion recognition apparatus, comprising:

the image matching module is used for acquiring a product image of a product to be recommended and a user image of a candidate user group of the product to be recommended, and calculating a matching value between the product image and each user image;

the first recommending module is used for selecting a first target user group from the candidate user groups according to the matching value and recommending the product to be recommended to the first target user group;

the portrait perfecting module is used for acquiring feedback data of the first target user group to the product to be recommended, and performing portrait perfecting on the user portrait of the first target user group according to the feedback data to obtain a target user portrait;

the image screening module is used for selecting a second target user group from the candidate user groups according to the target user image;

and the second recommending module is used for recommending the product to be recommended to the second target user group.

In order to solve the above problem, the present invention also provides an electronic device, including:

at least one processor; and the number of the first and second groups,

a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,

the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the lesion identification method described above.

In order to solve the above problem, the present invention also provides a computer-readable storage medium having at least one computer program stored therein, the at least one computer program being executed by a processor in an electronic device to implement the lesion identification method described above.

According to the embodiment of the invention, the image enhancement and the image expansion operation are carried out on the training focus image, so that the image quality of the training focus image is improved, the number of the training focus images is also expanded, and the method is favorable for improving the effect of training a more accurate focus recognition model by utilizing the enhanced and expanded image; meanwhile, feature extraction is carried out on the extended image by utilizing multi-dimensional cross convolution, iterative training is carried out on a focus recognition model which is constructed in advance by utilizing the extracted image features, and the accuracy of the extracted image features is improved, so that the accuracy of the trained focus recognition model is improved. Therefore, the method, the device, the electronic equipment and the computer readable storage medium for identifying the focus provided by the invention can solve the problem of low accuracy in identifying the focus type.

Drawings

Fig. 1 is a schematic flow chart illustrating a lesion identification method according to an embodiment of the present invention;

fig. 2 is a schematic diagram illustrating uniform segmentation of a training lesion image according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of image expansion of an enhanced image based on feature areas according to an embodiment of the present invention;

fig. 4 is a functional block diagram of a lesion recognition apparatus according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of an electronic device for implementing the lesion identification method according to an embodiment of the present invention.

The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The embodiment of the application provides a lesion identification method. The subject of the lesion identification method includes, but is not limited to, at least one of the electronic devices that can be configured to execute the method provided by the embodiments of the present application, such as a server, a terminal, and the like. In other words, the lesion recognition method may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like. The server may be an independent server, or may be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), a big data and artificial intelligence platform, and the like.

Fig. 1 is a schematic flow chart of a lesion identification method according to an embodiment of the present invention.

In this embodiment, the lesion recognition method includes:

s1, obtaining a training focus image and a focus label corresponding to the training focus image, and performing image enhancement on the training focus image to obtain an enhanced image.

In an embodiment of the present invention, the lesion training image is a medical image including different lesions, for example, an XCT image including a lesion, an OCT image including a lesion, an MRI image including a lesion, and the like; the lesion label is a label for marking a type of a lesion in the lesion training image, and for example, for eye diseases, a lesion label may include disc edema, ischemic disc lesion, disc tilt, glaucoma, and the like.

In detail, the training lesion image may be captured from a predetermined data storage area for storing medical images by using a computer sentence with image capture function (such as java sentence, python sentence, etc.), wherein the data storage area includes, but is not limited to, a database, a block chain node, and a network cache.

Specifically, the lesion label may be obtained by marking each training lesion image by a medical expert in advance.

In one practical application scenario of the present invention, because a large amount of noise pixels or interference information may exist in an acquired training focus image, in order to improve the accuracy of finally identifying a focus, image enhancement may be performed on the training focus image, where the image enhancement includes processing such as noise pixel elimination and texture enhancement.

In an embodiment of the present invention, the performing image enhancement on the training focus image to obtain an enhanced image includes:

In detail, referring to fig. 2, the following diagram is a schematic diagram of performing uniform segmentation on the training lesion image according to an embodiment of the present invention.

In fig. 2, the training lesion image of size 9X9 exists, and is divided into 9 lesion image blocks of 3X3 according to equal length and equal width.

In the embodiment of the invention, the training focus image is uniformly cut into a plurality of focus image blocks, which is beneficial to reducing the number of pixels in each focus image block, thereby improving the efficiency of eliminating noise pixels of the training focus image.

Specifically, in the embodiment of the present invention, a Gabor filter is adopted to perform pixel convolution on the plurality of focus image blocks, the Gabor filter performs convolution calculation on the plurality of focus image blocks according to a preset number of directions and a preset number of scales, only pixels that meet a preset standard are allowed to pass, and pixels that do not meet the filter are suppressed.

In the embodiment of the invention, a gaussian kernel function is used for performing gaussian smoothing on the plurality of convolution focus image blocks to obtain a plurality of smooth focus image blocks, the gaussian kernel function is also called as a radial basis function, and is a common smooth kernel function, and finite-dimension data (namely pixel values) can be smoothly mapped to a high-dimension space by using the rotational symmetry of the gaussian kernel function, so that the plurality of convolution focus image blocks can be subjected to gaussian smoothing.

The embodiment of the invention carries out noise pixel filtering on the training focus image, can inhibit the noise pixel in the training focus image, and is favorable for improving the accuracy of finally carrying out focus identification.

In another embodiment of the present invention, the image enhancement on the training focus image to obtain an enhanced image includes:

Optionally, the calculating a binary symbol of the central pixel of each image region by using a preset algorithm according to the central pixel of each image region and the neighborhood pixels of the central pixel includes:

calculating a binary symbol of a center pixel of the image area using the following algorithm

Wherein, P₀Is the central pixel of said image area, P_eIs the mean value of the neighborhood pixels of the central pixel, n is the number of the neighborhood pixels, s (P)₀-P_e) Is a quantization operation.

The embodiment of the invention performs detail enhancement processing on the training focus image, filters noise pixel points in the training focus image, and performs local texture deepening on image details, so that the detail characteristics in the image are highlighted, and the accuracy of finally performing focus identification is improved.

S2, extracting a characteristic region of the enhanced image, and performing image expansion on the enhanced image based on the characteristic region to obtain an expanded image set.

In one practical application scenario of the invention, because the number of the training medical images with higher quality is less, and the training medical images can be used for training only after focus marking is performed manually, in order to improve the training efficiency and reduce the process of manual marking, the number of the enhanced images obtained after the training images are subjected to image enhancement can be increased by performing image extension on the enhanced images so as to facilitate the subsequent training of a more accurate focus identification model by using a large number of extended image sets, and further improve the accuracy of final focus identification.

In the embodiment of the invention, the characteristic region of the enhanced image can be extracted, and the enhanced image is subjected to image expansion according to the characteristic region to obtain the expanded image set, so that the difference of the pixel distribution of each image in the expanded image set obtained by expansion and the enhanced image is favorably reduced, and the quality of each image in the expanded image set is improved.

In an embodiment of the present invention, the extracting the feature region of the enhanced image includes:

In detail, since the enhanced image includes a large amount of pixel information, but each pixel information is not the key information of the enhanced image, the enhanced image may be divided according to a preset ratio to divide the enhanced image into a plurality of enhanced image blocks, so that each enhanced image block is accurately analyzed subsequently, and the effect of image expansion of the enhanced image is improved.

Specifically, an image frame may be generated according to the preset size, and then the generated image frame is used to perform non-repetitive framing in the enhanced image to obtain a plurality of enhanced image blocks.

For example, if the length of the enhanced image is 10cm and the width of the enhanced image is 10cm, and the length of the image frame generated according to the preset size is 2cm and the width of the image frame is 2cm, 25 enhanced image blocks with the length of 2cm and the width of 2cm can be obtained by using the image frame to perform frame selection in the enhanced image.

Further, in order to perform a targeted analysis on each enhanced image block in the enhanced image, the image features corresponding to each enhanced image block in the plurality of enhanced image blocks may be extracted respectively.

In detail, the image features include global features and local features of each enhanced image block.

In one embodiment of the present invention, the global features of the target enhanced image block may be generated by using a Histogram of Oriented Gradients (HOG), a Discrete Part Model (DPM), a Local Binary Pattern (LBP), or the like, or may be extracted by using a pre-trained artificial intelligence Model with a specific image feature extraction function, where the artificial intelligence Model includes, but is not limited to, a VGG-net Model and a U-net Model.

In another embodiment of the present invention, the substrate is,

illustratively, the preset mapping function may be:

wherein, Y_iMapping the ith pixel point in the target enhanced image block to the pixel value, x, within the preset range_iAnd max (x) is the maximum pixel value in the target enhanced image block, and min (x) is the minimum pixel value in the target enhanced image block.

Further, a preset gradient algorithm can be used for calculating the pixel gradient of each row of pixels in the mapped target enhanced image block, wherein the gradient algorithm includes, but is not limited to, a two-dimensional discrete derivative algorithm, a cable operator and the like.

In the embodiment of the present application, the pixel gradient of each row of pixels may be converted into a row vector, and the row vector may be spliced into the global feature of the target enhanced image block.

For example, the selected target enhanced image block includes three rows of pixels, where the pixel gradient of the first row of pixels is q, w, e, the pixel gradient of the first row of pixels is a, s, d, and the pixel gradient of the first row of pixels is z, x, c, and then the pixel gradient of each row of pixels can be respectively used as a row vector to be spliced into the following global features:

further, the generating local features of the target enhanced image block according to the pixel values in each of the pixel windows includes:

selecting one pixel point from the pixel window one by one as a target pixel point;

judging whether the pixel value of the target pixel point is an extreme value in the pixel window;

when the pixel value of the target pixel point is not an extreme value in the pixel window, returning to the step of selecting one pixel point from the pixel window one by one as the target pixel point;

when the pixel value of the target pixel point is an extreme value in the pixel window, determining the target pixel point as a key point;

vectorizing the pixel values of all key points in all the pixel windows, and collecting the obtained vectors as the local features of the target enhanced image block.

In this embodiment of the application, the sliding window may be a pre-constructed selection box with a certain area, and may be used to perform frame selection on pixels in the target enhanced image block, for example, a square selection box constructed with 10 pixels as height and 10 pixels as width.

In detail, the extreme value includes a maximum value and a minimum value, and when the pixel value of the target pixel point is the maximum value or the minimum value in the pixel window, the target pixel point is determined to be the key point of the pixel window.

Specifically, the step of vectorizing the pixel values of all the key points in the pixel window is consistent with the step of calculating the pixel gradient of each line of pixels in the mapped target enhanced image block and converting the pixel gradient of each line of pixels into a line vector, which is not repeated again.

In an embodiment of the present invention, the performing image expansion on the enhanced image based on the feature region to obtain an expanded image set includes:

In detail, the enhanced image block corresponding to the target feature may be mapped to a corresponding position in the random lesion image in an image mapping manner, so as to generate an extended image of the enhanced image.

Specifically, the mapping may be a full mapping or a local mapping, for example, all image blocks corresponding to the target feature are mapped to corresponding positions in the random lesion image, or a part of image blocks corresponding to the target feature are randomly selected and the selected image blocks are mapped to corresponding positions in the random lesion image.

In the embodiment of the present invention, the position information refers to information of positions of selected extended image blocks in the enhanced image, and since the enhanced image and the random lesion image are both divided into a plurality of image blocks according to the same preset size, each extended image block can be mapped to a corresponding position of the random lesion image according to the position information of each extended image block in the enhanced image, that is, the extended image blocks are used to replace image blocks at corresponding positions in the random lesion image, so as to obtain the extended image.

Exemplarily, as shown in fig. 3, the enhanced image block includes an image block a, an image block B, an image block C, and an image block D, and the random lesion image includes an image block a, an image block B, an image block C, and an image block D, where the image block a and the image block D in the enhanced image block are extended image blocks, and the image block a and the image block D in the random lesion image may be replaced by the image block a and the image block D in the enhanced image, so as to obtain an extended image.

And S3, performing feature extraction on each expanded image in the expanded image set by utilizing multi-dimensional cross convolution to obtain image features corresponding to each expanded image.

In the embodiment of the invention, in order to obtain a lesion identification model with higher accuracy by training the extended image set, multidimensional cross convolution can be performed on each extended image in the extended image set so as to improve the accuracy of extracting image features from each extended image.

In this embodiment of the present invention, the extracting features of each expanded image in the expanded image set by using multidimensional cross convolution to obtain image features corresponding to each expanded image includes:

pooling the first convolution image to obtain an intermediate feature;

In detail, the first dimension convolution kernel and the second dimension convolution kernel may be multiple ones, and the first dimension convolution kernel and the second dimension convolution kernel are sequentially used in turn for convolution (i.e. multidimensional cross convolution).

For example, the first-dimension convolution kernel is a convolution kernel with a 3x3x3 dimension, the second-dimension convolution kernel is a convolution kernel with a 5x5x5 dimension, and the number of the first-dimension convolution kernel and the number of the second-dimension convolution kernel are both 2, the target image may be convolved by using a convolution kernel with a 3x3x3 dimension, the target image after the last convolution is convolved by using a convolution kernel with a 5x5x5 dimension, the target image after the last convolution is convolved by using a convolution kernel with a 3x3x3 dimension, and the target image after the last convolution is convolved by using a convolution kernel with a 5x5x5 dimension, so as to obtain a final convolution image.

In particular, the pooling process includes, but is not limited to, maximum pooling, average pooling, and the like.

And S4, performing iterative training on the pre-constructed focus recognition model by using the image characteristics to obtain the trained focus recognition model.

In the embodiment of the present invention, the pre-constructed lesion recognition model may be a network model such as an SVM (support vector machines), GoogleNet, and Alexnet.

In one embodiment of the present application, the training process of the region segmentation model includes:

performing convolution and pooling on each extended image in the extended image set by using the focus identification model to obtain the image characteristics of each extended image;

calculating a loss value between the image feature and the focus label by using a preset loss function;

and performing parameter optimization on the focus recognition model according to the loss value, and returning to the step of performing convolution and pooling on the training set by using a preset focus recognition model until the loss value is smaller than a preset loss threshold value to obtain the trained focus recognition model.

In detail, when the loss value is greater than or equal to the preset loss threshold, it indicates that the accuracy of the lesion recognition model is not sufficient, and more false judgments may occur, so that parameter optimization needs to be performed on the lesion recognition model to improve the accuracy of the lesion recognition model.

Specifically, according to the loss value, a preset gradient descent algorithm may be used to calculate an update gradient of a parameter in the lesion identification model, and then the parameter in the lesion identification model is adjusted according to the update gradient until the loss value is smaller than a preset loss threshold, where the gradient descent algorithm includes, but is not limited to, a small batch gradient descent algorithm, a batch gradient descent algorithm, and a random gradient descent algorithm.

And S5, acquiring a focus image to be recognized, and performing focus recognition on the focus image to be recognized by using the trained focus recognition model to obtain a focus type corresponding to the focus image to be recognized.

In the embodiment of the invention, the focus image to be recognized can be uploaded by a user in advance, and then the focus image to be recognized can be input to the trained focus recognition model, so that the trained focus recognition model is utilized to perform focus recognition on the focus image to be recognized, and the focus type corresponding to the focus image to be recognized is obtained.

According to the embodiment of the invention, the image enhancement and the image expansion operation are carried out on the training focus image, so that the image quality of the training focus image is improved, the number of the training focus images is also expanded, and the method is favorable for improving the effect of training a more accurate focus recognition model by utilizing the enhanced and expanded image; meanwhile, feature extraction is carried out on the extended image by utilizing multi-dimensional cross convolution, iterative training is carried out on a focus recognition model which is constructed in advance by utilizing the extracted image features, and the accuracy of the extracted image features is improved, so that the accuracy of the trained focus recognition model is improved. Therefore, the focus identification method provided by the invention can solve the problem of low accuracy of focus type identification.

Fig. 4 is a functional block diagram of a lesion recognition apparatus according to an embodiment of the present invention.

The lesion recognition apparatus 100 according to the present invention may be installed in an electronic device. According to the implemented functions, the lesion recognition apparatus 100 may include an image enhancement module 101, an image expansion module 102, a feature extraction module 103, a model training module 104, and a lesion recognition module 105. The module of the present invention, which may also be referred to as a unit, refers to a series of computer program segments that can be executed by a processor of an electronic device and can perform a fixed function, and are stored in a memory of the electronic device.

In the present embodiment, the functions regarding the respective modules/units are as follows:

the image enhancement module 101 is configured to obtain a training focus image and a focus label corresponding to the training focus image, and perform image enhancement on the training focus image to obtain an enhanced image;

the image expansion module 102 is configured to extract a feature region of the enhanced image, and perform image expansion on the enhanced image based on the feature region to obtain an expanded image set;

the feature extraction module 103 is configured to perform feature extraction on each expanded image in the expanded image set by using multi-dimensional cross convolution to obtain an image feature corresponding to each expanded image;

the model training module 104 is configured to perform iterative training on a pre-constructed focus recognition model by using the image features to obtain a trained focus recognition model;

the lesion identification module 105 is configured to obtain a lesion image to be identified, and perform lesion identification on the lesion image to be identified by using the trained lesion identification model to obtain a lesion type corresponding to the lesion image to be identified.

In detail, each module in the lesion recognition device 100 according to the embodiment of the present invention adopts the same technical means as the lesion recognition method described in fig. 1 to 3, and can produce the same technical effect, and will not be described herein again.

Fig. 5 is a schematic structural diagram of an electronic device for implementing a lesion recognition method according to an embodiment of the present invention.

The electronic device 1 may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may further include a computer program, such as a lesion recognition program, stored in the memory 11 and executable on the processor 10.

In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or may be composed of a plurality of integrated circuits packaged with the same function or different functions, and includes one or more Central Processing Units (CPUs), microprocessors, digital Processing chips, graphics processors, and combinations of various control chips. The processor 10 is a Control Unit (Control Unit) of the electronic device, connects various components of the electronic device by using various interfaces and lines, and executes various functions and processes data of the electronic device by operating or executing programs or modules (e.g., executing a lesion recognition program, etc.) stored in the memory 11 and calling data stored in the memory 11.

The memory 11 includes at least one type of readable storage medium including flash memory, removable hard disks, multimedia cards, card-type memory (e.g., SD or DX memory, etc.), magnetic memory, magnetic disks, optical disks, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device, for example a removable hard disk of the electronic device. The memory 11 may also be an external storage device of the electronic device in other embodiments, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, provided on the electronic device. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device. The memory 11 may be used not only to store application software installed in the electronic device and various types of data, such as codes of a lesion recognition program, etc., but also to temporarily store data that has been output or will be output.

The communication bus 12 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is arranged to enable connection communication between the memory 11 and at least one processor 10 or the like.

The communication interface 13 is used for communication between the electronic device and other devices, and includes a network interface and a user interface. Optionally, the network interface may include a wired interface and/or a wireless interface (e.g., WI-FI interface, bluetooth interface, etc.), which are typically used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a Display (Display), an input unit, such as a Keyboard (Keyboard), and optionally a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, or the like. The display, which may also be referred to as a display screen or display unit, is suitable, among other things, for displaying information processed in the electronic device and for displaying a visualized user interface.

Only electronic devices having components are shown, and those skilled in the art will appreciate that the structures shown in the figures do not constitute limitations on the electronic devices, and may include fewer or more components than shown, or some components in combination, or a different arrangement of components.

For example, although not shown, the electronic device may further include a power supply (such as a battery) for supplying power to each component, and preferably, the power supply may be logically connected to the at least one processor 10 through a power management device, so that functions such as charge management, discharge management, and power consumption management are implemented through the power management device. The power supply may also include any component of one or more dc or ac power sources, recharging devices, power failure detection circuitry, power converters or inverters, power status indicators, and the like. The electronic device may further include various sensors, a bluetooth module, a Wi-Fi module, and the like, which are not described herein again.

It is to be understood that the embodiments described are illustrative only and are not to be construed as limiting the scope of the claims.

The lesion recognition program stored in the memory 11 of the electronic device 1 is a combination of instructions that, when executed in the processor 10, may implement:

Specifically, the specific implementation method of the instruction by the processor 10 may refer to the description of the relevant steps in the embodiment corresponding to the drawings, which is not described herein again.

Further, the integrated modules/units of the electronic device 1, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. The computer readable storage medium may be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying said computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM).

The present invention also provides a computer-readable storage medium, storing a computer program which, when executed by a processor of an electronic device, may implement:

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.

The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.

In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.

It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof.

The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.

The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.

The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units or means recited in the system claims may also be implemented by one unit or means in software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.

Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method of lesion identification, the method comprising:

2. The lesion recognition method of claim 1, wherein the extracting the feature region of the enhanced image comprises:

3. The lesion identification method according to claim 2, wherein the image expansion of the enhanced image based on the feature region to obtain an expanded image set comprises:

4. The lesion identification method of claim 2, wherein the generating global features of the target enhanced image patch from pixel gradients in the target enhanced image patch comprises:

5. The lesion identification method according to claim 1, wherein the performing feature extraction on each expanded image in the expanded image set by using multi-dimensional cross convolution to obtain an image feature corresponding to each expanded image comprises:

pooling the first convolution image to obtain an intermediate feature;

6. The lesion recognition method of any one of claims 1 to 5, wherein the performing image enhancement on the training lesion image to obtain an enhanced image comprises:

respectively carrying out pixel convolution on the plurality of focus image blocks to obtain a plurality of convolution focus image blocks;

respectively carrying out Gaussian smoothing treatment on the plurality of convolution focus image blocks to obtain a plurality of smooth focus image blocks;

7. The lesion recognition method of any one of claims 1 to 5, wherein the image enhancement of the training lesion image to obtain an enhanced image comprises:

8. A lesion recognition device, the device comprising:

the image enhancement module is used for acquiring a training focus image and a focus label corresponding to the training focus image, and performing image enhancement on the training focus image to obtain an enhanced image;

the image expansion module is used for extracting a characteristic region of the enhanced image and carrying out image expansion on the enhanced image based on the characteristic region to obtain an expanded image set;

the feature extraction module is used for extracting features of each expanded image in the expanded image set by utilizing multi-dimensional cross convolution to obtain image features corresponding to each expanded image;

the model training module is used for carrying out iterative training on a pre-constructed focus identification model by utilizing the image characteristics to obtain a trained focus identification model;

and the focus identification module is used for acquiring a focus image to be identified, and performing focus identification on the focus image to be identified by using the trained focus identification model to obtain a focus type corresponding to the focus image to be identified.

9. An electronic device, characterized in that the electronic device comprises:

at least one processor; and (c) a second step of,

the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the lesion identification method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements a lesion identification method according to any one of claims 1 to 7.