CN113553460B

CN113553460B - Image retrieval method and device, electronic device and storage medium

Info

Publication number: CN113553460B
Application number: CN202110931833.XA
Authority: CN
Inventors: 郑御; 曾韦胜; 张良; 吴振洲
Original assignee: Beijing Ande Yizhi Technology Co ltd
Current assignee: Beijing Ande Yizhi Technology Co ltd
Priority date: 2021-08-13
Filing date: 2021-08-13
Publication date: 2022-06-28
Anticipated expiration: 2041-08-13
Also published as: CN113553460A

Abstract

The disclosure relates to an image retrieval method and apparatus, an electronic device and a storage medium, including determining a first feature characterizing a human tissue abnormality in a medical image to be retrieved; according to the first feature, a target first image matched with the first feature is searched in a search library; the target first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the target first image; and displaying the target first image and the remark information. According to the image retrieval method disclosed by the embodiment of the disclosure, the image matched with the medical image to be retrieved can be effectively retrieved, and the accuracy and efficiency of interpreting the medical image to be retrieved by a doctor are further improved.

Description

Image retrieval method and device, electronic device and storage medium

Technical Field

The present disclosure relates to the field of data processing, and in particular, to an image retrieval method.

Background

The medical image is: images of internal tissue properties of a human body or a part of a human body are obtained in a non-invasive manner for medical treatment or medical research. Generally, medical images interact with the human body by means of some medium (such as X-rays, electromagnetic fields, ultrasonic waves, etc.), and the information of the internal tissue organ structure, density, tissue composition, etc. of the human body is represented in an image mode.

However, because of the wide variety of diseases and the different clinical responses of the same disease at different locations, the treatment modalities and prognosis are quite different. This requires that the doctor be able to make an accurate interpretation of the medical image to make an accurate determination of the disease.

Generally, a doctor identifies a medical image of a patient according to his or her own medical experience to determine whether an abnormality occurs at a certain part of the patient, which is likely to cause misdiagnosis or delay treatment time due to irregular interpretation levels.

Disclosure of Invention

In view of this, the present disclosure provides an image retrieval method.

According to an aspect of the present disclosure, there is provided an image retrieval method, including:

determining a first characteristic representing human tissue abnormality in a medical image to be retrieved;

according to the first feature, a target first image matched with the first feature is searched in a search library; the target first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the target first image;

and displaying the target first image and the remark information.

In one possible implementation, the remark information includes at least one of:

Description of human tissue abnormality, position of human anatomy structure where the human tissue abnormality is located, medical image measurement data, and human tissue abnormality region in the medical image.

In one possible implementation, determining a first feature characterizing a human tissue abnormality in a medical image to be retrieved includes:

determining a segmentation mask indicating a human body tissue abnormal region in the medical image to be retrieved according to the medical image to be retrieved;

and taking the characteristic of the abnormal human tissue area in the medical image to be retrieved, which is indicated by the segmentation mask, as the first characteristic.

In one possible implementation manner, the retrieving, according to the first feature, a target first image matching the first feature in a search library includes:

determining a positioning mask indicating the position of a human anatomy structure in a medical image to be retrieved;

searching a first image in the search library according to the positioning mask and the segmentation mask;

and determining a target first image with the matching degree meeting preset conditions from the first image according to the matching degree of the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image.

In one possible implementation manner, the retrieving a first image from the search library according to the localization mask and the segmentation mask includes:

fusing the positioning mask and the segmentation mask to determine the position information of the human tissue abnormality in the medical image to be retrieved, wherein the position information represents the position of the human anatomy structure where the human tissue abnormality is located;

and determining the image with the same position information in the search library as a first image.

In a possible implementation manner, the determining, according to the degree of matching between the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image, a target first image from the first image whose degree of matching satisfies a preset condition includes:

according to the similarity between the features, determining a first image with the similarity between the first feature and the second feature higher than a similarity threshold value as a target first image; the second characteristic is a characteristic representing the human tissue abnormality in the first image.

In a possible implementation manner, the determining, according to a matching degree between the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image, a target first image from the first images, where the matching degree satisfies a preset condition, includes:

Determining a first image as a second image according to the range metric value of the abnormal human tissue region, wherein the difference value between the first range metric value and the second range metric value is smaller than a difference threshold value; the first range metric value is a metric value of the range of the abnormal region of the human tissue indicated by the segmentation mask; the second range measurement value is the measurement value of the range of the human body tissue abnormal region in the first image;

according to the similarity between the features, determining a second image with the similarity between the first feature and the third feature higher than the similarity threshold value as a target first image; the third characteristic is a characteristic representing the human tissue abnormality in the second image.

According to another aspect of the present disclosure, there is provided an image retrieval apparatus including:

the first characteristic determining unit is used for determining a first characteristic representing human tissue abnormality in the medical image to be retrieved;

the retrieval unit is used for retrieving a target first image matched with the first characteristic from a retrieval library according to the first characteristic; the first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the first image;

and the display unit is used for displaying the target first image and the remark information.

In one possible implementation manner, the first feature determining unit includes:

the segmentation mask determining unit is used for determining a segmentation mask indicating a human body tissue abnormal region in the medical image to be retrieved according to the medical image to be retrieved;

and the first feature determination subunit is configured to use, as the first feature, a feature of the abnormal human tissue region in the medical image to be retrieved, which is indicated by the segmentation mask.

In a possible implementation manner, the retrieving unit includes:

the positioning mask determining unit is used for determining a positioning mask indicating the position of the human anatomy structure in the medical image to be retrieved;

a first image retrieval unit, configured to retrieve a first image from the search library according to the positioning mask and the segmentation mask;

and the target first image retrieval unit is used for determining a target first image of which the matching degree meets a preset condition from the first image according to the matching degree of the human tissue abnormal region indicated by the segmentation mask and the human tissue abnormal region in the first image.

In one possible implementation manner, the first image retrieving unit includes:

the position information determining unit is used for fusing the positioning mask and the segmentation mask to determine the position information of the human tissue abnormity in the medical image to be retrieved, and the position information represents the position of the human anatomy structure where the human tissue abnormity is located;

and the first image searching subunit is used for determining the images with the same position information in the searching library as the first images.

In one possible implementation manner, the target first image retrieving unit includes:

the target first image first retrieval subunit is used for determining a first image, as a target first image, of which the similarity between the first feature and the second feature is higher than a similarity threshold value according to the similarity between the features; the second characteristic is a characteristic representing human tissue abnormality in the first image.

the second image retrieval unit is used for determining a first image of which the difference value between the first range metric value and the second range metric value is smaller than a difference threshold value as a second image according to the range metric value of the human tissue abnormal region; the first range metric value is a metric value of the range of the abnormal region of the human tissue indicated by the segmentation mask; the second range metric value is the metric value of the range of the human body tissue abnormal region in the first image;

The target first image second retrieval subunit is used for determining a second image, as a target first image, of which the similarity between the first feature and the third feature is higher than the similarity threshold value according to the similarity between the features; the third characteristic is a characteristic representing human tissue abnormality in the second image.

According to an aspect of the present disclosure, there is provided an electronic device including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the memory-stored instructions to perform the above-described method.

According to an aspect of the present disclosure, there is provided a computer readable storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement the method of any one of claims 1 to 7.

In the embodiment of the disclosure, a first feature representing human tissue abnormality is determined in a medical image to be retrieved, and a target first image matched with the first feature is retrieved in a retrieval library by using the first feature. The doctor interprets the medical image to be retrieved according to remark information such as description of the human tissue abnormality in the target first image, the position of the human anatomical structure where the human tissue abnormality is located, measurement data of the medical image, the abnormal region of the human tissue in the medical image and the target first image, so that interpretation efficiency is improved, and interpretation accuracy is improved due to the fact that the target first image is used as a reference.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

Fig. 1 shows a flowchart of an image retrieval method according to an embodiment of the disclosure.

Fig. 2 shows a schematic diagram of a medical image to be retrieved, a segmentation mask, according to an embodiment of the present disclosure.

Fig. 3 illustrates a flow diagram for generating a segmentation mask according to an embodiment of the disclosure.

Fig. 4 shows a flowchart of extracting a first feature according to an embodiment of the present disclosure.

Fig. 5 shows a schematic diagram of a medical image to be retrieved, a localization mask, according to an embodiment of the present disclosure.

Fig. 6 illustrates a flow diagram for generating a localization mask according to an embodiment of the present disclosure.

Fig. 7 shows a schematic diagram of human tissue abnormality location information according to an embodiment of the present disclosure.

Fig. 8 illustrates a flow chart of a process of generating a segmentation mask, a localization mask, a first feature, and location information according to an embodiment of the disclosure.

Fig. 9 shows a flowchart of extracting a second feature according to an embodiment of the present disclosure.

Fig. 10 shows a block diagram of an image retrieval device according to an embodiment of the present disclosure.

FIG. 11 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure.

Fig. 12 shows a block diagram of an electronic device according to an embodiment of the disclosure.

Detailed Description

Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present disclosure.

Generally, medical images interact with the human body by means of some medium (such as X-rays, electromagnetic fields, ultrasonic waves, etc.), and the information of the internal tissue organ structure, density, tissue composition, etc. of the human body is represented in an image mode. Thus, the doctor can recognize the abnormality of the human body part according to the medical image.

However, because of the wide variety of diseases and the different clinical responses of the same disease at different locations, the treatment modalities and prognosis are quite different. Such as epidermoid cyst of brain, in the pono-cerebellar horn, manifested by trigeminal neuralgia or as pono-cerebellar horn syndrome, with less impairment of hearing and vestibular function; and in the surface skin cyst of the saddle area, the vision cross compression expressions such as primary optic nerve atrophy, bistotemporal hemianopsia and the like can appear: the cranial flap butterfly saddle is normal in size, but the optic nerve hole, the optic cross groove and the like can be damaged in a limited way. For cysts with light adhesion to surrounding tissues, especially cysts of the fourth ventricle, total resection is expected. For cysts with severe adhesion to blood vessels and other important structures, a part of the cyst wall can be left unless the physician is conscious. Therefore, doctors are required to be able to make accurate interpretation of medical images to make accurate judgment of diseases.

Generally, doctors identify medical images of patients according to their own medical experiences to determine whether an abnormality occurs in a certain part of the patients, which is likely to cause misdiagnosis or delay treatment time due to uneven interpretation levels.

Therefore, the embodiment of the present disclosure provides an image retrieval method, which can retrieve, in a retrieval library, an influence matched with an abnormal region of a human tissue in a medical image to be retrieved. So that the doctor can obtain the reference image and the related information of the diseases reflected in the image when the image to be searched is read. The doctor is helped to judge and read the images, and the efficiency and the accuracy of image judgment and reading of the doctor are improved.

The image retrieval method of the embodiment of the disclosure may be applied to a certain electronic device, and the certain electronic device may be a server or a terminal device that runs the image retrieval method of the disclosure. The terminal device may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, or the like, which is not limited by the present disclosure.

Fig. 1 shows a flowchart of an image retrieval method according to an embodiment of the disclosure. As shown in fig. 1, the flowchart includes:

in step S11, a first feature characterizing the human tissue abnormality is determined in the medical image to be retrieved.

The electronic device receives the medical image to be retrieved. The medical image may be one of an ultrasound image, a Computed Tomography (CT) image, and a magnetic resonance image that reflects a target object such as a human body or any part of the human body.

Then, the electronic device identifies the human tissue abnormality in the medical image to be retrieved, and divides one or more human tissue abnormal regions, wherein the human tissue abnormal regions are composed of pixels representing human tissue abnormality.

And then, extracting features in the abnormal human tissue region of the medical image to be retrieved to obtain a first feature representing the abnormal human tissue. The first feature may include a plurality of categories, such as: color, gray scale, texture, shape, spatial relationship between various human tissue abnormalities, and the like, to which the present disclosure does not limit.

In step S12, a target first image matching the first feature is retrieved from a search library based on the first feature; the target first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the target first image.

The electronic device searches in a search library for storing medical images similar to the medical image to be searched according to the first feature. And obtaining a first image of the target containing the characteristic matched with the first characteristic through retrieval. The characteristic matching process can match the first characteristics of one or more categories of color, gray scale, texture and the like displayed in the image to be retrieved by the human tissue abnormity with the characteristics representing the human tissue abnormity on the image in the retrieval library. The matching can be performed by using the same kind of first features, and the matching can also be performed by using the combination of the first features of a plurality of categories.

For example, for diseases in which human tissue abnormalities tend to exhibit symmetry, feature matching may be performed according to the spatial relationship between human tissue abnormalities. For example, two abnormal regions of human tissue appear in the brain image of the patient to be retrieved, and the two abnormal regions of human tissue are symmetric to each other, which can be used as the first feature to search in the search library, and the image of the symmetric abnormal region of human tissue can be retrieved.

For example, the first features such as gray scale, texture, shape, etc. may be combined for feature matching. For example, according to the shape of the human tissue abnormality in the medical image to be searched, an image containing the human tissue abnormality having a similar shape is searched in the search library. Then, the texture and grayscale are continuously compared. Until the image with the shape, texture and gray level similar to the human body tissue abnormity in the medical image to be searched is searched.

Through feature matching, the matching degree of the first feature and the feature of the image in the search library can be obtained. When the matching degree falls within a preset threshold value range, the image stored in the search library is indicated to have the characteristic similar to the first characteristic. Then the image is taken as the target first image.

The target first image has remark information, and the remark information and the target first image are stored in the search library together. The remark information here is related information describing a disease corresponding to a human tissue abnormality expressed in the target first image. The information such as the type of the disease, the description of the focus, the symptom of the corresponding patient, the focus area range and the like corresponding to the target first image can be known through the remark information.

Further, the remark information includes at least one of: description of human tissue abnormality, position of human anatomy structure where the human tissue abnormality is located, medical image measurement data, and human tissue abnormality region in the medical image.

The description of the human tissue abnormality can be the description of the shape, size, physiological structure, detailed position in the human body of the human tissue abnormality through medical images, and data obtained by a matched test.

The disclosed embodiments can divide various parts of the human body according to the human anatomy, and perform further detailed division on the various parts. For example: the human body comprises lower limbs, upper limbs, spine region, head, neck, chest, abdomen, and pelvis. And can be subdivided into brain, five sense organs, bone, muscle tissue, etc., for the head, each of which can be further subdivided; taking the brain as an example, the method can be divided into: left frontal lobe, right frontal lobe, left cerebellar hemisphere, right cerebellar hemisphere, third ventricle, fourth ventricle, pontine, left cerebral foot, right cerebral foot, left thalamus, right thalamus, left lateral ventricle, right lateral ventricle, and the like.

After the human body is divided according to the human anatomy structure, the human body is divided into a plurality of parts, each part corresponds to one human anatomy structure, and all parts form a complete human body. And if the area for representing the human tissue abnormality in the medical image to be retrieved corresponds to a human anatomy structure, the human anatomy structure is the position of the human anatomy structure where the human tissue abnormality is located.

The medical image measurement data may be data representing a local state of the human body obtained from each medical image. For example, the CT value in the CT image, the blood flow velocity and the blood flow direction in the doppler ultrasound image, and the like.

The abnormal human tissue region in the medical image may be a region formed by pixels in the medical image to be retrieved, the pixels in the region being used for representing the abnormal human tissue, and the boundary of the region being a boundary of the abnormal human tissue.

In step S13, the target first picture and the memo information are displayed.

After the target first image is retrieved in step S12, the electronic device displays the target first image and the remark information of the target first image, which may be displayed on a screen or displayed by projection, and the embodiment of the disclosure is not limited thereto. Thus, a doctor or other user can read and refer to the first image and the remark information thereof.

First characteristics which represent human tissue abnormity are determined in the medical image to be retrieved, and the first characteristics are used for retrieving a target first image matched with the first characteristics from a retrieval base. The doctor interprets the medical image to be retrieved according to the remark information such as the description of the human tissue abnormality in the target first image, the position of the human anatomy structure where the human tissue abnormality is located, the measurement data of the medical image, the abnormal region of the human tissue in the medical image and the target first image, so that the interpretation efficiency is improved, and the interpretation accuracy is improved by using the target first image as a reference.

In one possible implementation manner, the determining, by step S11, a first feature characterizing the human tissue abnormality in the medical image to be retrieved includes: determining a segmentation mask indicating a human body tissue abnormal region in the medical image to be retrieved according to the medical image to be retrieved; and taking the characteristic of the abnormal human tissue area in the medical image to be retrieved, which is indicated by the segmentation mask, as the first characteristic.

Fig. 2 shows a medical image to be retrieved, a segmentation mask image and a segmentation mask schematic diagram according to an embodiment of the disclosure.

As shown in fig. 2(a), after receiving a medical image (201) to be retrieved, the electronic device identifies pixels in the medical image (201) that represent abnormalities in human tissue. The medical image (201) to be retrieved is taken in relation to a target object (202). As shown in fig. 2(b), the target object (202) is divided, and the identified abnormal region (203) of the human tissue is represented by white in fig. 2(b) by assigning 1 to the pixel value of the pixel representing the abnormal human tissue; in the medical image, the pixel value of the pixel outside the human tissue abnormal region is given 0, and is shown in gray in fig. 2 (b). The image after the division processing is a division mask image (204). The size and the pixel size of the segmentation mask image are the same as those of the medical image to be retrieved.

A region composed of pixels having a pixel value of 1 in the division mask image (204) constitutes a division mask (205). A segmentation mask (205) is used to characterize the extent of the abnormal region of human tissue. Furthermore, in the medical image (201) to be retrieved, features within a region corresponding to the position of the segmentation mask (205) in the segmentation mask image (204) are determined. The feature is used as an image feature of the abnormal region of the human tissue, namely a first feature.

As shown in fig. 3, a process of generating a segmentation mask according to an embodiment of the present disclosure: a nuclear magnetic resonance image of which the size of the medical image to be retrieved is 256 (length) × 256 (width) × 24 (image layer); carrying out convolution downsampling operation on the medical image to be retrieved for 4 times by using an encoder, and outputting a 128 x 24 image to an encoder 2 after the convolution operation of the encoder 1; after the convolution operation of the encoder 2 is completed, outputting a 64 × 64 × 12 image to the encoder 3; outputting the 32 × 32 × 12 image to the encoder 4 after the convolution operation of the encoder 3; after the encoder 4 performs the convolution operation, another encoder (or a decoder) is used to schematically display the image by the encoder 5 in fig. 3, the image is taken as an intermediate layer to continue feature extraction, and then the decoder is used to perform 4 times of convolution upsampling operations, wherein the size and the number of layers of the output image after each decoder is convolved are the same as the size and the number of layers of the input image of the encoder corresponding to the decoder. For example, a 32 × 32 × 12 video is output through the processing of the decoder 1, and the size and layer of the video input to the encoder are the same. Thus, until the decoder 4 finishes processing the image, the segmentation mask of the human tissue abnormal region is obtained.

Fig. 4 shows a flow of extracting a first feature from a medical image to be retrieved according to an embodiment of the present disclosure: the medical image to be retrieved and the segmentation mask corresponding thereto are input to a decoder. And performing convolution operation on the medical image to be retrieved in the segmentation mask range by a decoder. In fig. 4, there are 5 decoders altogether, resulting in 5 convolution results. In practical applications, the number of decoders can be determined according to practical needs. The specific operation, input, and output of the decoder may also obtain the first characteristic representing the human tissue abnormality according to each convolution result, which is not described herein again.

The human tissue abnormal region in the medical image to be retrieved is segmented, so that the position and the range of the human tissue abnormal region in the medical image to be retrieved can be determined, the range of extracting the first feature is narrowed from the whole medical image to be retrieved to the range indicated by the segmentation mask, and the efficiency and the accuracy of extracting the first feature are improved.

In one possible implementation, a second neural network may be used to segment regions of abnormal human tissue in the medical image to be retrieved. The second neural network may be any kind of image segmentation neural network.

Preferably, the second neural network may be a Unet (unity networking) neural network.

The second neural network outputs a minimized segmentation loss parameter L of a segmentation mask generation process in the process of segmenting the abnormal region of the human tissue_SEG. In the current segmentation process, L is used_SEGThe parameters of the second neural network are adjusted so that the segmentation mask generated in the next round has higher precision.

Preferably, the loss function may use dice loss, and its calculation formula is detailed in formula (1).

Wherein L is_SEGA minimum segmentation loss parameter for a segmentation mask generation process; omega_C1The training scale of the channel c1 is (0, 1);

Is a binary mask model;

is a soft prediction mask model.

In one possible implementation, a first neural network may be used for a first feature extraction of the medical image to be retrieved. The first feature network may be any kind of image feature extraction neural network. The first neural network outputs the loss parameter L of the inter-class maximization and the intra-class minimization of the characteristic generation process in the first characteristic generation process_triplet. In the first feature generation process, L is used_tripletTo adjust the firstAnd parameters of the neural network enable the first feature generated in the next round to be more accurate.

Preferably, the loss function can be expressed as a triplet loss, and the calculation formula is detailed in formula (2).

Wherein L is_tripletA minimum distance loss parameter for the signature sequence generation process; | x | is the euclidean distance;

euclidean distance measurement is carried out on homogeneous data;

euclidean distance measurement is carried out on non-homogeneous data; alpha is the minimum interval between the Euclidean distance of the same kind data and the Euclidean distance of the different kind data, and + represents [ [ alpha ] ]]When the internal value is more than 0, taking the loss; when less than 0, the loss is 0; i is the ith convolution operation.

The neural network is used for carrying out human tissue abnormal region segmentation and feature extraction on the medical image to be retrieved, so that the efficiency of feature extraction can be improved. Parameters of the neural network are adjusted by using the loss function, so that the aim of training the neural network is fulfilled, and the accuracy of the neural network is improved. As the accuracy of the neural network is improved, the accuracy of feature extraction will also be improved.

In one possible implementation manner, the step S12 of retrieving, according to the first feature, a target first image matching the first feature from a search library includes: determining a positioning mask indicating the position of a human anatomy structure in a medical image to be retrieved; and determining a target first image of which the matching degree meets a preset condition from the first image according to the matching degree of the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image.

The electronic device can identify the human anatomy structure of the target object such as the human body or the local part of the human body in the medical image to be retrieved according to the human anatomy structure, so that the target object in the medical image is divided according to the human anatomy structure.

Taking a human brain nuclear magnetic image as an image to be retrieved as an example, fig. 5 shows a schematic diagram of a localization mask according to an embodiment of the disclosure.

The electronic device divides an image area representing a human brain on an image to be retrieved shown in fig. 5(a) according to a human anatomy structure. As shown in fig. 5(b), the image region representing the human brain is divided into 6 parts according to the human anatomy structure, each part is represented by a fill line type for distinguishing the parts, and the divided parts correspond to the corresponding human anatomy structure.

By identifying and dividing the human anatomy structure of the target object in the medical image to be retrieved, a localization mask image (503) indicating the position of the human anatomy structure is obtained.

In practical applications, the portions may be distinguished by different pixel values. Thus, the same human anatomy node is assigned with the same pixel value on the localization mask image (503), and pixels with the same pixel value representing a certain human anatomy structure constitute the localization mask (502-507).

For example, in a human brain nuclear magnetic resonance image, a localization mask image (501) corresponding to the human brain nuclear magnetic resonance image is configured by assigning a pixel value of a pixel representing the right temporal lobe to 20 (dot filling in fig. 5), assigning a pixel value of a pixel representing the left temporal lobe to 21 (diagonal filling in fig. 5), and the like. Then in the localization mask image (501), pixels with a pixel value of 20 constitute the right temporal lobe (504); similarly, a pixel having a pixel value of 21 constitutes the left temporal lobe (507).

The positioning mask represents the position and the range of each human anatomy structure in the medical image to be retrieved after the medical image to be retrieved is divided according to the human anatomy structure.

Since the positioning mask image is generated based on the medical image to be retrieved, the positioning mask image is consistent with the size and pixel size of the corresponding medical image.

Fig. 6 shows a process of generating a location mask according to an embodiment of the present disclosure: and performing 4 times of convolution downsampling operation on the medical image to be retrieved by using an encoder, performing continuous feature extraction on the image by using another encoder (or a decoder) as an intermediate layer, and performing 4 times of convolution upsampling operation by using the decoder to obtain a positioning mask representing the human anatomy structure of the target object in the medical image to be retrieved. The input, output and operation of the encoder and decoder are not described in detail. In practical applications, the number of decoders and encoders can be adjusted as needed.

In a possible implementation manner, the third neural network may be used to perform human anatomy structure recognition and division on a target object in the medical image to be retrieved, so as to generate a localization mask. The third neural network may be any image segmentation neural network.

Preferably, the third neural network may be a Unet (unity networking) neural network.

The third neural network outputs the minimum segmentation loss parameter L of the process in the process of identifying and dividing the human anatomy structure_altas. In the current segmentation process, L is used_altasTo adjust parameters of the third neural network. Preferably, the loss function may be expressed as dice loss. L is _altasSee the formula (3)

Wherein L is_altasA segment loss minimization parameter for a segmentation mask generation process; omega_C2The training scale for channel C2, the span of values is (0, 1);

is a binary mask model;

is a soft prediction mask model.

The electronic device can search in a search library according to the positioning mask and the segmentation mask corresponding to the medical image to be searched. And searching the first image in which the human anatomy structure position where the human tissue abnormality is located and the range of the human tissue abnormality region in the search library both accord with preset standards. The first image includes an abnormal region of human tissue.

For example, the preset criteria for the position of the human anatomy structure where the abnormal region of the human tissue is located are: the positions are the same; in the human brain nuclear magnetic resonance image, the human tissue abnormality is located in the pons, and the image area representing the pons in the retrieved first image contains the human tissue abnormality.

Then, the electronic device compares the human tissue abnormality in the area represented by the segmentation mask with the human tissue abnormality in the first image to obtain the matching degree of the human tissue abnormality and the human tissue abnormality in the first image. And taking the first image with the matching degree meeting the preset condition as a target first image to finish the image retrieval.

The segmentation mask and the positioning mask corresponding to the medical image to be retrieved are used for retrieving the first image from the retrieval base, and the purpose of further narrowing the retrieval range is achieved. And further determining a target first image matched with the medical image to be retrieved according to the matching degree of the abnormal human tissue area in the medical image to be retrieved and the abnormal human tissue area in the first image. Therefore, the retrieval range is gradually reduced through step-by-step retrieval, the target first image is gradually locked, mutual interference caused by excessive retrieval according to elements is avoided, and the retrieval accuracy is improved.

In one possible implementation, the retrieving a first image from the search library according to the location mask and the segmentation mask includes: fusing the positioning mask and the segmentation mask to determine the position information of the human tissue abnormality in the medical image to be retrieved, wherein the position information represents the position of the human anatomy structure where the human tissue abnormality is located; and determining the image with the same position information in the search library as a first image.

Fig. 7 illustrates a schematic diagram of human tissue abnormality location information according to an embodiment of the present disclosure.

The electronic device can keep the segmentation mask image and the positioning mask image corresponding to the medical image to be retrieved in the same direction as the medical image to be retrieved for superposition, and multiply the pixel values of the segmentation mask image and the pixel values of the pixel points corresponding to the positioning mask to complete the fusion processing of the segmentation mask image and the positioning mask image, so as to obtain a fused image (701). The fused image (701) retains the pixel values of the localization mask at the position of the abnormal human tissue range within the range of the segmentation mask (fig. 7 continues to show the pixel values representing the human anatomy structure in a filled line type, and fills the localization mask with the line type of the human anatomy structure at the same position in fig. 5), and the pixel values of the region on the fused image except the range of the segmentation mask are 0, which is shown in gray in fig. 7. In this way, the position of the human anatomy structure where the human tissue abnormality is located is obtained, i.e. position information (702) of the human tissue abnormality is obtained, the position information (702) comprising the aforementioned pixel values representing the localization mask.

The position information of the human tissue abnormality is used as a search element to search in a search library. And taking the image with the same position information as a first image.

In actual operation, the position of the human anatomy structure where the human tissue abnormality is located in the remark information of the medical image in the search library is stored in a character form, the position of the human anatomy structure corresponds to a number, and the number is consistent with the pixel value of the positioning mask corresponding to the human anatomy structure. The electronic equipment extracts non-zero pixel values after fusing the segmentation mask image and the positioning mask image, and searches in a search library according to the non-zero pixel values, namely, the aim of searching by using the position information of the human tissue abnormity is fulfilled.

Fig. 8 shows a flow of generating a segmentation mask, a location mask, a first feature, and location information according to an embodiment of the present disclosure.

After receiving the medical image to be retrieved, the electronic device inputs the medical image to be retrieved into a second neural network (402) and a third neural network (403). Obtaining a segmentation mask corresponding to the medical image to be retrieved through processing of a second neural network (402); and obtaining a positioning mask corresponding to the medical image to be retrieved through the processing of the third neural network (403). And then, acquiring the position information of the human tissue abnormality in the medical image to be retrieved through the fusion processing of the segmentation mask and the positioning mask. The first neural network (401) extracts the influence characteristics of the human tissue abnormity by using the medical image to be retrieved and the segmentation mask to obtain first characteristics.

And fusing the positioning mask and the segmentation mask corresponding to the medical image to be retrieved to obtain the position information of the human tissue abnormality in the medical image to be retrieved. According to the position information, searching is carried out in a search base, and a first image is obtained. And further searching to narrow the searching range for follow-up. The accuracy of the retrieval operation is improved.

In a possible implementation manner, the determining, according to the degree of matching between the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image, a target first image from the first image whose degree of matching satisfies a preset condition includes: according to the similarity between the features, determining a first image with the similarity between the first feature and the second feature higher than a similarity threshold value as a target first image; the second characteristic is a characteristic representing the human tissue abnormality in the first image.

Fig. 9 shows a second feature extraction process of the medical image in the search library according to the embodiment of the present disclosure: the embodiment of the disclosure can determine the abnormal region of the human tissue in the first image by using the abnormal region information of the human tissue in the first image remark information. Then, performing convolution operation on the first image in the abnormal region range of the human tissue by using 5 decoders to obtain 5 convolution results; and then, according to each convolution result, obtaining second characteristics representing human tissue abnormity, and carrying out image characteristic extraction on the human tissue abnormity region in the first image to extract the second characteristics representing the human tissue abnormity. In practical applications, the number of decoders can be determined according to practical situations.

And comparing the first characteristic with the second characteristic to obtain the similarity of the first characteristic and the second characteristic. And when the similarity is higher than a preset similarity threshold, taking the first image corresponding to the second feature as a target first image.

And in the first image searched by the search library, performing secondary search by using the similarity of the first characteristic and the second characteristic, wherein the human tissue abnormal region of the searched target first image is similar to the image characteristic in the human tissue abnormal region of the medical image to be searched. The retrieval accuracy is further improved.

In one possible implementation, the second feature may be extracted using the first neural network and the parameters.

In a possible implementation manner, the determining, according to a matching degree between the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image, a target first image from the first images, where the matching degree satisfies a preset condition, includes: determining a first image with a difference value between the first range metric value and the second range metric value smaller than a difference threshold value as a second image according to the range metric value of the abnormal region of the human tissue; the first range metric value is a metric value of the range of the abnormal region of the human tissue indicated by the segmentation mask; the second range measurement value is the measurement value of the range of the human body tissue abnormal region in the first image; according to the similarity between the features, determining a second image with the similarity between the first feature and the third feature higher than the similarity threshold value as a target first image; the third characteristic is a characteristic representing human tissue abnormality in the second image.

In the embodiment of the present disclosure, a first metric value of a range of a human tissue abnormal region in a medical image to be retrieved, which is indicated by a segmentation mask, may be determined. The metric here may be a quantized value used to represent the size of an abnormal region of human tissue, such as: volume or maximum cross-sectional area. For the human body tissue abnormal region close to the sphere, the diameter of the human body tissue abnormal region can be used as a measurement value. The present disclosure does not limit the specific form of the metric value. And then, measuring a second metric value of the abnormal area range of the human tissue in the first image. And then, calculating the absolute value of the difference between the first metric value and the second metric value, and taking the first image corresponding to the second metric value as a second image when the absolute value of the difference is smaller than a preset difference threshold.

For example, the degree of matching V between the abnormal region of human tissue in the medical image to be retrieved and the abnormal region of human tissue in the first image_sCan be expressed by equation (4).

Wherein D_iA second range metric, p, representing the j th abnormal region of the human tissue of the medical image with sequence number i in the search library D_mAnd a first range metric value segment representing the m & ltth & gt human body tissue abnormal region of the medical image with the sequence number p in the medical image to be retrieved, wherein the larger the Vs is, the higher the matching degree is.

Next, image feature extraction may be performed on the abnormal human tissue region in the second image, and a third feature representing the abnormal human tissue may be extracted. And comparing the first characteristic with the third characteristic to obtain the similarity of the first characteristic and the third characteristic. And when the similarity is higher than the preset similarity threshold, taking the second image corresponding to the third feature as the target first image.

The comparison of the first feature with the second feature, and the comparison of the first feature with the third feature herein, can use any similarity measure algorithm according to actual needs, such as: and (4) Euclidean distance algorithm.

In one possible implementation, the third feature may be extracted using the first neural network and the parameters.

And in the first image searched by the search library, performing second search by using the measurement value of the abnormal region of the human tissue to further obtain a second image. And then, carrying out three times of retrieval on the similarity of the first characteristic and the third characteristic to obtain a target first image, so that the human tissue abnormal area of the target first image is similar to the human tissue abnormal area of the medical image to be retrieved in terms of satisfying the range, human anatomy structure position and characteristic. Because each retrieval operation is only carried out on one element without interference of other factors, the efficiency and the accuracy of the retrieval process are improved.

Fig. 10 is a block diagram of an image retrieval apparatus according to an embodiment of the disclosure, and as shown in fig. 10, the apparatus 30 includes:

a first feature determining unit 301, configured to determine a first feature that represents a human tissue abnormality in a medical image to be retrieved;

a retrieving unit 302, configured to retrieve, according to the first feature, a target first image matching the first feature from a search library; the target first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the target first image;

a display unit 303, configured to display the target first image and the remark information.

In a possible implementation manner, the first feature determining unit 301 includes:

a segmentation mask determining unit, configured to determine, according to the medical image to be retrieved, a segmentation mask indicating an abnormal region of a human tissue in the medical image to be retrieved;

A first feature determining subunit, configured to use, as the first feature, a feature of the abnormal human tissue region in the medical image to be retrieved, which is indicated by the segmentation mask.

In a possible implementation manner, the retrieving unit 302 includes:

the system comprises a positioning mask determining unit, a searching unit and a searching unit, wherein the positioning mask determining unit is used for determining a positioning mask indicating the position of a human anatomy structure in a medical image to be searched;

a first image retrieval unit, configured to retrieve a first image from the search library according to the localization mask and the segmentation mask;

And the first image searching subunit is used for determining the image with the same position information in the searching library as the first image.

In one possible implementation manner, the target first image retrieval unit includes:

the target first image first retrieval subunit is used for determining a first image, as a target first image, of which the similarity between the first feature and the second feature is higher than a similarity threshold value according to the similarity between the features; the second characteristic is a characteristic representing the human tissue abnormality in the first image.

the second image retrieval unit is used for determining a first image as a second image according to the range metric value of the abnormal human tissue region, wherein the difference value between the first range metric value and the second range metric value is smaller than a difference threshold value; the first range metric value is a metric value of the range of the abnormal region of the human tissue indicated by the segmentation mask; the second range metric value is the metric value of the range of the human body tissue abnormal region in the first image;

Fig. 11 is a block diagram of an image retrieval apparatus 800 according to an exemplary embodiment. For example, the apparatus 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, an exercise device, a personal digital assistant, and the like.

Referring to fig. 10, the apparatus 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input/output (I/O) interface 812, sensor component 814, and communications component 816.

The processing component 802 generally controls overall operation of the device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to perform all or a portion of the steps of the methods described above. Further, the processing component 802 can include one or more modules that facilitate interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations at the apparatus 800. Examples of such data include instructions for any application or method operating on device 800, contact data, phonebook data, messages, pictures, videos, and so forth. The memory 804 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.

A power supply component 806 provides power to the various components of the device 800. The power components 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 800.

The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front facing camera and/or a rear facing camera. The front camera and/or the rear camera may receive external multimedia data when the device 800 is in an operating mode, such as a shooting mode or a video mode. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a Microphone (MIC) configured to receive external audio signals when the apparatus 800 is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may further be stored in the memory 804 or transmitted via the communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to: a home button, a volume button, a start button, and a lock button.

The sensor assembly 814 includes one or more sensors for providing various aspects of state assessment for the device 800. For example, the sensor assembly 814 may detect the open/closed status of the device 800, the relative positioning of components, such as a display and keypad of the device 800, the sensor assembly 814 may also detect a change in the position of the device 800 or a component of the device 800, the presence or absence of user contact with the device 800, the orientation or acceleration/deceleration of the device 800, and a change in the temperature of the device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact. The sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 816 is configured to facilitate communication between the apparatus 800 and other devices in a wired or wireless manner. The apparatus 800 may access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communications. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

In an exemplary embodiment, the apparatus 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components for performing the above-described methods.

In an exemplary embodiment, a non-transitory computer-readable storage medium, such as the memory 804, is also provided that includes computer program instructions executable by the processor 820 of the device 800 to perform the above-described methods.

Fig. 12 is a block diagram of an image retrieval device 1900 according to an exemplary embodiment. For example, the apparatus 1900 may be provided as a server. Referring to fig. 10, the apparatus 1900 includes a processing component 1922 further including one or more processors and memory resources, represented by memory 1932, for storing instructions, e.g., applications, executable by the processing component 1922. The application programs stored in memory 1932 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1922 is configured to execute instructions to perform the methods described above.

The device 1900 may also include a power component 1926 configured to perform power management of the device 1900, a wired or wireless network interface 1950 configured to connect the device 1900 to a network, and an input/output (I/O) interface 1958. The device 1900 may operate based on an operating system stored in memory 1932, such as Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, or the like.

In an exemplary embodiment, a non-transitory computer readable storage medium, such as the memory 1932, is also provided that includes computer program instructions executable by the processing component 1922 of the apparatus 1900 to perform the above-described methods.

The present disclosure may be systems, methods, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for causing a processor to implement various aspects of the present disclosure.

The computer-readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be interpreted as a transitory signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or an electrical signal transmitted through an electrical wire.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device over a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.

Computer program instructions for carrying out operations of the present disclosure may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, aspects of the disclosure are implemented by personalizing an electronic circuit, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA), with state information of computer-readable program instructions, which can execute the computer-readable program instructions.

Various aspects of the present disclosure are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.

These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, but is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terms used herein were chosen in order to best explain the principles of the embodiments, the practical application, or technical improvements to the techniques in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An image retrieval method, comprising:

in a medical image to be retrieved, determining a first characteristic representing a human tissue abnormality, wherein the first characteristic at least comprises the following components: a spatial relationship between each of said human tissue anomalies;

displaying the target first image and the remark information;

The method for determining the first characteristic of the human tissue abnormality in the medical image to be retrieved comprises the following steps:

taking the feature of the abnormal human tissue region in the medical image to be retrieved, which is indicated by the segmentation mask, as the first feature;

the searching out the target first image matched with the first characteristic from the search library according to the first characteristic comprises the following steps:

determining a target first image of which the matching degree meets a preset condition from the first image according to the matching degree of the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image;

the determining, according to the matching degree between the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image, a target first image whose matching degree satisfies a preset condition from the first image includes:

determining a first image with a difference value between the first range metric value and the second range metric value smaller than a difference threshold value as a second image according to the range metric value of the abnormal region of the human tissue; the first range metric value is a metric value of the range of the abnormal region of the human tissue indicated by the segmentation mask; the second range metric value is the metric value of the range of the human body tissue abnormal region in the first image;

According to the similarity between the features, determining a second image with the similarity between the first feature and the third feature higher than a similarity threshold value as a target first image; the third characteristic is a characteristic representing the human tissue abnormality in the second image.

2. The method of claim 1, wherein the remark information includes at least one of:

3. The method of claim 1, wherein retrieving the target first image matching the first feature in a search library according to the first feature comprises:

and searching a first image in the search library according to the positioning mask and the segmentation mask.

4. The method of claim 3, wherein retrieving the first image from the search library according to the location mask and the segmentation mask comprises:

5. The method according to claim 3, wherein the determining, from the first images, a target first image whose matching degree satisfies a predetermined condition according to the matching degree of the abnormal human tissue region indicated by the segmentation mask and the abnormal human tissue region in the first image comprises:

6. An image retrieval apparatus, comprising:

a first feature determination unit, configured to determine, in a medical image to be retrieved, a first feature characterizing a human tissue abnormality, where the first feature at least includes: a spatial relationship between each of said human tissue anomalies;

the retrieval unit is used for retrieving a target first image matched with the first characteristic from a retrieval library according to the first characteristic; the target first image is provided with remark information, and the remark information is used for expressing relevant information of diseases corresponding to the target first image;

The display unit is used for displaying the target first image and the remark information;

the first feature determination unit includes:

a first feature determining subunit, configured to use, as the first feature, a feature of the abnormal human tissue region in the medical image to be retrieved, which is indicated by the segmentation mask;

the retrieval unit includes:

the target first image retrieval unit is used for determining a target first image of which the matching degree meets a preset condition from the first image according to the matching degree of the human tissue abnormal region indicated by the segmentation mask and the human tissue abnormal region in the first image;

the target first image retrieval unit includes:

The target first image second retrieval subunit is used for determining a second image, as a target first image, of which the similarity between the first feature and the third feature is higher than a similarity threshold value according to the similarity between the features; the third characteristic is a characteristic representing the human tissue abnormality in the second image.

7. An electronic device, comprising:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to invoke the memory-stored instructions to perform the method of any of claims 1 to 5.

8. A non-transitory computer readable storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement the method of any of claims 1 to 5.