CN117036376B - Lesion image segmentation method, device and storage medium based on artificial intelligence - Google Patents

Abstract

The application provides a lesion image segmentation method, device and storage medium based on artificial intelligence, which are characterized in that image segmentation is carried out on medical image images to be segmented to obtain image segmentation sets to be identified corresponding to the medical image images to be segmented, a reference image segmentation set is determined according to medical image segmentation to be identified, wherein the medical image segmentation sets to be identified meet identification conditions, the reference image segmentation sets are determined according to suspicious image segmentation in the reference image segmentation sets, identification image segmentation identification is carried out on identification image segmentation blocks in the identification image segmentation sets, and image segmentation marking is carried out when marking critical conditions are deeply identified according to identification results. Therefore, the identification image block set corresponding to the medical image to be segmented can be automatically identified, so that the identification image block identification is carried out on the identification image block set, and the image segmentation efficiency is improved.

Description

Lesion image segmentation method, device and storage medium based on artificial intelligence

Technical field

The present application relates to, but is not limited to, the technical fields of image processing and artificial intelligence, and in particular, to a method, device and storage medium for segmenting lesion images based on artificial intelligence.

Background technique

Lesion identification is one of the important tasks in medical image analysis, which aims to accurately locate and extract abnormal areas or lesions of interest from medical images. Among them, a variety of image processing technologies are needed for auxiliary collaboration, such as image enhancement, image segmentation, feature extraction, feature selection, classifier construction, object positioning and tracking, etc. In medical image segmentation, by extracting and separating areas of interest in medical images, especially lesion images, it is helpful to subsequently identify and locate lesions based on the segmentation results. Common segmentation methods are only based on threshold segmentation, edge detection, region growing, horizontal line segmentation, etc., which are of limited help for subsequent lesion identification. In addition, existing lesion image segmentation is limited by the limitations of manual labeling, and there is still room for improvement in segmentation accuracy and efficiency.

Contents of the invention

In view of this, embodiments of the present application at least provide an artificial intelligence-based lesion image segmentation method, device, and storage medium.

The technical solution of the embodiment of this application is implemented as follows:

In the first aspect, this application provides an artificial intelligence-based lesion image segmentation method, which method includes:

Obtain the medical image image to be segmented, perform an image segmentation operation on the medical image image to be segmented, and obtain a set of image segments to be identified corresponding to the medical image image to be segmented, wherein the medical image image to be segmented is rendered through volume rendering get;

Determine the pathological auxiliary promotion index of each to-be-identified medical image block in the to-be-identified image block set, and use the pathological auxiliary promotion index of each to-be-identified medical image block from the to-be-identified image block set. Determine to obtain a reference image block set; the medical image blocks in the reference image block set are to-be-identified medical image blocks whose corresponding pathological auxiliary promotion indicators in the to-be-identified image block set satisfy the identification conditions;

Determine a recognition image segment set according to the reference image segment set, and the recognition image segment in the recognition image segment set is a suspicious image segment;

Perform recognition image segmentation recognition according to the recognition image segmentation set, and perform image segmentation marking when it is determined based on the depth of the recognition result that the critical condition for marking is met;

The medical image image to be segmented is segmented based on the image segment labeling result of each medical image segment to be identified.

In some embodiments, determining the pathological auxiliary promotion index of each to-be-identified medical image block in the set of image blocks to be identified includes:

For any first medical image block to be identified in the set of image blocks to be identified whose block size is greater than a size threshold, determine the internal correlation evaluation index of the first medical image block to be identified and the first medical image to be identified The external correlation evaluation index of the block, and the determination of the first medical image block to be identified based on the internal correlation evaluation index of the first medical image block to be identified and the external correlation evaluation index of the first medical image block to be identified. Pathological auxiliary promotion indicators;

For any second medical image block to be identified in the set of image blocks to be identified whose block size is less than or equal to the size threshold, determine the external correlation evaluation index of the second medical image block to be identified, and set The external correlation evaluation index of the second medical image block to be identified is used as the pathological auxiliary promotion index of the second medical image block to be identified.

In some embodiments, the medical image image to be segmented is any slice in the target three-dimensional medical image data; and determining the internal correlation evaluation index of the first medical image segment to be identified includes:

Determine the pixel cluster group corresponding to the first medical image segment to be identified. Any pixel cluster group is composed of multiple cut pixel clusters obtained by cutting the first medical image segment to be identified. Each cut pixel cluster is Composed of one pixel or multiple surrounding pixels in the first medical image block to be identified;

For any pixel cluster group, obtain the confidence that each cut pixel cluster in the pixel cluster group appears in the target three-dimensional medical imaging data;

According to the confidence that each cut pixel cluster appears in the target three-dimensional medical imaging data and the confidence that the first medical image block to be identified appears in the target three-dimensional medical imaging data, determine the pixel cluster grouping The corresponding internal correlation evaluation index;

Determine the minimum internal correlation evaluation index among the internal correlation evaluation indicators corresponding to each pixel group grouping as the internal correlation evaluation index of the first to-be-identified medical image block;

Determining the external correlation evaluation index of the first medical image block to be identified includes:

Determine a neighborhood pixel set of the first to-be-identified medical image block in the target three-dimensional medical image data, where the neighborhood pixel set includes no less than one neighborhood pixel;

Obtain the confidence that the fused medical image blocks to be identified, obtained by combining each neighborhood pixel and the first medical image block to be identified, respectively appear in the target three-dimensional medical image data;

The external correlation evaluation index of the first medical image segment to be identified is determined based on the confidence of each fused medical image segment to be identified appearing in the target three-dimensional medical image data.

In some embodiments, the neighborhood pixels included in the neighborhood pixel set are surrounding pixels of the first to-be-identified medical image block in the target three-dimensional medical image data; the acquisition of each neighborhood pixel and The confidence that the fused medical image blocks to be identified respectively appear in the target three-dimensional medical image data obtained by combining the first medical image blocks to be identified includes obtaining each surrounding pixel in the neighborhood pixel set and The confidence that the first fused medical image blocks to be identified obtained by combining the first medical image blocks to be identified appear in the target three-dimensional medical image data respectively;

Determining the external correlation evaluation index of the first medical image segment to be identified based on the confidence that each fused medical image segment to be identified appears in the target three-dimensional medical image data includes:

Obtain a plurality of first uncertainty measurement values based on the confidence of each first to-be-identified fused medical image segment appearing in the target three-dimensional medical image data;

The minimum uncertainty metric value among the plurality of first uncertainty metric values is determined as the external correlation evaluation index of the first to-be-identified medical image block.

In some embodiments, the medical image image to be segmented is any slice in the target three-dimensional medical image data, and the target three-dimensional medical image data is obtained at the current acquisition moment; the block set is based on the reference image Determine the set of recognized image blocks, including:

Acquire a reference lesion image set, the reference lesion image set being composed of normal image segments and identification image segments determined based on three-dimensional medical imaging data acquired before the current acquisition moment;

Each medical image block in the reference image block set is matched in the reference lesion image set, and the identification image block set is determined based on the unsuccessfully matched medical image block.

In some embodiments, the target recognition image segment is any recognition image segment in the recognition image segment set, and the method further includes:

Match the matching medical image segment that matches the target recognition image segment in the reference lesion image set; if matched, determine the new recognition image through the target recognition image segment and the matching medical image segment. block, and iterating the set of recognition image blocks based on the newly added recognition image blocks; and iterating the set of reference lesion images based on the iterated set of recognition image blocks.

In some embodiments, the target recognition image block is any recognition image block in the recognition image block set, and the recognition image block recognition based on the recognition image block set includes:

Obtaining the first occurrence frequency of the target recognition image segment in the three-dimensional medical imaging data at R acquisition times, and acquiring the second occurrence frequency of the target recognition image segment in the three-dimensional medical imaging data at the target acquisition time frequency, the target acquisition time is the acquisition time after the R acquisition moments among the S acquisition times; calculate the quotient between the first occurrence frequency and the second occurrence frequency; divide the first occurrence frequency into The quotient between the frequency and the second frequency of occurrence is determined as the recognition result of the target recognition image block;

Alternatively, the step of performing recognition image segment recognition based on the recognition image segment set includes:

Obtaining the first frequency of occurrence of the target recognition image segment in the three-dimensional medical imaging data at R acquisition times, and acquiring the third frequency of the three-dimensional medical imaging data in the S acquisition time of the target recognition image segment frequency of occurrence;

Calculate the quotient between the first frequency of occurrence and the third frequency of occurrence;

Determine the quotient between the first frequency of occurrence and the third frequency of occurrence as the recognition result of the target recognition image block;

Wherein, the S acquisition moments include the current acquisition moment and S-1 acquisition moments before the current acquisition moment; the R acquisition moments include the current acquisition moment and R acquisition moments before the current acquisition moment. -1 acquisition time, R<S.

In some embodiments, the image block marking is performed when the critical conditions for marking are met according to the depth of the recognition result, including:

Get the preset mark reference value;

When the quotient corresponding to the recognition result of the target recognition image block is greater than or equal to the preset marking reference value, it is determined that the marking critical condition is met, and the corresponding target recognition image block is marked accordingly.

In some embodiments, obtaining a preset mark reference value includes:

Obtain the frequency of occurrence of the target recognition image segment in the three-dimensional medical imaging data obtained at the current acquisition time and the previous time;

Determine the recognition type corresponding to the target recognition image block according to the frequency of occurrence;

Obtain the preset mark reference value corresponding to the recognition type; the recognition image blocks in the recognition image block set are classified into no less than one recognition type based on the frequency of occurrence of each recognition image segment, and one recognition type corresponds to A preset marker reference value.

In a second aspect, this application provides a lesion image segmentation device, including:

An image acquisition module, configured to acquire a medical image image to be segmented, perform an image segmentation operation on the medical image image to be segmented, and obtain a set of image segments to be identified corresponding to the medical image image to be segmented, wherein the medical image image to be segmented is Medical imaging images are obtained through volume rendering;

An index determination module, configured to determine the pathological auxiliary promotion index of each to-be-identified medical image block in the to-be-identified image block set, and use the pathological auxiliary promotion index of each of the to-be-identified medical image blocks to obtain the pathological auxiliary promotion index from the to-be-identified medical image block. A reference image block set is determined from the identified image block set; the medical image blocks in the reference image block set are to-be-identified medical images whose corresponding pathological auxiliary promotion indicators in the to-be-identified image block set satisfy the identification conditions. image tiles;

A block determination module, configured to determine a set of identified image blocks based on the set of reference image blocks, and the identified image blocks in the set of identified image blocks are suspicious image blocks;

A block marking module, configured to perform recognition image block recognition based on the recognition image block set, and perform image block marking when it is deeply recognized based on the recognition result that the critical condition for marking is met;

An image segmentation module, configured to segment the medical image image to be segmented based on the image segment labeling result of each medical image segment to be identified.

In a third aspect, the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method are implemented.

The at least beneficial effects of this application include:

This application performs image segmentation on the medical image image to be segmented to obtain a set of image segments to be identified corresponding to the medical image image to be segmented, and determines the segments of the medical image to be recognized that meet the recognition conditions based on the corresponding pathological auxiliary promotion indicators in the image segment set to be recognized. Referring to the image block set, determine the identification image block set based on the suspicious image blocks in the reference image block set, and perform identification image block recognition on the identification image blocks in the identification image block set, and based on the depth of the recognition result When it is recognized that the critical conditions for marking are met, the image is marked into blocks. In this way, the recognition image block set corresponding to the medical image image to be segmented can be automatically identified to perform recognition image block recognition on the recognition image block set, which increases the efficiency of image segmentation. At the same time, the identified image blocks in the identified image block set are suspicious image blocks. The suspicious image blocks can be used to characterize new suspicious image blocks that may appear. Therefore, the suspicious image blocks are identified and the suspicious image blocks are identified based on the suspicious image blocks. The recognition results of image blocks are deeply recognized and marked when the critical conditions for marking are met. New suspicious image blocks that may appear can be pre-marked to accurately and quickly complete the segmentation of lesion images.

It should be understood that the above general description and the following detailed description are only exemplary and explanatory, but do not limit the technical solution of the present application.

Description of the drawings

The accompanying drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application, and together with the description, are used to explain the technical solutions of the present application.

Figure 1 is a schematic flow chart of the implementation of an artificial intelligence-based lesion image segmentation method provided by an embodiment of the present application.

Figure 2 is a schematic structural diagram of a lesion image segmentation device provided by an embodiment of the present application.

Figure 3 is a schematic diagram of a hardware entity of a computer device provided by an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated below in conjunction with the accompanying drawings and examples. The described embodiments should not be regarded as limiting the present application. Those of ordinary skill in the art will All other embodiments obtained without creative work fall within the scope of protection of this application.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or a different subset of all possible embodiments, and Can be combined with each other without conflict. The terms "first/second/third" involved are only used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first/second/third" can be used interchangeably if permitted. The specific order or sequence may be changed so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs. The terminology used herein is for the purpose of describing the application only and is not intended to be limiting.

The embodiment of the present application provides an artificial intelligence-based lesion image segmentation method, which can be executed by a processor of a computer device. Among them, computer equipment may refer to servers, laptops, tablets, desktop computers, smart TVs, mobile devices (such as mobile phones, portable video players, personal digital assistants) and other devices with data processing capabilities.

Figure 1 is a schematic flow chart of the implementation of an artificial intelligence-based lesion image segmentation method provided by an embodiment of the present application. As shown in Figure 1, the method includes the following operations 101 to 105:

In operation S101, a medical image image to be segmented is acquired, and an image block operation is performed on the medical image image to be segmented to obtain a set of image blocks to be identified corresponding to the medical image image to be segmented.

In the embodiment of this application, the medical image image to be segmented is obtained through volume rendering (Volume Rendering). The medical image image to be segmented is any slice in the target three-dimensional medical image data. The target three-dimensional medical image data is obtained at the current acquisition time. The target three-dimensional medical image data includes no less than one medical image image to be segmented. After obtaining the medical image image to be segmented, the image segmentation operation can be performed on the medical image image to be segmented to obtain the image segmentation set to be identified corresponding to the medical image image to be segmented. The image segmentation set to be recognized includes the different image segments obtained by the image segmentation operation. Less than one medical image patch to be identified. In the embodiment of the present application, the process of image segmentation operation may specifically be a process of basic image segmentation of the medical imaging image to be segmented, such as segmentation according to pixel grayscale and preset grayscale threshold, or through edge detection (such as Canny Edge detection, Sobel operator) method is used to segment the image into different image segments. In other embodiments, the medical image image to be segmented can also be segmented into multiple image segments with the same pixel size according to the mean segmentation method. piece.

Operation S102, determine the pathological auxiliary promotion index of each medical image block to be identified in the set of image blocks to be identified, and determine the reference from the set of image blocks to be identified through the pathological auxiliary promotion index of each medical image block to be identified. A collection of image tiles.

In the embodiment of the present application, the pathological auxiliary promotion index of the medical image segmentation to be identified is an evaluation index used for secondary image segmentation. The reference basis of the evaluation index is the recognition promotion element of pathology, which may include at least one of the following indicators. One: the internal correlation evaluation index of the medical image block to be identified, the external correlation evaluation index of the medical image block to be identified; the medical image block in the reference image block set is the corresponding pathological auxiliary in the image block set to be identified To-be-identified medical image blocks whose promotion indicators meet the recognition conditions are determined as medical image blocks in the reference image block set whose corresponding pathological auxiliary promotion indicators satisfy the recognition conditions in the set of image blocks to be identified. . For example, for any first medical image block to be identified in the set of image blocks to be identified whose block size is greater than the size threshold, the size threshold is, for example, 0.5 mm × 0.5 mm, specifically depending on the image modality (such as CT, MRI), The specific disease type is related to the lesion characteristics and is not specifically limited. Determining the pathological auxiliary promotion index of the first medical image block to be identified includes: determining the internal correlation evaluation index of the first medical image block to be identified and the external correlation evaluation index of the first medical image block to be identified, and based on the first medical image block to be identified. The internal correlation evaluation index of the first medical image block to be identified and the external correlation evaluation index of the first medical image block to be identified determine the pathological auxiliary promotion index of the first medical image block to be identified. At this time, the pathology auxiliary promotion index of the first medical image block to be identified satisfies the identification condition, for example, the pathology auxiliary promotion index of the first medical image block to be identified is greater than or equal to the first index value. In other embodiments, for any second medical image block to be identified whose block size is less than or equal to the size threshold in the set of image blocks to be identified, determining the pathological auxiliary promotion index of the second medical image block to be identified includes: determining The external correlation evaluation index of the second medical image block to be identified is used as the pathological auxiliary promotion index of the second medical image block to be identified. The second medical image block to be identified is The pathological auxiliary promotion index of the image block satisfying the identification condition means that the pathology auxiliary promotion index of the second medical image block to be identified is greater than or equal to the second index value.

Optionally, the set of image blocks to be identified includes no less than one first medical image block to be identified, and each first to be identified medical image block can be classified according to the order of decreasing pathological auxiliary promotion index of each first medical image block to be identified. The medical image blocks are arranged to obtain a first medical image block sequence to be identified, and the first medical image block to be identified in the first medical image block sequence to be identified and located before the first distribution order is determined as the reference image block Medical image blocks in the set, where the first distribution order is the first M of the first sequence of medical image blocks to be identified. According to the same idea, the set of image blocks to be identified includes no less than one second medical image block to be identified, and each second medical image block to be identified is also classified according to the order of decreasing pathological auxiliary promotion index of each second medical image block to be identified. The medical image blocks are arranged to obtain a second medical image block sequence to be identified, and the second medical image block to be identified that is located before the second distribution order in the second medical image block sequence to be identified is determined as the reference image block The medical image blocks in the collection; the second distribution order refers to the first N blocks of the second medical image block sequence to be identified. The specific values of M and N can be set according to actual needs.

In the embodiment of the present application, the internal correlation evaluation index of the medical image block to be identified can be based on the involvement between each cut pixel group in the medical image block to be identified (that is, the correlation between each cut pixel group) To evaluate whether the medical image block to be identified is suitable for independent existence, each cutting pixel cluster consists of one pixel or multiple surrounding pixels (that is, multiple adjacent pixels) in the medical image block to be identified. Among them, independent existence means that the medical image block to be identified is regarded as an independent area. The greater the internal correlation evaluation index of the medical image block to be identified, the smaller the number of cut pixel clusters in the medical image block to be identified. The stronger the correlation between them, the more suitable the medical image blocks to be identified are to exist independently. The external correlation evaluation index of the medical image block to be identified can be obtained by combining the medical image block to be identified and the neighborhood pixels to obtain the uncertainty measurement value of the fused medical image block to be identified (the uncertainty of the probability distribution can be used to determine Measurement, for example, by counting the frequency of each pixel value in the fused medical image block to be identified, a histogram of pixel values is obtained, and then dividing the frequency of each pixel value by the total number of pixels in the fused medical image block to be identified, Obtain the probability of each pixel value, then perform a logarithmic operation on each probability value and multiply it by the negative number of the logarithm. Finally, sum the results of all pixel values to obtain the uncertainty measure of the fused medical image block to be identified. value) to evaluate the extent to which the medical image blocks to be identified are suitable for independent existence. The neighborhood pixels may include surrounding pixels of the medical image block to be identified in the target three-dimensional medical imaging data; the uncertainty measurement value of the fused medical image block to be identified is used to characterize the unpredictability of the fused medical image block to be identified; The smaller the uncertainty measure value of the fused medical image block to be identified, the smaller the unpredictability of the fused medical image block to be identified, and the higher the possibility that the neighborhood pixels are components of the medical image block to be identified. That is to say, the smaller the uncertainty measurement value of the fused medical image block to be identified, the smaller the external correlation evaluation index of the medical image block to be identified, and the less suitable the medical image block to be identified to exist independently. The medical image blocks in the reference image block set are determined based on the medical image blocks to be identified that are suitable for independent existence in the image block set to be identified. For any first medical image block to be identified whose block size is not less than the size threshold in the set of image blocks to be identified, determining the pathological auxiliary promotion index of the first medical image block to be identified includes the following operations:

Operation 1: Determine the internal correlation evaluation index of the first medical image block to be identified.

Specifically, the pixel cluster group corresponding to the first medical image segment to be identified can be determined. Any pixel cluster group is composed of multiple cut pixel clusters obtained by cutting the first medical image segment to be identified. Each cut pixel cluster is It consists of one pixel or multiple surrounding pixels in the first medical image block to be identified. For any pixel group group, obtain the confidence that each cut pixel group in the pixel group appears in the target 3D medical imaging data, and use the confidence level and first appearance of each cut pixel group in the target 3D medical imaging data The confidence that the medical image segment to be identified appears in the target three-dimensional medical imaging data determines the internal correlation evaluation index corresponding to the pixel cluster grouping. The minimum internal correlation evaluation index among the internal correlation evaluation indicators corresponding to each pixel group group is used as the internal correlation evaluation index of the first medical image block to be identified. The confidence that the cut pixel cluster appears in the target 3D medical imaging data is the quotient between the number of occurrences of the cut pixel cluster in the target 3D medical imaging data and the total number of image blocks in the target 3D medical imaging data, or the number of times the cut pixel cluster appears in the target 3D medical imaging data. The quotient between the number of occurrences in the target three-dimensional medical imaging data and the total number of image blocks in the target three-dimensional medical imaging data. The confidence level that the first medical image segment to be identified appears in the target 3D medical image data is between the number of times the first medical image segment to be identified appears in the target 3D medical image data and the total number of image segments in the target 3D medical image data. , or the quotient of the number of times the first medical image block to be identified is in the target three-dimensional medical image data and the total number of image blocks in the target three-dimensional medical image data.

The calculation method of the internal correlation evaluation index of the pixel cluster group can refer to the general PMI (PointwiseMutual Information) calculation formula. I will not go into details here. The minimum internal correlation evaluation among the internal correlation evaluation indicators corresponding to multiple pixel clusters is used. The indicator is used as the internal correlation evaluation indicator of the first medical image block to be identified.

Operation 2: Determine the external correlation evaluation index of the first medical image block to be identified.

Specifically, the neighborhood pixel set of the first to-be-identified medical image block can be determined in the target three-dimensional medical image data. The neighborhood pixel set includes no less than one neighborhood pixel, and the neighborhood pixels in the neighborhood pixel set are The surrounding pixels of the first medical image block to be identified in the target three-dimensional medical image data are obtained, and the fused medical image blocks to be identified obtained by combining each neighborhood pixel with the first medical image block to be identified are respectively in the target three-dimensional medical image. The external correlation evaluation index of the first medical image segment to be identified is determined based on the confidence level appearing in the data and the appearance confidence level of each fused medical image segment to be identified in the target three-dimensional medical image data. Wherein, the fused medical image block to be identified is obtained by combining a neighborhood pixel and the first medical image block to be identified.

For example, the neighborhood pixels included in the neighborhood pixel set are the surrounding pixels of the first medical image segment to be identified in the target three-dimensional medical image data, and each surrounding pixel in the neighborhood pixel set is obtained with the first medical image segment to be identified. The confidence of each first fusion medical image block to be identified appearing in the target three-dimensional medical imaging data obtained by block combination, and the confidence of the appearance of each first fused medical image block to be identified in the target three-dimensional medical imaging data is obtained. A plurality of first uncertainty measurement values, and a minimum uncertainty measurement value among the plurality of first uncertainty measurement values is determined as an external correlation evaluation index of the first medical image segment to be identified. The confidence that the first fused medical image segment to be identified appears in the target 3D medical imaging data is the number of occurrences of the first fused medical image segment to be identified in the target 3D medical imaging data and the image segmentation of the target 3D medical imaging data. The quotient between the total number or the quotient of the number of times the first fused medical image block to be identified is in the target three-dimensional medical image data and the total number of image blocks of the target three-dimensional medical image data.

The calculation method of the first uncertainty measure value can refer to the general information entropy calculation formula, which will not be described in detail here. As another implementation manner, the neighborhood pixels included in the neighborhood pixel set are the first to-be-identified medical image blocks in the target Surrounding pixels in the three-dimensional medical image data are obtained. Each surrounding pixel in the neighborhood pixel set is combined with the first to-be-identified medical image block to obtain the first to-be-identified fused medical image block, which appears in the target three-dimensional medical image data respectively. The confidence level of each first fusion medical image segment to be identified appears in the target three-dimensional medical image data, and the first uncertainty measurement value is obtained based on the confidence level of each first fusion medical image segment to be identified, and the first uncertainty measurement value is used as the first medical image segment to be identified. External correlation evaluation index of image block.

Operation 3: Determine the pathological auxiliary promotion index of the first medical image segment to be identified based on the internal correlation evaluation index and the external correlation evaluation index of the first medical image segment to be identified.

Specifically, the first importance parameter corresponding to the internal correlation evaluation index of the first medical image block to be identified is obtained, and the second importance parameter corresponding to the internal correlation evaluation index of the first medical image block to be identified is obtained, The first to be identified is obtained according to the first importance parameter, the internal correlation evaluation index of the first medical image block to be identified, the second importance parameter, and the external correlation evaluation index of the first medical image block to be identified. Pathology-assisted facilitation metrics for medical image patching. The calculation formula of the pathological auxiliary promotion index of the first medical image block to be identified is:

G＝αI+βO

G is the pathological auxiliary promotion index of the first medical image block to be identified, α is the first importance parameter, I is the internal correlation evaluation index of the first medical image block to be identified, β is the second importance parameter, O is the external correlation evaluation index of the first medical image block to be identified.

For any second to-be-identified medical image block in the set of to-be-identified image block segments whose block size is not greater than the size threshold, determining the pathological auxiliary promotion index of the second to-be-identified medical image block specifically includes: determining the second to-be-identified medical image block The external correlation evaluation index of the image block is used as the pathological auxiliary promotion index of the second medical image block to be identified. The method of determining the external correlation evaluation index of the second medical image block to be identified may refer to determining the external correlation evaluation index of the first medical image block to be identified.

The neighborhood pixels included in the neighborhood pixel set of the second medical image block to be identified are the surrounding pixels of the second medical image block to be identified in the target three-dimensional medical image data. The confidence of the appearance of the third fusion medical image segment to be identified in the target three-dimensional medical image data obtained by combining each surrounding pixel with the second medical image segment to be identified can be obtained, and based on each third fusion medical image segment to be identified. Obtain the third uncertainty measure value based on the confidence of the block appearing in the target three-dimensional medical image data; determine the minimum uncertainty measure value in the third uncertainty measure value as the external correlation of the second medical image block to be identified sexual evaluation index. Wherein, the confidence of the appearance of the third fused medical image segment to be identified in the target three-dimensional medical imaging data is the number of occurrences of the third fused medical image segment to be identified in the target three-dimensional medical imaging data and the image of the target three-dimensional medical imaging data. The quotient between the total number of blocks or the quotient of the number of occurrences of the third fusion medical image block to be identified in the target three-dimensional medical image data and the total number of image blocks of the target three-dimensional medical imaging data.

In other embodiments, the neighborhood pixels included in the neighborhood pixel set are surrounding pixels of the second to-be-identified medical image block in the target three-dimensional medical image data. Each surrounding pixel in the neighborhood pixel set and the second to-be-identified medical image block can be obtained. Identify the confidence that the third fusion medical image blocks to be identified, which are combined by the medical image blocks, appear in the target three-dimensional medical image data, and the appearance of each third fusion medical image block to be identified in the target three-dimensional medical image data. The third uncertainty measure value is obtained from the confidence level that appears, and the calculated third uncertainty measure value is used as the external correlation evaluation index of the second medical image block to be identified.

In operation S103, the identified image block set is determined according to the reference image block set.

The recognized image segments in the recognized image segment set are medical image segments in the reference image segment set that do not appear in the reference lesion image set. The medical image image to be segmented is any slice in the target three-dimensional medical image data. The target three-dimensional medical image data is obtained at the current acquisition time. Determining the identified image block set according to the reference image block set includes: obtaining the reference lesion image set, The reference lesion image set is composed of normal image segments and identified image segments determined based on the three-dimensional medical image data acquired before the current acquisition time. In the reference lesion image set, each medical image segment in the reference image segment set is The blocks are matched, and a set of identified image blocks is determined based on the unsuccessfully matched medical image blocks.

Optionally, the reference lesion image set is iterated using the recognition image segment set, that is, the recognition image segment in the recognition image segment set is added to the reference lesion image set. During the iterative process of the reference lesion image set, the image blocks in the current reference lesion image set are retained, and the identified image block set determined based on the medical image image to be segmented each time is added to the reference lesion image set. Based on this, the completeness of the reference lesion image set is improved.

In operation S104, the recognition image block recognition is performed according to the recognition image block set, and the image block marking is performed when it is deeply recognized according to the recognition result that the critical condition for marking is met.

According to the frequency of occurrence of each recognition image segment in the recognition image segmentation set in all three-dimensional medical imaging data, the recognition image segmentation in the recognition image segmentation set is recognized, and the recognition result of each recognition image segmentation is obtained. , identify the recognition results of each recognized image block, and mark the image block when the critical conditions for marking are met through deep recognition. Among them, all three-dimensional medical imaging data includes all three-dimensional medical imaging data obtained at all acquisition times, and the frequency of occurrence of identified image blocks in all three-dimensional medical imaging data is the number of occurrences of identified image blocks in all three-dimensional medical imaging data, Or identify the quotient between the number of occurrences of image blocks in all three-dimensional medical imaging data and the total number of image blocks in all three-dimensional medical imaging data.

In operation S105, the medical image image to be segmented is segmented based on the image segment labeling result of each medical image segment to be identified.

Specifically, each image block in the medical image image to be segmented is marked according to the image block labeling result, and the corresponding labeling result is obtained. The labeling process completes the segmentation of the medical image image to be segmented.

In the embodiment of the present application, the medical image image to be segmented is divided into image blocks to obtain a set of image blocks to be identified corresponding to the medical image image to be segmented. The medical image segment to be identified satisfies the identification conditions according to the corresponding pathological auxiliary promotion index in the image block set to be identified. The block determines the reference image block set. The medical image blocks in the reference image block set are medical image blocks to be identified that are suitable for independent existence in the image block set to be identified. According to the suspicious image blocks in the reference image block set, The block determines the recognition image block set, and performs recognition image block recognition on the recognition image block in the recognition image block set, and performs image segmentation marking when it is deeply recognized based on the recognition result that the critical conditions for marking are met. In this way, the recognition image block set corresponding to the medical image image to be segmented can be automatically identified to perform recognition image block recognition on the recognition image block set, thereby increasing the efficiency of image segmentation. At the same time, the identified image blocks in the identified image block set are suspicious image blocks. The suspicious image blocks are used to characterize new suspicious image blocks that may appear. Therefore, the suspicious image blocks are identified and the suspicious image blocks are identified based on the suspicious image blocks. When the block recognition results are deeply recognized and meet the critical conditions for marking, image block labeling can be pre-marked for new suspicious image blocks that may appear, and the segmentation of the lesion image can be completed accurately and quickly.

In another embodiment, the artificial intelligence-based lesion image segmentation method provided by the embodiment of the present application may include the following operations:

Operation 201: Obtain a medical image image to be segmented, perform an image block operation on the medical image image to be segmented, and obtain a set of image blocks to be identified corresponding to the medical image image to be segmented.

Operation 202: Determine the pathological auxiliary promotion index of each medical image block to be identified in the set of image blocks to be identified, and determine the reference from the set of image blocks to be identified through the pathological auxiliary promotion index of each medical image block to be identified. A collection of image tiles.

Operation 203: Determine the identified image block set according to the reference image block set.

For the process of operation 201 to operation 203, please refer to the aforementioned operations 101 to 103.

Operation 204: Perform recognition image block recognition according to the recognition image block set.

Before operation 204, optionally, the target recognition image segment is any recognition image segment in the recognition image segment set, for example, using image vector clustering to match the matching medical image segment matching the target recognition image segment in the reference lesion image set. Image block, if matched, determine the new recognition image block based on the target recognition image block and the matching medical image block, for example, combine the target recognition image block and the matching medical image block to obtain the new recognition image Block. Among them, image vector clustering is used to match the matching medical image blocks that match the target identification image blocks in the reference lesion image set. Specifically, the medical image blocks to be matched can be determined in the reference lesion image set, and the target identification is obtained through a vector conversion algorithm. Image feature vectors of image blocks and image feature vectors of medical image blocks to be matched (when obtaining image feature vectors, you can first perform image preprocessing and then perform feature extraction, such as using LBP, HOG, color histogram or SIFT, etc. method to extract, then perform feature encoding, and finally normalize the feature vector to obtain the image feature vector). When the vector difference between the image feature vector of the target recognition image block and the image feature vector of the medical image block to be matched is less than the threshold, or when the image feature vector of the target recognition image block and the image of the medical image block to be matched If the inner product product between the feature vectors is greater than the preset inner product product, it is determined that the medical image block to be matched is a matching medical image block that matches the target recognition image block.

In other embodiments, matching medical image blocks that match the target recognition image blocks can be matched in the reference lesion image set through image feature vector clustering. If matched, the determination is based on the target recognition image blocks and the matching medical image blocks. Add new recognition image blocks, such as combining target recognition image blocks with matching medical image blocks to obtain new recognition image blocks. You can also iterate the recognition image block collection based on the new recognition image blocks, and The reference lesion image set is iterated based on the iterated recognition image segment set, that is, the recognition image segments in the iterated recognition image segment set are added to the reference lesion image set. Based on this, by combining the matching medical image blocks in the reference lesion image set with the target identification image blocks, new suspicious image blocks can be obtained, the identification image block set can be expanded, and the identification interval can be increased.

In operation 204, the recognition of image segmentation based on the recognition image segmentation set may specifically include one or more of the following strategies:

Strategy 1: First obtain the frequency of occurrence of each recognized image block in the recognized image block set in all three-dimensional medical imaging data. Among them, all three-dimensional medical imaging data include all three-dimensional medical imaging data obtained at all acquisition times. Recognition The frequency of occurrence of image blocks in all three-dimensional medical imaging data is to identify the number of occurrences of image blocks in all three-dimensional medical imaging data, or to identify the number of occurrences of image blocks in all three-dimensional medical imaging data and all three-dimensional medical imaging data. The quotient between the total number of image patches in .

Strategy 2. Target recognition image segmentation is to recognize any recognition image segmentation in the image segmentation set. When identifying the target recognition image segmentation, the target recognition image segmentation is obtained at R acquisition moments to obtain three-dimensional medical imaging data. The first frequency of occurrence in , and the second frequency of occurrence in the three-dimensional medical imaging data obtained by acquiring the target recognition image block at the target acquisition time. The target acquisition time is the acquisition time after the R acquisition moments among the S acquisition moments. Calculate the quotient between the first frequency of appearance and the second frequency of appearance; determine the quotient between the first frequency of appearance and the second frequency of appearance as the recognition result of the target recognition image block. To identify target recognition image blocks, you can refer to the following formula:

θ=(i+t)÷(j+t)

θ is the recognition result of the target recognition image block, i is the first frequency of occurrence in the three-dimensional medical imaging data obtained by the target recognition image block at R acquisition times, and j is the three-dimensional target recognition image block obtained at the target acquisition time. The second frequency of occurrence in medical imaging data, t is a constant. The S acquisition moments include the current acquisition time and S-1 acquisition moments before the current acquisition time; the R acquisition moments include the current acquisition time and R-1 acquisition moments before the current acquisition time, R<S. The recognition results of the target recognition image segmentation through strategy 2 can represent the increased value of the first frequency of appearance of the target recognition image segmentation at R acquisition times compared to the second appearance frequency of the target recognition image segmentation within the target acquisition time.

Strategy 3. Target recognition image segmentation is to recognize any recognition image segment in the image segmentation set. Recognition of the target recognition image segmentation includes: acquiring the target recognition image segmentation and obtaining three-dimensional medical imaging data at R acquisition moments. The first frequency of occurrence in , and the target recognition image segmentation is obtained at S acquisition moments to obtain the third frequency of occurrence in the three-dimensional medical imaging data, the quotient between the first frequency of occurrence and the third frequency of occurrence is obtained, and the first frequency of occurrence is obtained The quotient between the frequency of occurrence and the third frequency of occurrence is determined as the recognition result of the target recognition image block. The formula for identifying target recognition image blocks is:

λ=i÷ε

λ is the recognition result obtained by the target recognition image segmentation through strategy three, i is the first frequency of occurrence of the target recognition image segmentation in the three-dimensional medical imaging data obtained at R acquisition moments, ɛ is the target recognition image segmentation at S times The acquisition time is to obtain the third frequency of occurrence in the three-dimensional medical imaging data. The recognition result of the target recognition image segment through strategy three is used to represent the ratio of the first frequency of appearance of the target recognition image segment within the R acquisition moments to the third frequency of appearance of the target recognition image segment within the S acquisition moments.

Operation 205: Obtain a preset mark reference value.

The recognition image blocks in the recognition image block set are classified into no less than one recognition type based on the frequency of occurrence of each recognition image segment in all three-dimensional medical imaging data. One recognition type corresponds to a preset mark reference value. Obtaining the preset mark reference value includes: obtaining the frequency of occurrence of the target recognition image block in the current acquisition time and the previous time (i.e., all acquisition times) in the three-dimensional medical imaging data; determining the identification corresponding to the target recognition image block based on the frequency of occurrence Type to obtain the preset mark reference value corresponding to the identification type. For example, the recognition image blocks in the recognition image block set are classified into multiple data recognition types, general data recognition types and few data recognition types based on the frequency of occurrence of each recognition image segment in all three-dimensional medical imaging data. The recognition type corresponds to the multi-data preset mark reference value, the general data recognition type corresponds to the general data preset mark reference value, and the less data recognition type corresponds to the less data preset mark reference value. When the target recognition image is divided into blocks in all three-dimensional medical imaging data The frequency of occurrence in determines that when the target recognition image is divided into multiple data categories, the obtained preset mark reference value is the multi-data preset mark reference value.

Operation 206: When the quotient corresponding to the recognition result of the target recognition image segment is greater than or equal to the preset marking reference value, it is determined that the marking critical condition is met, and the corresponding target recognition image segment is marked accordingly.

When the quotient corresponding to the recognition result of the target recognition image block is not less than the preset mark reference value, it is determined that the mark critical condition is met. Optionally, arrange the recognition image blocks in the order of decreasing quotients corresponding to the recognition results of each recognition image block to obtain the recognition image block sequence, and classify the recognition image blocks in the sequence before the third distribution order. The image blocks are determined as recognition image blocks that satisfy the labeling critical condition, and the third distribution order is, for example, the first K recognition image blocks in the recognition image block sequence.

Operation 207: Mark the image blocks according to the image block labeling strategy to complete the image segmentation.

The medical image image to be segmented can be obtained, and the medical image image to be segmented can be subjected to an image segmentation operation to obtain a set of image segments to be identified corresponding to the medical image image to be segmented; and each medical image segment to be recognized in the image segmentation set to be recognized can be determined. The pathological auxiliary promotion index of each medical image block to be identified is determined from the image block set to be identified to obtain the reference image block set; the medical image block in the reference image block set is to be identified The medical image segments to be identified whose corresponding pathological auxiliary promotion indicators in the image segmentation set meet the recognition conditions; the recognition image segmentation set is determined based on the reference image segmentation set, and the recognition image segmentation in the recognition image segmentation set is suspicious Image segmentation; perform recognition of image segmentation based on the set of recognized image segments, and mark when conditions are met based on the recognition results.

In the embodiment of the present application, the medical image image to be segmented is divided into image blocks to obtain a set of image blocks to be identified corresponding to the medical image image to be segmented. The medical image segment to be identified satisfies the identification conditions according to the corresponding pathological auxiliary promotion index in the image block set to be identified. The blocks determine the reference image block set. The medical image blocks in the reference image block set are medical image blocks to be identified that are suitable for independent existence in the image block set to be identified. According to the suspicious image blocks in the reference image block set, The block determines the recognition image block set, and performs recognition image block recognition on the recognition image block in the recognition image block set. When it is deeply recognized based on the recognition result that the critical condition for marking is met, the image segmentation is marked. In this way, the recognition image block set corresponding to the medical image image to be segmented can be automatically identified to perform recognition image block recognition on the recognition image block set, which increases the efficiency of image segmentation. At the same time, the identified image blocks in the identified image block set are suspicious image blocks. The suspicious image blocks can be used to characterize new suspicious image blocks that may appear. Therefore, the suspicious image blocks are identified and the suspicious image blocks are identified based on the suspicious image blocks. The recognition results of image blocks are deeply recognized and marked when the critical conditions for marking are met. New suspicious image blocks that may appear can be pre-marked to accurately and quickly complete the segmentation of lesion images. At the same time, by combining the identified image blocks in the identified image block set with the matching medical image blocks in the reference lesion image set that match the identified image blocks, the suspicious image blocks can be restored and the identified image blocks can be restored. The identified image blocks in the set are combined with the matching medical image blocks in the reference lesion image set that match the identified image blocks. New suspicious image blocks can be obtained by combining the identified image blocks. The set of identified image blocks can be expanded and added. Recognition interval, classify the recognition image blocks according to the frequency of occurrence of the recognition image blocks in all three-dimensional medical imaging data, and match a preset mark reference value for each recognition type, which helps to match the recognition images with different frequency of occurrence Blocking to improve the accuracy of image block labeling.

Based on the foregoing embodiments, embodiments of the present application provide a lesion image segmentation device. Each unit included in the device and each module included in each unit can be implemented by a processor in a computer device; of course, it can also be implemented by Specific logic circuit implementation; during the implementation process, the processor can be a central processing unit (CPU), a microprocessor (Microprocessor Unit, MPU), a digital signal processor (Digital SignalProcessor, DSP) or field programmable Gate array (Field Programmable Gate Array, FPGA), etc.

Figure 2 is a schematic structural diagram of a lesion image segmentation device provided by an embodiment of the present application. As shown in Figure 2, the lesion image segmentation device 200 includes:

The image acquisition module 210 is used to acquire a medical image image to be segmented, perform an image segmentation operation on the medical image image to be segmented, and obtain a set of image segments to be identified corresponding to the medical image image to be segmented, wherein the medical image image to be segmented is Segmented medical imaging images are obtained through volume rendering;

The indicator determination module 220 is used to determine the pathological auxiliary promotion index of each medical image block to be identified in the set of image blocks to be identified, and use the pathology auxiliary promotion index of each medical image block to be identified to obtain the pathological auxiliary promotion index from the A reference image block set is determined from the set of image blocks to be identified; the medical image blocks in the reference image block set are to be identified whose corresponding pathological auxiliary promotion indicators in the set of image blocks to be identified satisfy the identification conditions. Medical image segmentation;

The block determination module 230 is configured to determine a set of identified image blocks according to the set of reference image blocks, and the identified image blocks in the set of identified image blocks are suspicious image blocks;

The block marking module 240 is configured to perform recognition image block recognition according to the recognition image block set, and perform image block marking when it is deeply recognized according to the recognition result that the critical condition for marking is met;

The image segmentation module 250 is configured to segment the medical image image to be segmented based on the image segment labeling result of each medical image segment to be identified.

The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules provided by the device provided by the embodiments of this application can be used to perform the methods described in the above method embodiments. For technical details not disclosed in the device embodiments of this application, please refer to the methods of this application. be understood from the description of the embodiments.

It should be noted that in the embodiments of the present application, if the above-mentioned artificial intelligence-based lesion image segmentation method is implemented in the form of a software function module and is sold or used as an independent product, it can also be stored in a computer-readable storage. in the medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of software products in essence or that contribute to related technologies. The software products are stored in a storage medium and include a number of instructions to enable a A computer device (which may be a personal computer, a server, a network device, etc.) executes all or part of the methods described in various embodiments of this application. The aforementioned storage media include: U disk, mobile hard disk, read only memory (Read Only Memory, ROM), magnetic disk or optical disk and other various media that can store program codes. In this way, the embodiments of the present application are not limited to any specific hardware, software, or firmware, or any combination of hardware, software, and firmware.

An embodiment of the present application provides a computer device, including a memory and a processor. The memory stores a computer program that can be run on the processor. When the processor executes the program, some or all of the steps in the above method are implemented.

Embodiments of the present application provide a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium may be transient or non-transitory.

Embodiments of the present application provide a computer program, which includes computer readable code. When the computer readable code is run in a computer device, the processor in the computer device executes a part for implementing the above method or All steps.

Embodiments of the present application provide a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, some of the above methods are implemented or All steps. The computer program product can be implemented specifically through hardware, software or a combination thereof. In some embodiments, the computer program product is embodied as a computer storage medium. In other embodiments, the computer program product is embodied as a software product, such as a software development kit (Software Development Kit, SDK) and so on.

It should be noted here that the above description of various embodiments tends to emphasize the differences between the various embodiments, and the similarities or similarities may be referred to each other. The description of the above embodiments of equipment, storage media, computer programs and computer program products is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the embodiments of the equipment, storage media, computer programs and computer program products of this application, please refer to the description of the method embodiments of this application for understanding.

Figure 3 is a schematic diagram of the hardware entity of a computer device provided by an embodiment of the present application. As shown in Figure 3, the hardware entity of the computer device 1000 includes: a processor 1001 and a memory 1002, where the memory 1002 stores information that can be stored in the processor. The computer program running on 1001 implements the steps in the method of any of the above embodiments when the processor 1001 executes the program.

The memory 1002 stores computer programs that can be run on the processor. The memory 1002 is configured to store instructions and applications executable by the processor 1001. It can also cache data to be processed or processed by the processor 1001 and each module in the computer device 1000. Data (for example, image data, audio data, voice communication data and video communication data) can be realized through flash memory (FLASH) or random access memory (Random Access Memory, RAM).

When the processor 1001 executes the program, the steps of any of the above artificial intelligence-based lesion image segmentation methods are implemented. Processor 1001 generally controls the overall operation of computer device 1000 .

Embodiments of the present application provide a computer storage medium. The computer storage medium stores one or more programs. The one or more programs can be executed by one or more processors to implement the artificial intelligence-based system as in any of the above embodiments. Steps of the lesion image segmentation method.

It should be pointed out here that the above description of the storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding. The above-mentioned processor can be an application specific integrated circuit (ASIC), a digital signal processor (DigitalSignal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), or a programmable logic device (Programmable Logic Device). , PLD), Field Programmable Gate Array (Field Programmable Gate Array, FPGA), Central Processing Unit (Central Processing Unit, CPU), controller, microcontroller, and microprocessor. It is understandable that the electronic device that implements the above processor function can also be other, which is not specifically limited in the embodiments of this application.

The above-mentioned computer storage medium/memory can be read-only memory (Read Only Memory, ROM), programmable read-only memory (Programmable Read-Only Memory, PROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Ferromagnetic Random Access Memory (FRAM), Flash Memory, Magnetic Surface Memory, Memories such as optical disks or Compact Disc Read-Only Memory (CD-ROM); it can also be various terminals including one or any combination of the above memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc. .

It will be understood that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic associated with the embodiment is included in at least one embodiment of the present application. Thus, the appearances of "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily referring to the same embodiment. Furthermore, the particular features, structures or characteristics may be combined in any suitable manner in one or more embodiments. It should be understood that in various embodiments of the present application, the size of the serial numbers of the above steps/processes does not mean the order of execution. The execution order of each step/process should be determined by its function and internal logic, and should not be The implementation process of the embodiments of this application does not constitute any limitations. The above serial numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments. It should be noted that, in this document, the terms "comprising", "comprises" or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or device that includes a series of elements not only includes those elements, It also includes other elements not expressly listed or inherent in the process, method, article or apparatus. Without further limitation, an element defined by the statement "comprises a..." does not exclude the presence of additional identical elements in a process, method, article or apparatus that includes that element.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components may be combined, or can be integrated into another system, or some features can be ignored, or not implemented. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the devices or units may be electrical, mechanical, or other forms. of.

The units described above as separate components may or may not be physically separated; the components shown as units may or may not be physical units; they may be located in one place or distributed to multiple network units; Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be separately used as a unit, or two or more units can be integrated into one unit; the above-mentioned integration The unit can be implemented in the form of hardware or in the form of hardware plus software functional units.

Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed through hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the execution includes: The steps of the above method embodiment; and the aforementioned storage media include: mobile storage devices, read-only memory (Read Only Memory, ROM), magnetic disks or optical disks and other various media that can store program codes.

Alternatively, if the integrated units mentioned above in this application are implemented in the form of software function modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or that contributes to related technologies. The computer software product is stored in a storage medium and includes a number of instructions to enable a computer. A computer device (which may be a personal computer, a server, a network device, etc.) executes all or part of the methods described in various embodiments of this application. The aforementioned storage media include: mobile storage devices, ROMs, magnetic disks or optical disks and other media that can store program codes.

The above are only embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person familiar with the technical field can easily think of changes or replacements within the technical scope disclosed in the present application. are covered by the protection scope of this application.

Claims (7)

1. An artificial intelligence-based lesion image segmentation method, characterized in that the method includes: Obtain the medical image image to be segmented, perform an image segmentation operation on the medical image image to be segmented, and obtain a set of image segments to be identified corresponding to the medical image image to be segmented, wherein the medical image image to be segmented is rendered through volume rendering get; Determine the pathological auxiliary promotion index of each to-be-identified medical image block in the to-be-identified image block set, and use the pathological auxiliary promotion index of each to-be-identified medical image block to obtain the pathological auxiliary promotion index from the to-be-identified image block set. Determine to obtain a reference image block set; the medical image blocks in the reference image block set are to-be-identified medical image blocks whose corresponding pathological auxiliary promotion indicators in the to-be-identified image block set satisfy the identification conditions; Wherein, determining the pathological auxiliary promotion index of each to-be-identified medical image block in the to-be-identified image block set includes: targeting any first block size in the to-be-identified image block set that is greater than a size threshold. A medical image block to be identified, determining an internal correlation evaluation index of the first medical image block to be identified and an external correlation evaluation index of the first medical image block to be identified, and determining an internal correlation evaluation index based on the first medical image block to be identified. The internal correlation evaluation index and the external correlation evaluation index of the first medical image block to be identified determine the pathological auxiliary promotion index of the first medical image block to be identified; wherein, based on the internal correlation of the first medical image block to be identified The evaluation index and the external correlation evaluation index determine the pathological auxiliary promotion index of the first medical image block to be identified, including: obtaining the first importance parameter corresponding to the internal correlation evaluation index of the first medical image block to be identified, and obtaining The second importance parameter corresponding to the internal correlation evaluation index of the first medical image block to be identified is based on the first importance parameter, the internal correlation evaluation index of the first medical image block to be identified, the second importance parameter, The external correlation evaluation index of the first medical image block to be identified is obtained to obtain the pathology auxiliary promotion index of the first medical image block to be identified; the calculation formula of the pathology auxiliary promotion index of the first medical image block to be identified is: G＝αI+βO Among them, G is the pathological auxiliary promotion index of the first medical image block to be identified, α is the first importance parameter, I is the internal correlation evaluation index of the first medical image block to be identified, and β is the second importance parameter. , O is the external correlation evaluation index of the first medical image block to be identified; For any second medical image block to be identified in the set of image blocks to be identified whose block size is less than or equal to the size threshold, determine the external correlation evaluation index of the second medical image block to be identified, and set The external correlation evaluation index of the second medical image block to be identified is used as the pathological auxiliary promotion index of the second medical image block to be identified; The identification image block set is determined according to the reference image block set, the identification image block in the identification image block set is a suspicious image block; the medical image image to be segmented is any of the target three-dimensional medical imaging data. A slice, the target three-dimensional medical image data is obtained at the current acquisition moment; determining the identification image block set according to the reference image block set includes: obtaining a reference lesion image set, the reference lesion image set is It is composed of normal image blocks and identification image blocks determined according to the three-dimensional medical image data acquired before the current acquisition moment; in the reference lesion image set, each medical image in the reference image block set is Matching is performed in blocks, and the set of recognized image blocks is determined based on the unsuccessfully matched medical image blocks; Recognition image segmentation is performed according to the recognition image segmentation set, and image segmentation is marked when the critical condition for marking is determined according to the depth of the recognition result; wherein the recognition image segmentation is performed according to the recognition image segmentation set. Block identification, including at least one of the following strategies: The first strategy, the target recognition image block is any recognition image block in the recognition image block set, and the first occurrence of the target recognition image block in the three-dimensional medical imaging data is obtained at R acquisition times. frequency, and the second frequency of occurrence in the three-dimensional medical imaging data obtained by acquiring the target recognition image segment at the target acquisition time, which is the acquisition time after the R acquisition times among the S acquisition times. ; Calculate the quotient between the first frequency of occurrence and the second frequency of occurrence; determine the quotient between the first frequency of occurrence and the second frequency of occurrence as the recognition result of the target recognition image segmentation ; Among them, the formula for identifying target recognition image blocks is: θ=(i+t)÷(j+t) Among them, θ is the recognition result of the target recognition image block, i is the first frequency of appearance in the three-dimensional medical imaging data obtained at R acquisition times, and j is the target recognition image block obtained at the target acquisition time. Obtain the second frequency of occurrence in the three-dimensional medical imaging data, t is a constant; S acquisition times include the current acquisition time and S-1 acquisition moments before the current acquisition time; R acquisition times include the current acquisition time and the current acquisition time before R-1 acquisition moments, R<S; the recognition results of the target recognition image block through strategy 2 can represent the first frequency of appearance of the target recognition image block at the R acquisition moments, compared with the target recognition image block in the target acquisition Increased value of the second frequency of occurrence within time; The second strategy is to obtain the first frequency of occurrence in the three-dimensional medical imaging data of the target recognition image segment at R acquisition times, and obtain the three-dimensional medical imaging data of the target recognition image segment at S acquisition times. the third frequency of occurrence in; calculate the quotient between the first frequency of occurrence and the third frequency of occurrence; determine the quotient between the first frequency of occurrence and the third frequency of occurrence as the target recognition Recognition results of image blocks; wherein, the S acquisition moments include the current acquisition moment and S-1 acquisition moments before the current acquisition moment; the R acquisition moments include the current acquisition moment and the Describe the R-1 acquisition times before the current acquisition time, R<S; The formula for identifying target recognition image blocks is: λ=i÷ε λ is the recognition result obtained by the target recognition image segmentation through strategy three, i is the first frequency of occurrence of the target recognition image segmentation in the three-dimensional medical imaging data obtained at R acquisition moments, ε is the target recognition image segmentation at S times. The third frequency of occurrence in the three-dimensional medical imaging data is obtained at the acquisition moment; the recognition result of the target recognition image segment through strategy three is used to represent the first frequency of occurrence of the target recognition image segment within the R acquisition moments accounting for the target recognition image segment. The proportion of the third occurrence frequency of a block within S acquisition moments; The medical image image to be segmented is segmented based on the image segment labeling result of each medical image segment to be identified. 2. The method according to claim 1, wherein the medical image image to be segmented is any slice in the target three-dimensional medical image data; the internal correlation evaluation of the first medical image segment to be identified is determined. Indicators, including: Determine the pixel cluster group corresponding to the first medical image segment to be identified. Any pixel cluster group is composed of multiple cut pixel clusters obtained by cutting the first medical image segment to be identified. Each cut pixel cluster is Composed of one pixel or multiple surrounding pixels in the first medical image block to be identified; For any pixel cluster group, obtain the confidence that each cut pixel cluster in the pixel cluster group appears in the target three-dimensional medical imaging data; According to the confidence that each cut pixel cluster appears in the target three-dimensional medical imaging data and the confidence that the first medical image block to be identified appears in the target three-dimensional medical imaging data, determine the pixel cluster grouping The corresponding internal correlation evaluation index; Determine the minimum internal correlation evaluation index among the internal correlation evaluation indicators corresponding to each pixel group grouping as the internal correlation evaluation index of the first to-be-identified medical image block; Determining the external correlation evaluation index of the first medical image block to be identified includes: Determine a neighborhood pixel set of the first to-be-identified medical image block in the target three-dimensional medical image data, where the neighborhood pixel set includes no less than one neighborhood pixel; Obtain the confidence that the fused medical image blocks to be identified, obtained by combining each neighborhood pixel and the first medical image block to be identified, respectively appear in the target three-dimensional medical image data; The external correlation evaluation index of the first medical image segment to be identified is determined based on the confidence of each fused medical image segment to be identified appearing in the target three-dimensional medical image data. 3. The method according to claim 2, wherein the neighborhood pixels included in the neighborhood pixel set are surrounding pixels of the first medical image block to be identified in the target three-dimensional medical image data; Obtaining the confidence that each neighborhood pixel is combined with the first medical image segment to be identified and the fused medical image segment to be identified appears in the target three-dimensional medical image data, includes acquiring the neighborhood The confidence that the first to-be-identified fused medical image block obtained by combining each surrounding pixel in the pixel set and the first to-be-identified medical image block appears in the target three-dimensional medical image data respectively; Determining the external correlation evaluation index of the first medical image segment to be identified based on the confidence that each fused medical image segment to be identified appears in the target three-dimensional medical image data includes: Obtain a plurality of first uncertainty measurement values based on the confidence of each first to-be-identified fused medical image segment appearing in the target three-dimensional medical image data; The minimum uncertainty metric value among the plurality of first uncertainty metric values is determined as the external correlation evaluation index of the first to-be-identified medical image block. 4. The method according to claim 1, characterized in that the target recognition image block is any recognition image block in the recognition image block set, and the method further includes: Match the matching medical image segment that matches the target recognition image segment in the reference lesion image set; if matched, determine the new recognition image through the target recognition image segment and the matching medical image segment. block, and iterating the set of recognition image blocks based on the newly added recognition image blocks; and iterating the set of reference lesion images based on the iterated set of recognition image blocks. 5. The method according to claim 1, characterized in that, performing image block marking when it is recognized based on the depth of the recognition result that the critical condition for marking is met, including: Get the preset mark reference value; When the quotient corresponding to the recognition result of the target recognition image segment is greater than or equal to the preset marking reference value, it is determined that the marking critical condition is met, and the corresponding target recognition image segment is marked accordingly; Wherein, obtaining the preset mark reference value includes: Obtain the frequency of occurrence of the target recognition image segment in the three-dimensional medical imaging data obtained at the current acquisition time and the previous time; Determine the recognition type corresponding to the target recognition image block according to the frequency of occurrence; Obtain the preset mark reference value corresponding to the recognition type; the recognition image blocks in the recognition image block set are classified into no less than one recognition type based on the frequency of occurrence of each recognition image segment, and one recognition type corresponds to A preset marker reference value. 6. A lesion image segmentation device, characterized in that it includes: An image acquisition module, configured to acquire a medical image image to be segmented, perform an image segmentation operation on the medical image image to be segmented, and obtain a set of image segments to be identified corresponding to the medical image image to be segmented, wherein the medical image image to be segmented is Medical imaging images are obtained through volume rendering; An index determination module, configured to determine the pathological auxiliary promotion index of each to-be-identified medical image block in the to-be-identified image block set, and use the pathological auxiliary promotion index of each of the to-be-identified medical image blocks to obtain the pathological auxiliary promotion index from the to-be-identified medical image block. A reference image block set is determined from the identified image block set; the medical image blocks in the reference image block set are to-be-identified medical images whose corresponding pathological auxiliary promotion indicators in the to-be-identified image block set satisfy the identification conditions. Image segmentation; wherein, determining the pathological auxiliary promotion index of each to-be-identified medical image segment in the to-be-identified image segment set includes: targeting a segment size in the to-be-identified image segment set that is greater than a size threshold Any first medical image block to be identified, determine the internal correlation evaluation index of the first medical image block to be identified and the external correlation evaluation index of the first medical image block to be identified, and based on the first medical image block to be identified The internal correlation evaluation index of the image block and the external correlation evaluation index of the first medical image block to be identified determine the pathological auxiliary promotion index of the first medical image block to be identified; wherein, based on the first medical image block to be identified, The internal correlation evaluation index and the external correlation evaluation index determine the pathological auxiliary promotion index of the first medical image block to be identified, including: obtaining the first importance corresponding to the internal correlation evaluation index of the first medical image block to be identified. parameters, and obtain the second importance parameter corresponding to the internal correlation evaluation index of the first medical image block to be identified, according to the first importance parameter, the internal correlation evaluation index of the first medical image block to be identified, the second The importance parameter and the external correlation evaluation index of the first medical image block to be identified are obtained to obtain the pathological auxiliary promotion index of the first medical image block to be identified; the pathological auxiliary promotion index of the first medical image block to be identified is obtained The calculation formula is: G＝αI+βO Among them, G is the pathological auxiliary promotion index of the first medical image block to be identified, α is the first importance parameter, I is the internal correlation evaluation index of the first medical image block to be identified, and β is the second importance parameter. , O is the external correlation evaluation index of the first medical image block to be identified; For any second medical image block to be identified in the set of image blocks to be identified whose block size is less than or equal to the size threshold, determine the external correlation evaluation index of the second medical image block to be identified, and set The external correlation evaluation index of the second medical image block to be identified is used as the pathological auxiliary promotion index of the second medical image block to be identified; A block determination module, configured to determine a recognition image block set based on the reference image block set, where the recognition image blocks in the recognition image block set are suspicious image blocks; the medical image image to be segmented is the target Any slice in the three-dimensional medical image data, the target three-dimensional medical image data is obtained at the current acquisition moment; determining the identification image block set according to the reference image block set includes: obtaining a reference lesion image set, The reference lesion image set is composed of normal image segments and identification image segments determined based on three-dimensional medical imaging data acquired before the current acquisition time; the reference image segmentation is performed in the reference lesion image set. Each medical image segment in the set is matched, and the set of identified image segments is determined based on the unsuccessfully matched medical image segments; A block marking module, configured to perform recognition image block recognition according to the recognition image block set, and perform image block marking when it is deeply recognized according to the recognition result that the critical condition for marking is met; the recognition image block according to the recognition result The set is used for recognition image segmentation recognition, including at least one of the following strategies: The first strategy, the target recognition image block is any recognition image block in the recognition image block set, and the first occurrence of the target recognition image block in the three-dimensional medical imaging data is obtained at R acquisition times. frequency, and the second frequency of occurrence in the three-dimensional medical imaging data obtained by acquiring the target recognition image segment at the target acquisition time, which is the acquisition time after the R acquisition times among the S acquisition times. ; Calculate the quotient between the first frequency of occurrence and the second frequency of occurrence; determine the quotient between the first frequency of occurrence and the second frequency of occurrence as the recognition result of the target recognition image segmentation ; Among them, the formula for identifying target recognition image blocks is: θ=(i+t)÷(j+t) Among them, θ is the recognition result of the target recognition image block, i is the first frequency of appearance in the three-dimensional medical imaging data obtained at R acquisition times, and j is the target recognition image block obtained at the target acquisition time. Obtain the second frequency of occurrence in the three-dimensional medical imaging data, t is a constant; S acquisition times include the current acquisition time and S-1 acquisition moments before the current acquisition time; R acquisition times include the current acquisition time and the current acquisition time before R-1 acquisition moments, R<S; the recognition results of the target recognition image block through strategy 2 can represent the first frequency of appearance of the target recognition image block at the R acquisition moments, compared with the target recognition image block in the target acquisition Increased value of the second frequency of occurrence within time; The second strategy is to obtain the first frequency of occurrence in the three-dimensional medical imaging data of the target recognition image segment at R acquisition times, and obtain the three-dimensional medical imaging data of the target recognition image segment at S acquisition times. the third frequency of occurrence in; calculate the quotient between the first frequency of occurrence and the third frequency of occurrence; determine the quotient between the first frequency of occurrence and the third frequency of occurrence as the target recognition Recognition results of image blocks; wherein, the S acquisition moments include the current acquisition moment and S-1 acquisition moments before the current acquisition moment; the R acquisition moments include the current acquisition moment and the Describe the R-1 acquisition times before the current acquisition time, R<S; The formula for identifying target recognition image blocks is: λ=i÷ε λ is the recognition result obtained by the target recognition image segmentation through strategy three, i is the first frequency of occurrence of the target recognition image segmentation in the three-dimensional medical imaging data obtained at R acquisition moments, ε is the target recognition image segmentation at S times. The third frequency of occurrence in the three-dimensional medical imaging data is obtained at the acquisition moment; the recognition result of the target recognition image segment through strategy three is used to represent the first frequency of occurrence of the target recognition image segment within the R acquisition moments accounting for the target recognition image segment. The proportion of the third occurrence frequency of a block within S acquisition moments; An image segmentation module, configured to segment the medical image image to be segmented based on the image segment labeling result of each medical image segment to be identified. 7. A computer-readable storage medium with a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps in the method of any one of claims 1 to 5 are implemented.