CN111369525A - Image analysis method, apparatus and storage medium - Google Patents

Image analysis method, apparatus and storage medium Download PDF

Info

Publication number
CN111369525A
CN111369525A CN202010137683.0A CN202010137683A CN111369525A CN 111369525 A CN111369525 A CN 111369525A CN 202010137683 A CN202010137683 A CN 202010137683A CN 111369525 A CN111369525 A CN 111369525A
Authority
CN
China
Prior art keywords
coronary artery
image
coronary
detected
distance
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010137683.0A
Other languages
Chinese (zh)
Other versions
CN111369525B (en
Inventor
董昢
吴迪嘉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Lianying Intelligent Medical Technology Beijing Co ltd
Original Assignee
Lianying Intelligent Medical Technology Beijing Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Lianying Intelligent Medical Technology Beijing Co ltd filed Critical Lianying Intelligent Medical Technology Beijing Co ltd
Priority to CN202010137683.0A priority Critical patent/CN111369525B/en
Publication of CN111369525A publication Critical patent/CN111369525A/en
Application granted granted Critical
Publication of CN111369525B publication Critical patent/CN111369525B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/68—Analysis of geometric attributes of symmetry
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30101—Blood vessel; Artery; Vein; Vascular
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

The present application relates to an image analysis method, apparatus, and storage medium. The method comprises the following steps: detecting a coronary image to be detected to obtain reference information of a coronary in the coronary image to be detected; the reference information is related to position information of an atrium and a ventricle surrounded by a coronary artery in the coronary image to be detected; performing segmentation processing on the coronary artery image to be detected to obtain segmentation images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmentation images of the left and right coronary arteries; analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; and the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected. The method can save labor.

Description

Image analysis method, apparatus and storage medium
Technical Field
The present application relates to the field of image processing technologies, and in particular, to an image analysis method, an image analysis device, and a storage medium.
Background
The heart is used as a muscular power organ for pumping blood, and the heart also needs enough nutrition and energy, namely a coronary artery and a vein, namely coronary circulation, the coronary artery is an artery for supplying heart blood, starts from the root of an aorta, is divided into a left branch and a right branch, runs on the surface of the heart, and can be divided into three types of dominant types according to the source of a blood supply vessel (a rear descending branch) of the back one third of the ventricular interval, namely a right dominant type, a left dominant type and a balanced type. The dominant type is simply the right and left coronary arteries of the heart that take more myocardial blood supply. Aiming at the coronary artery distribution with different dominant types, the processing means adopted by doctors are different, so that the judgment of the dominant type of the coronary artery is very important.
In the related art, when determining the dominant coronary artery, most clinicians determine the acquired coronary artery image based on experience to obtain the determination result of the dominant coronary artery.
However, the above technique has a problem of being labor-consuming.
Disclosure of Invention
In view of the above, it is necessary to provide an image analysis method, an apparatus, and a storage medium capable of saving manpower in view of the above technical problems.
A method of image analysis, the method comprising:
detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of an atrium and a ventricle surrounded by the coronary artery in the coronary artery image to be detected;
segmenting the coronary artery image to be detected to obtain segmented images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmented images of the left and right coronary arteries;
analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
In one embodiment, the detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected includes:
detecting and processing the atrioventricular intersection point in the coronary image to be detected by adopting a preset detection model to obtain the position information of the atrioventricular intersection point in the coronary image to be detected, and taking the position information of the atrioventricular intersection point as the reference information of the coronary;
the detection model is obtained by training based on a first sample coronary image and a gold standard image corresponding to the first sample coronary image, the gold standard image corresponding to the first sample coronary image comprises a position mark of an atrioventricular intersection point corresponding to the first sample coronary image, and the atrioventricular intersection point is a point at a junction of a central vein between ventricles and a coronary sinus of the heart room.
In one embodiment, the analyzing the coronary image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result includes:
calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the left coronary artery to obtain a first distance; calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the right coronary artery to obtain a second distance;
and analyzing the coronary artery image to be detected according to the first distance, the second distance and a preset distance threshold range to obtain an analysis result.
In one embodiment, the analyzing the coronary artery image to be detected according to the first distance, the second distance and the preset distance threshold range to obtain an analysis result includes:
matching the first distance and the second distance with a first distance threshold range; if the first distance and the second distance do not exceed the first distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type according to the analysis result;
or, matching the first distance with a second distance threshold range; if the first distance does not exceed the second distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a left dominant type;
or matching the second distance with a third distance threshold range; and if the second distance does not exceed the third distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is the right dominant type.
In one embodiment, the first distance threshold range includes a left distance threshold range and a right distance threshold range, and the matching of the first distance and the second distance with the first distance threshold range; if the first distance and the second distance do not exceed the first distance threshold range, determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type, and the method comprises the following steps:
matching the first distance with the threshold range of the first distance of the left branch, and matching the second distance with the threshold range of the first distance of the right branch;
and if the first distance does not exceed the first distance threshold range of the left branch and the second distance does not exceed the first distance threshold range of the right branch, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type.
In one embodiment, the detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected includes:
performing segmentation processing on the coronary artery image to be detected to obtain a segmentation result of the heart chamber; the segmentation result of the heart chamber comprises left and right atria and left and right ventricles of the heart;
establishing an atrioventricular groove coordinate system according to the segmentation result of the heart chamber, and taking the atrioventricular groove coordinate system as the reference information of the coronary artery; the transverse axis direction of the atrioventricular groove coordinate system is the direction of the heart atrioventricular groove boundary line, and the longitudinal axis direction of the atrioventricular groove coordinate system is the direction of the heart ventricular groove boundary line.
In one embodiment, the analyzing the coronary image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result includes:
calculating the coordinates of each point on the central line of the left coronary artery under the atrioventricular groove coordinate system to obtain the coordinates of at least one left fulcrum, and calculating the coordinates of each point on the central line of the right coronary artery under the atrioventricular groove coordinate system to obtain the coordinates of at least one right fulcrum;
and analyzing the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and a preset coordinate threshold range to obtain an analysis result.
In one embodiment, the analyzing the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and the preset threshold range of coordinates to obtain an analysis result includes:
respectively matching the coordinates of at least one left fulcrum and the coordinates of at least one right fulcrum with the coordinate threshold range to obtain the number of left supporting target points and the number of right supporting target points, comparing the number of the left supporting target points and the number of the right supporting target points with a preset number threshold, and obtaining an analysis result according to the comparison result;
the left supporting target point is a point of which the coordinate in at least one left supporting point does not exceed the range of the coordinate threshold, and the right supporting target point is a point of which the coordinate in at least one right supporting point does not exceed the range of the coordinate threshold.
In one embodiment, if the coordinate threshold range includes a left-branch first coordinate threshold range and a right-branch first coordinate threshold range, the number threshold includes a left-branch first number threshold and a right-branch first number threshold, the matching between the coordinate of at least one left fulcrum and the coordinate of at least one right fulcrum and the coordinate threshold range to obtain the number of left-branch target points and the number of right-branch target points, comparing the number of left-branch target points and the number of right-branch target points with a preset number threshold, and obtaining an analysis result according to a comparison result, the method includes:
matching the coordinate of at least one left fulcrum with the threshold range of the first left coordinate to obtain the number of first target points of the left fulcrum, and matching the coordinate of at least one right fulcrum with the threshold range of the first right coordinate to obtain the number of first target points of the right fulcrum;
comparing the number of the first target points of the left branch with a first threshold value of the left branch, and comparing the number of the first target points of the right branch with a first threshold value of the right branch;
and if the number of the left first target points is greater than the left first number threshold and the number of the right first target points is greater than the right first number threshold, determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type.
In one embodiment, if the coordinate threshold range includes a left second coordinate threshold range and a right second coordinate threshold range, and the number threshold includes a left second number threshold and a right second number threshold, the matching the coordinate of at least one left fulcrum and the coordinate of at least one right fulcrum with the coordinate threshold range respectively to obtain the number of left branch target points and the number of right branch target points, comparing the number of left branch target points and the number of right branch target points with a preset number threshold, and obtaining an analysis result according to a comparison result, the method includes:
matching the coordinates of at least one left fulcrum with the threshold range of the second left coordinates to obtain the number of second left target points, comparing the number of the second left target points with the threshold of the second left number, and determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a left dominant type if the number of the second left target points is greater than the threshold of the second left number; alternatively, the first and second electrodes may be,
and matching the coordinate of at least one right pivot with the threshold range of the right second coordinate to obtain the number of right second target points, comparing the number of the right second target points with the threshold of the right second number, and determining that the dominant type of the coronary artery in the coronary artery image to be detected is the right dominant type if the number of the right second target points is greater than the threshold of the right second number.
In one embodiment, the segmenting the to-be-detected coronary artery image to obtain segmented images of left and right coronary arteries corresponding to the to-be-detected coronary artery image includes:
inputting the coronary artery image to be detected into a preset segmentation model to obtain a probability map of left and right coronary arteries corresponding to the coronary artery image to be detected; the pixel values of the positions on the probability graph of the left and right coronary arteries are the probability that the pixel values of the corresponding positions on the coronary artery image to be detected belong to the left and right coronary arteries, the segmentation model is obtained by training based on a second sample coronary artery image and a gold standard image corresponding to the second sample coronary artery image, and the gold standard image corresponding to the second sample coronary artery image comprises left and right coronary artery marks corresponding to the second sample coronary artery image;
and carrying out binarization processing on the probability maps of the left and right coronary arteries according to a preset first probability threshold value to obtain segmentation images of the left and right coronary arteries corresponding to the probability maps of the left and right coronary arteries.
In one embodiment, the detecting the atrioventricular intersection point in the coronary artery image to be detected by using the preset detection model to obtain the position information of the atrioventricular intersection point in the coronary artery image to be detected includes:
inputting the coronary artery image to be detected into a detection model to obtain a probability map of the atrioventricular intersection point corresponding to the coronary artery image to be detected; the pixel value of each position on the probability graph of the atrioventricular intersection point is the probability that the pixel value of the corresponding position on the coronary image to be detected belongs to the atrioventricular intersection point;
carrying out binarization processing on the probability map of the chamber intersection point according to a preset second probability threshold value to obtain a binarization mask image corresponding to the probability map of the chamber intersection point;
labeling connected domains in the binary mask image, and determining the maximum connected domain according to the labeled connected domains;
and acquiring a weighted central point of the probability value corresponding to the maximum connected domain, and determining the position information of the weighted central point as the position information of the chamber intersection point.
An image analysis apparatus, the apparatus comprising:
the detection module is used for detecting and processing the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of an atrium and a ventricle surrounded by the coronary artery in the coronary artery image to be detected;
the segmentation module is used for performing segmentation processing on the coronary artery image to be detected to obtain segmentation images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmentation images of the left and right coronary arteries;
the analysis module is used for analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of an atrium and a ventricle surrounded by the coronary artery in the coronary artery image to be detected;
segmenting the coronary artery image to be detected to obtain segmented images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmented images of the left and right coronary arteries;
analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of an atrium and a ventricle surrounded by the coronary artery in the coronary artery image to be detected;
segmenting the coronary artery image to be detected to obtain segmented images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmented images of the left and right coronary arteries;
analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
According to the image analysis method, the image analysis device, the computer equipment and the storage medium, the coronary artery image to be detected is detected to obtain the reference information of the coronary artery, the reference information of the coronary artery is related to the position information of the ventricle and the atrium surrounded by the coronary artery, the coronary artery image to be detected is segmented and the midline is extracted for processing to obtain the midlines of the left and right coronary arteries, and the coronary artery image to be detected is analyzed based on the reference information of the coronary artery and the midlines of the left and right coronary arteries to obtain the analysis result of the dominant type category capable of representing the coronary artery. In the method, the coronary image can be detected, and the dominant type category of the coronary is obtained based on the obtained reference information of the coronary and the central lines of the left and right coronary, so that the dominant type category does not need to be judged manually, and therefore, the method can save the labor cost to a certain extent; in addition, because the coronary image is detected by the computer equipment to obtain the dominant type category, compared with the method of manually judging the dominant type category according to experience, the obtained judging result of the dominant type category is more accurate, and meanwhile, the judging process of the dominant type category is quicker, so that the judging time of the dominant type category can be saved.
Drawings
FIG. 1 is a diagram of the internal structure of one embodiment of a computer device;
FIG. 2 is a schematic flow chart diagram of a method for image analysis in one embodiment;
FIG. 3a is a schematic view of an atrioventricular intersection marked on a first sample coronary image;
FIG. 3b is a schematic flow chart of a method for image analysis in another embodiment;
FIG. 4 is a schematic flow chart diagram of a method for image analysis in another embodiment;
FIG. 5a is a schematic flow chart of a method for image analysis in another embodiment;
FIG. 5b is a schematic view of an atrioventricular groove coordinate system established on a heart chamber;
FIG. 6 is a schematic flow chart diagram illustrating a method for image analysis in accordance with another embodiment;
FIG. 7a is a schematic flow chart of a method for image analysis in another embodiment;
FIG. 7b is a schematic illustration of left and right coronary vessels marked on a second sample coronary image;
fig. 8 is a block diagram showing the structure of an image analysis apparatus according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The coronary arteries can be divided into three types according to the source of blood supply vessels (posterior descending branches) in the posterior one third of the heart ventricular septum, namely a right dominant type, a left dominant type and a balanced type, namely if the blood supply vessels of the posterior descending branches are from the Right Coronary Artery (RCA), the coronary arteries are of the right dominant type; left dominant if the posterior descending blood supply vessel is from the left circumflex coronary artery (LCX); the posterior descending branch is balanced if the blood supply vessels are derived from both the Right Coronary Artery (RCA) and the left circumflex coronary artery (LCX). Currently, in clinical work, doctors determine the left and right dominant coronary arteries according to the origin of posterior descending branches and the distribution of the left and right coronary arteries in the septal plane. The right dominant right coronary artery crosses the cross, and sends out the posterior descending branch at the septal plane and sends out the right posterior lateral branch; the left dominant posterior descending branch is emitted by the left crown circumflex branch and the left crown circumflex over the cross; the balanced left and right coronary arteries are distributed in the septal plane of the heart in a balanced manner, and do not cross the crossing of the atrioventricular intersection, and the posterior descending branch can be sent out by the right coronary artery or come from the coronary arteries on both sides. The judgment method has higher requirements on the anatomical experience of doctors, particularly for the condition of great coronary artery vessel variation, the artificial experience is very important, and the primary doctor often has the condition of judgment error due to insufficient reading experience; in addition, manual judgment, especially when doctors process a lot of reports, needs to consume a lot of efforts of the doctors, and increases the risk of human misjudgment. The present application provides an image detection method, apparatus, device and storage medium, which can solve the above technical problems.
The image detection method provided by the embodiment can be applied to a computer device, which can be a terminal or a server, and can perform wired or wireless communication with a medical scanning device. Taking a computer device as an example, the internal structure diagram thereof can be as shown in fig. 1. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless communication can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement an image detection method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 1 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
The execution subject of the embodiment of the present application may be a computer device, or may be an image detection apparatus, and the following description will be given taking a computer device as an execution subject.
In one embodiment, an image detection method is provided, and the embodiment relates to a specific process of detecting and analyzing a coronary image and obtaining a coronary dominant type category based on the analysis result of the coronary image. As shown in fig. 2, the method may include the steps of:
s202, detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of the atrium and ventricle surrounded by the coronary artery in the image of the coronary artery to be detected.
The atrium and ventricle surrounded by the coronary artery generally include left atrium, left ventricle, right atrium and right ventricle, the reference information may be information related to position information of the left atrium and the right atrium and position information of the left ventricle and the right ventricle, for example, the position of an intersection of the ventricle and the atrium, or a coordinate system of the ventricle and the atrium established according to the position of the ventricle and the position of the atrium, or other information related to the position of the ventricle and the atrium, which is not limited in this embodiment. In addition, the coronary artery image to be detected may be a one-dimensional image, a two-dimensional image, a three-dimensional image, a four-dimensional image, or the like, the embodiment mainly uses a three-dimensional image, and the modality of the coronary artery image to be detected may be a CT image, a PET image, an MR image, or the like, for example, when the coronary artery image to be detected is a CT image, the coronary artery image to be detected may be a CTA image; the coronary arteries in the image to be detected generally include both the left coronary artery and the right coronary artery.
Specifically, before detecting the object to be detected, a scanning device (not limited to a scanning device such as CT, PET, MR, etc.) may be used to scan the chest or the whole body of the object to be detected to obtain a chest image of an image to be detected, where the chest image includes coronary artery and heart and may be recorded as a coronary artery image to be detected; then, the computer device can detect the coronary artery and the heart in the coronary artery image to be detected, and obtain the position information related to the coronary artery and the ventricle atrium, and the position information is recorded as the reference information of the coronary artery.
S204, the coronary artery image to be detected is segmented to obtain segmented images of left and right coronary arteries corresponding to the coronary artery image to be detected, and central lines of the left and right coronary arteries of the segmented images of the left and right coronary arteries are extracted.
When the coronary artery to be detected is segmented, the coronary artery in the coronary artery to be detected can be segmented by adopting a segmentation model, an image segmentation algorithm and the like to obtain a left coronary artery and a right coronary artery. After the coronary artery image to be detected is obtained, the coronary artery image to be detected can be segmented firstly, and then the detection processing of the S202 is carried out; of course, the detection process of S202 may be performed first, and then the segmentation process of this step may be performed; of course, the detection process of S202 and the segmentation process of this step may be performed simultaneously, and the sequence of S202 and this step is not specifically limited in this embodiment. In addition, when extracting the central lines of the left and right coronary arteries, the central lines may be extracted by a skeletonization method, or may be extracted by a neural network, or may be extracted by another method, which is not particularly limited in this embodiment. In addition, the midline of the left and right coronary arteries may be the centerline of the left and right coronary arteries, or may be a line that can characterize the left and right coronary arteries, such as the line near the outer wall of the left and right coronary arteries.
Specifically, after the computer device obtains the coronary artery image to be detected in S202, the computer device may use a segmentation model, an image segmentation algorithm, and the like to segment the coronary artery in the coronary artery to be detected, so as to obtain a segmentation image of the left coronary artery and a segmentation image of the right coronary artery, and then extract a centerline of the segmentation image of the left coronary artery and a centerline of the segmentation image of the right coronary artery respectively, so as to obtain a centerline of the left coronary artery corresponding to the segmentation image of the left coronary artery and a centerline of the right coronary artery corresponding to the segmentation image of the right coronary artery.
S206, analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
The dominant coronary artery type includes balanced type, left dominant type and right dominant type.
Specifically, after obtaining the reference information of the coronary artery, the central line of the left coronary artery and the central line of the right coronary artery, the computer device may obtain an analysis result by analyzing the relative position between the ventricular atrium position and the central lines of the left and right coronary arteries based on the reference information of the coronary artery, and obtain the dominant type category based on the analysis result. For example, if the reference information of the coronary artery is closer to the relative position of the ventricular atrium and the midline of the left coronary artery, the coronary artery may be considered as a left dominant type, if the reference information of the coronary artery is closer to the relative position of the ventricular atrium and the midline of the right coronary artery, the coronary artery may be considered as a right dominant type, and if the reference information of the coronary artery is not as much as the relative position of the ventricular atrium and the midline of the left and right coronary arteries, the coronary artery may be considered as a balanced type. After the dominant type category of the coronary artery in the coronary artery image to be detected is obtained, the corresponding dominant type category can be marked on the radiograph interpretation of the object to be detected, and meanwhile, a doctor can also process the object to be detected by adopting a targeted processing means according to the dominant type category so as to obtain a better processing result.
In the image analysis method, the coronary artery image to be detected is detected to obtain the reference information of the coronary artery, the reference information of the coronary artery is related to the position information of the ventricle and the atrium surrounded by the coronary artery, the coronary artery image to be detected is segmented and the midline is extracted for processing to obtain the midline of the left and right coronary arteries, and the coronary artery image to be detected is analyzed based on the reference information of the coronary artery and the midline of the left and right coronary arteries to obtain the analysis result of the dominant type category which can represent the coronary artery. In the method, the coronary image can be detected, and the dominant type category of the coronary is obtained based on the obtained reference information of the coronary and the central lines of the left and right coronary, so that the dominant type category does not need to be judged manually, and therefore, the method can save the labor cost to a certain extent; in addition, because the coronary image is detected by the computer equipment to obtain the dominant type category, compared with the method of manually judging the dominant type category according to experience, the obtained judging result of the dominant type category is more accurate, and meanwhile, the judging process of the dominant type category is quicker, so that the judging time of the dominant type category can be saved.
It should be noted that there may be two types of reference information for obtaining the coronary artery, one is the position information of the atrioventricular intersection point, and the other is the atrioventricular groove coordinate system, and the dominant type category of the coronary artery is analyzed with respect to the reference information in the two cases. First, a case where the reference information is position information of a room intersection will be described below.
In another embodiment, another image detection method is provided, and this embodiment relates to a specific process of how to detect a coronary artery image to be detected by using a detection model to obtain position information of an atrioventricular intersection point in the coronary artery image. On the basis of the above embodiment, the above S202 may include the following step a:
step A, detecting atrioventricular intersection points in a coronary image to be detected by adopting a preset detection model to obtain position information of the atrioventricular intersection points in the coronary image to be detected, and taking the position information of the atrioventricular intersection points as reference information of the coronary; the detection model is obtained by training based on a first sample coronary image and a gold standard image corresponding to the first sample coronary image, the gold standard image corresponding to the first sample coronary image comprises a position mark of an atrioventricular intersection point corresponding to the first sample coronary image, and the atrioventricular intersection point is a point at a junction of a central vein between ventricles and a coronary sinus of the heart room.
In step a, the central vein refers to a longitudinal vein between the left ventricle and the right ventricle of the septal surface of the heart, the coronary sinus refers to a longitudinal venous sinus between the left atrium and the right atrium of the septal surface of the heart, the septal surface of the heart refers to the back surface of the heart, a plurality of points may exist at the junction of the central vein and the coronary sinus, one of the points may be selected as an atrioventricular intersection point, and which point at the junction is selected may be determined according to actual conditions.
The position information of the room intersection may be coordinates of the room intersection, and the coordinates of the room intersection may be one-dimensional coordinates, two-dimensional coordinates, three-dimensional coordinates, or the like.
In addition, before the detection model is used, the detection model also needs to be trained, and the training process of the detection model may include the following three steps:
1) a training data set is generated. Marking N atrioventricular intersection points P in N first sample coronary images by using labeling softwareiStoring coordinates (wherein i refers to an index of a atrioventricular intersection point on each first sample coronary image, the range of the index is 1-N, namely, one atrioventricular intersection point is marked on each first sample coronary image), then generating N spherical binary mask images with the radius of r and the coordinates of each atrioventricular intersection point as the center of a circle in the N first sample coronary images, and recording the N spherical binary mask images as gold standard images corresponding to each first sample coronary image (namely, each first sample coronary image corresponds to one gold standard image), wherein the gold standard images and the first sample coronary images are paired to form a training data set, and the number of the training data sets is also N; in addition, the modality of the first sample coronary image is the same as that of the coronary image to be detected, the sizes of r and N can be determined according to actual conditions, for example, r can be 6 pixels, and the number of N can be 1000, 2000, and the like. For example, refer to fig. 3a, which is a schematic diagram of atrioventricular intersection points marked on a first sample coronary image, and the four diagrams are respectively atrioventricular intersection points marked on a sagittal plane, a coronal plane, a horizontal plane and a three-dimensional solid (it should be noted that the four diagrams are only an example and do not affect the essence of the embodiment of the present application).
2) And establishing a detection model, wherein the detection model can be a convolutional neural network model. Constructing a convolutional neural network model, and setting hyper-parameters of the convolutional neural network model, wherein an input channel of the convolutional neural network is 1, the input channel is a first sample coronary image, an output channel is 2, and the input channel and the output channel are respectively a detection probability map of N atrioventricular intersection points corresponding to the first sample coronary image and a detection probability map of a background; in addition, the training data set can be divided into a training set X1, a verification set X2 and a test set X3, wherein the training set, the verification set and the test set are independent of each other, the number of the training set, the number of the verification set and the number of the test set are N1, N2 and N3, the training set, the verification set and the test set are natural numbers, N1+ N2+ N3 is equal to N, and N1 is equal to or more than 1/2N; for example, assuming that the number of first sample coronary images is 1000, n 1-500, n 2-200, and n 3-300 may be used.
3) And training a detection model. Training the detection model established in the step 2) by using the training data set generated in the step 1), wherein a training set X1 is used for training the detection model, a verification set X2 is used for evaluating the current performance of the model, and a test set X3 is used for checking the generalization performance of the model; in the training process, the training set is divided into a plurality of batches (for example, 100 rounds of training in 100 batches) and input into the detection model repeatedly for training a plurality of rounds, meanwhile, the difference between the output detection probability image of the crossing point of the atrioventricular system and the gold standard image is calculated by using a cost function, the difference is used as a training error and fed back to the detection model, and the model parameters are updated by a learning algorithm; and after the training of each batch is finished, performing performance test on the detection model by using the verification set, considering that the training of the detection model is finished when the performance test indexes tend to be stable, and storing the trained network model. In addition, the network structure of the detection model may include an input module, two down-sampling modules, two up-sampling modules and an output module, except for the output module, the other modules all use a batch normalization layer and a nonlinear activation function Relu, and the nonlinear activation function in the output module uses softmax, the output value of which is within a (0,1) interval, it should be noted that the last softmax is performed between each output channel, so that the sum of corresponding position elements in each probability map finally output is 1, and they respectively represent the probability that the pixel at the current position in the original first sample coronary image belongs to each background or foreground.
In addition, the detection model can be a deep convolutional neural network CNN, a generative countermeasure network GAN, convolutional neural networks U-Net and V-Net, a recurrent neural network RNN, or the like; the hyper-parameters may include the number of network layers, convolution kernels, learning rate, parameter initialization, number of training rounds, and batch size. The cost function may be a set similarity metric function (Dice loss) or a focus loss function (Focal loss), and then one of a Stochastic Gradient Descent (SGD), an Adaptive Moment Estimation optimization algorithm (Adam), and a Momentum algorithm (Momentum) may also be used to minimize a training error to train the detection model.
When the trained detection model is used to detect the coronary artery image to be detected to obtain the position information of the atrioventricular intersection, as shown in fig. 3b, optionally, the step a may include the following steps S302-S308:
s302, inputting the coronary artery image to be detected into a detection model to obtain a probability map of the atrioventricular intersection point corresponding to the coronary artery image to be detected; and the pixel value of each position on the probability graph of the atrioventricular intersection point is the probability that the pixel value of the corresponding position on the coronary image to be detected belongs to the atrioventricular intersection point.
And S304, carrying out binarization processing on the probability map of the chamber intersection point according to a preset second probability threshold value to obtain a binarization mask image corresponding to the probability map of the chamber intersection point.
S306, labeling the connected domains in the binary mask image, and determining the maximum connected domain according to the labeled connected domains.
S308, acquiring a weighted center point of the probability value corresponding to the maximum connected domain, and determining the position information of the weighted center point as the position information of the chamber intersection point.
In S302-S308, the size of the second probability threshold may be determined according to practical situations, and may be, for example, 0.2, 0.3, 0.35, and so on.
Specifically, after obtaining the coronary artery image to be detected, the computer device may input the coronary artery image to be detected into the detection model to obtain a probability map of the atrioventricular intersection point corresponding to the coronary artery image to be detected, then compare the second probability threshold with probability values of respective positions on the probability map of the atrioventricular intersection point, set the probability value at a position on the probability map where the probability value is smaller than the second probability threshold to 0, set the probability value at a position on the probability map where the probability value is greater than or equal to the second probability threshold to 1, and obtain a probability map after binarization, which is marked as a binarization mask image; then, the connected domains with the probability values of 1 on the mask image can be marked (for example, the connected domains with 1 can be marked with different colors), the connected domain with the maximum area in the connected domains with the probability values of 1 is found out from the connected domains, the connected domain with the maximum area is used as the maximum connected domain, and then the maximum connected domain corresponds to a probability map of the intersection point of the original chamber and the room output by the detection model, so that the probability value and the coordinate of the corresponding position of the maximum connected domain are obtained; or corresponding each connected domain with the probability value of 1 to a probability graph of the original room-room intersection point according to the corresponding position, finding the connected domain with the maximum probability density of each connected domain on the probability graph of the original room-room intersection point, taking the connected domain with the maximum probability density as the maximum connected domain, then obtaining the probability value and the coordinate of the corresponding position of the maximum connected domain, then taking the probability value as the weight, carrying out weighted summation on the coordinates corresponding to the weights, obtaining the weighted summed coordinates, and recording the weighted summed coordinates as the position information of the room-room intersection point; the probability value on the probability map of the intersection of the original atrioventricular system obtained by the detection model is not a binary probability value but an actually calculated probability value, and may be, for example, 0.95, 0.88, 0.02, 0.23, 0.45, or the like.
In the image analysis method provided by this embodiment, a preset detection model may be adopted to perform detection processing on the atrioventricular intersection point in the coronary artery image to be detected, so as to obtain position information of the atrioventricular intersection point in the coronary artery image to be detected, and the position information of the atrioventricular intersection point is used as reference information of the coronary artery; the detection model is obtained by training based on the first sample coronary image and the position mark of the atrioventricular intersection point corresponding to the first sample coronary image. In this embodiment, since the trained detection model can be used to obtain the position information of the atrioventricular intersection point on the coronary artery image to be detected, and the trained detection model is trained based on the position mark of the atrioventricular intersection point corresponding to the sample coronary artery image, the trained detection model is accurate, so that the analysis result obtained when the dominant type of the coronary artery is analyzed subsequently by using the position information of the atrioventricular intersection point is also accurate.
In another embodiment, another image detection method is provided, and this embodiment relates to a specific process of analyzing the coronary artery image to be detected based on the above-mentioned position information of the atrioventricular intersection and the central line of the coronary artery to obtain the analysis result of the dominant type. On the basis of the above embodiment, as shown in fig. 4, the above S206 may include the following steps:
s402, calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the left coronary artery to obtain a first distance; and calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the right coronary artery to obtain a second distance.
Specifically, when the distance is calculated, the position information of the intersection point of the atria on the coronary artery image to be detected and the distance between each point on the central line of the left coronary artery can be calculated to obtain a plurality of distances, and the shortest distance is found out from the distances and is recorded as a first distance; similarly, the position information of the intersection point of the atria on the coronary image to be detected and the distance between each point on the central line of the right coronary artery can be calculated to obtain a plurality of distances, and the shortest distance can be found out from the distances and recorded as the second distance. The formula for calculating the distance may be calculated by using an existing formula for calculating the distance between the point and the line, or may be calculated directly by using an existing formula for calculating the distance between the point and the line to obtain the first distance and the second distance.
S404, analyzing the coronary artery image to be detected according to the first distance, the second distance and a preset distance threshold range to obtain an analysis result.
The preset distance threshold range can be calculated according to the position information of the atrioventricular intersection point marked on the sample coronary image and the central line of the left coronary artery and the right coronary artery marked on the sample coronary image, and the sample coronary image can be marked with the dominant type category. In addition, the preset distance threshold range includes a first distance threshold range, a second distance threshold range, and a third distance threshold range, where the first distance threshold range may be a distance threshold range of an equilibrium type coronary artery calculated according to an equilibrium type sample coronary artery image, the second distance threshold range may be a distance threshold range of a left superior type coronary artery calculated according to a left superior type sample coronary artery image, and the third distance threshold range may be a distance threshold range of a right superior type coronary artery calculated according to a right superior type sample coronary artery image, and correspondingly, the method for calculating the preset distance threshold range may be as follows:
calculating the shortest distance from the atrioventricular intersection point marked on the balanced sample coronary image to the central line of the left and right coronary arteries to obtain the shortest distance corresponding to each balanced sample coronary image, and then calculating the mean value mu of the plurality of shortest distances1And standard deviation σ1And will [ mu ] be1-2σ1，μ1+2σ1]The range is used as a first distance threshold range; similarly, calculate the left dominant modeThe shortest distance from the atrioventricular intersection point marked on the sample coronary image to the central line of the left coronary is obtained to obtain the corresponding shortest distance of each left-side dominant-type sample coronary image, and then the average value mu of the shortest distances is calculated2And standard deviation σ2And will [ mu ] be2-2σ2，μ2+2σ2]The range is used as a second distance threshold range; similarly, the shortest distance from the atrioventricular intersection point marked on the right dominant coronary artery image to the central line of the right coronary artery is calculated to obtain the corresponding shortest distance of each right dominant coronary artery image, and then the mean value mu of the shortest distances is calculated3And standard deviation σ3And will [ mu ] be3-2σ3，μ3+2σ3]The range serves as a third distance threshold range.
Specifically, after obtaining the first distance, the second distance and the preset three distance threshold ranges, optionally, the first distance, the second distance and the first distance threshold range may be matched; if the first distance and the second distance do not exceed the first distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type according to the analysis result; or, matching the first distance with a second distance threshold range; if the first distance does not exceed the second distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a left dominant type; or matching the second distance with a third distance threshold range; and if the second distance does not exceed the third distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is the right dominant type.
That is, here, when comparing the first distance and the second distance with the preset distance threshold range, the first distance and the second distance may be respectively compared with the first distance threshold range [ μ [ mu ] ]1-2σ1，μ1+2σ1]Comparing, and if the first distance and the second distance do not exceed the first distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type; if any one of the first distance and the second distance exceeds the first distance threshold rangeThen the first distance and the second distance threshold range [ mu ] are continued2-2σ2，μ2+2σ2]Comparing, and if the first distance does not exceed the second distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a left dominant type; if the first distance exceeds the second distance threshold range, the second distance and a third distance threshold range [ mu ] are continued3-2σ3，μ3+2σ3]Comparing, and if the second distance does not exceed the third distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a right dominant type; if the second distance exceeds the third distance threshold range, outputting that the dominant type of the coronary artery in the coronary artery image to be detected is uncertain, and requiring further determination by a doctor.
When the first distance, the second distance and the preset distance threshold range are compared, there may be 6 matching orders, the 1 st: matching balanced types according to the method, if the balanced types are not matched, matching a left side dominant type, and if the left side dominant types are not matched, matching a right side dominant type; the 2 nd: firstly, matching balanced types, if the balanced types are not matched, matching a right side dominant type, and if the right side dominant types are not matched, matching a left side dominant type; and (3) type: matching a left dominant type, if the left dominant type is not matched, matching a right dominant type, and if the right dominant type is not matched, matching a balanced type; and 4, the method comprises the following steps: matching a left dominant type, if the left dominant type is not matched, matching an equilibrium type, and if the equilibrium type is not matched, matching a right dominant type; and (5) the following steps: matching the right side dominant type, if the right side dominant type is not matched, matching the balanced type, and if the balanced type is not matched, matching the left side dominant type; the 6 th: and matching the right side dominant type firstly, if the right side dominant type is not matched, matching the left side dominant type, and if the left side dominant type is not matched, matching the balanced type.
Based on the above, if the first distance threshold range may include the left branch first distance threshold range and the right branch first distance threshold range, optionally, the matching is performed according to the first distance, the second distance, and the first distance threshold rangeThe step of obtaining the results of the analysis of the equalization type may comprise: matching the first distance with the threshold range of the first distance of the left branch, and matching the second distance with the threshold range of the first distance of the right branch; and if the first distance does not exceed the first distance threshold range of the left branch and the second distance does not exceed the first distance threshold range of the right branch, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type. That is, when the first distance threshold range is determined, the shortest distance from the atrioventricular intersection marked on the balanced sample coronary image to the central line of the left coronary artery can be calculated to obtain the corresponding left branch shortest distance of each balanced sample coronary image, and then the average value μ of the plurality of left branch shortest distances can be calculated1And standard deviation σ1And the left branch [ mu ]1-2σ1，μ1+2σ1]The range is used as a left-branch first distance threshold range; meanwhile, the shortest distance from the atrioventricular intersection point marked on the balanced sample coronary image to the central line of the right coronary artery can be calculated to obtain the right branch shortest distance corresponding to each balanced sample coronary image, and then the mean value mu of the plurality of right branch shortest distances is calculated1And standard deviation σ1And the right branch [ mu ]1-2σ1，μ1+2σ1]The range is used as a right branch first distance threshold range; after the first distance and the second distance are calculated and obtained for the coronary artery image to be detected, the first distance and the second distance can be matched with the corresponding first distance threshold range of the left branch and the corresponding first distance threshold range of the right branch, and if the first distance and the second distance do not exceed the corresponding distance threshold ranges, the dominant type of the coronary artery is determined to be a balanced type; if any one of the first distance and the second distance exceeds the corresponding distance threshold range, the 6 comparison sequences can be operated to obtain the final analysis result of the dominant coronary artery.
The image analysis method provided in this embodiment may calculate a shortest distance between the position information of the atrioventricular intersection point and the central line of the left coronary artery to obtain a first distance, calculate a shortest distance between the position information of the atrioventricular intersection point and the central line of the right coronary artery to obtain a second distance, and analyze the coronary artery image to be detected according to the first distance, the second distance, and a preset distance threshold range to obtain an analysis result. In this embodiment, since the final dominant type can be obtained by comparing the threshold value with the shortest distance between the atrioventricular intersection point and the central lines of the left and right coronary arteries, the method is simple and direct, and manual analysis is not required, so that manpower can be saved to a certain extent, the analysis time of the dominant type of the coronary artery can be saved, and the analysis speed of the dominant type of the coronary artery can be increased.
The case where the reference information is the atrioventricular groove coordinate system will be described next.
In another embodiment, another image detection method is provided, and this embodiment relates to a specific process of how to detect a coronary artery image to be detected by using a detection model to obtain an atrioventricular groove coordinate system in the coronary artery image. On the basis of the above embodiment, as shown in fig. 5a, the above S202 may include the following steps:
s502, segmenting the coronary artery image to be detected to obtain a segmentation result of the heart chamber; the results of the segmentation of the heart chamber include the left and right atria and the left and right ventricles of the heart.
Wherein, the heart chamber includes temple ditch chamber, is left atrium, left ventricle, right atrium, right ventricle respectively. When the coronary artery image to be detected is segmented, a cavity segmentation model or an image segmentation algorithm and the like can be adopted to segment the cavity in the coronary artery image to be detected; if the chamber segmentation model is used for segmentation, the chamber segmentation model can be trained in advance, and the chamber segmentation model can be obtained by training according to the sample coronary artery image and the gold standard images of the four chambers corresponding to the sample coronary artery image.
Specifically, when the coronary artery image to be detected is segmented, the coronary artery image to be detected can be input into a pre-trained cavity segmentation model to obtain segmented images of four cavities, and the segmented images of the four cavities can be one image or each cavity respectively corresponds to one segmented image; in addition, the four chamber segmentation images may include position information of each chamber.
S504, establishing an atrioventricular groove coordinate system according to the heart chamber segmentation result, and taking the atrioventricular groove coordinate system as the reference information of the coronary artery; the transverse axis direction of the atrioventricular groove coordinate system is the direction of the heart atrioventricular groove boundary line, and the longitudinal axis direction of the atrioventricular groove coordinate system is the direction of the heart ventricular groove boundary line.
The atrioventricular groove boundary refers to the boundary between the atria and the ventricles, and the ventricular groove boundary refers to the boundary between the left ventricle and the right ventricle of the heart septal surface.
Specifically, after obtaining the segmented image of each chamber, the computer device may also obtain position information of each chamber, including position information of boundary points of each chamber, and then may perform fitting processing (for example, least square fitting or the like) on the positions of each point on the atrioventricular groove boundary line to obtain an X axis, which may also be recorded as a horizontal axis, where a positive direction of the horizontal axis may be a direction of the right atrium of the right ventricle; similarly, the position of each point on the ventricular sulcus boundary line may be subjected to fitting processing to obtain a Y-axis, which may also be referred to as a vertical axis, where the positive direction of the vertical axis may be the direction of the left ventricle and the right ventricle, where the X-axis and the Y-axis are perpendicular to each other, where the origin may be the intersection of the X-axis and the Y-axis, or may also be the above-mentioned atrioventricular intersection, and through the obtained origin, the X-axis and the Y-axis may obtain an atrioventricular sulcus coordinate system, and the established atrioventricular sulcus coordinate system may be shown in fig. 5 b.
The image analysis method provided by the embodiment can be used for segmenting the coronary artery image to be detected to obtain the segmentation result of the heart chamber; the segmentation result of the heart chamber comprises left and right atria and left and right ventricles of the heart; and establishing an atrioventricular groove coordinate system according to the segmentation result of the heart chamber, and taking the atrioventricular groove coordinate system as the reference information of the coronary artery. In the embodiment, since the atrioventricular groove coordinate system can be established according to the segmentation result of the heart chamber, a calculation basis can be provided for the subsequent analysis of the dominant coronary artery; in addition, because the coordinate system is fitted by adopting the segmentation result of segmenting the heart chamber, rather than establishing the coordinate system directly according to the segmentation result, the coordinate system fitted by the embodiment is also more accurate, so that the subsequent analysis result of the dominant coronary artery is more accurate.
In another embodiment, another image detection method is provided, and the embodiment relates to a specific process of analyzing the coronary artery image to be detected based on the atrioventricular groove coordinate system and the central lines of the left and right coronary arteries to obtain the analysis result of the dominant type. On the basis of the above embodiment, as shown in fig. 6, the above S206 may include the following steps:
s602, calculating coordinates of each point on the central line of the left coronary artery under the atrioventricular groove coordinate system to obtain coordinates of at least one left fulcrum, and calculating coordinates of each point on the central line of the right coronary artery under the atrioventricular groove coordinate system to obtain coordinates of at least one right fulcrum.
Specifically, after the atrioventricular artery coordinate system is established, the coordinates of each point on the central line of the left coronary artery and the coordinates of each point on the central line of the right coronary artery under the standard of the atrioventricular artery coordinate system can be calculated, wherein the coordinates of each point on the central line of the left coronary artery are recorded as the coordinates of a left fulcrum, and the coordinates of each point on the central line of the right coronary artery are recorded as the coordinates of a right fulcrum. The coordinates here may be one-dimensional coordinates, two-dimensional coordinates, three-dimensional coordinates, and the like, and the two-dimensional coordinates of each point, which are coordinates in the X-axis direction and coordinates in the Y-axis direction, respectively, are mainly used here.
S604, analyzing the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and a preset coordinate threshold range to obtain an analysis result.
In this step, optionally, the following step B may be adopted to analyze the coronary artery image to be detected:
b, respectively matching the coordinates of at least one left fulcrum and the coordinates of at least one right fulcrum with a coordinate threshold range to obtain the number of left supporting target points and the number of right supporting target points, comparing the number of the left supporting target points and the number of the right supporting target points with a preset number threshold, and obtaining an analysis result according to a comparison result; the left supporting target point is a point of which the coordinate in at least one left supporting point does not exceed the range of the coordinate threshold, and the right supporting target point is a point of which the coordinate in at least one right supporting point does not exceed the range of the coordinate threshold.
The preset coordinate threshold range may be an X-axis coordinate threshold range, which may be calculated according to a central line of a posterior descending branch marked on the sample coronary image, and the sample coronary image may also be marked with an advantageous type category. Specifically, first, a left posterior descending branch and a right posterior descending branch may be marked on the equilibrium type sample coronary image in advance, then the coordinates of the X axis of the left posterior descending branch in the atrioventricular sulcus coordinate system on each equilibrium type sample coronary image are obtained through calculation, the maximum coordinate and the minimum coordinate thereof are used as the first coordinate threshold range of the left branch, the coordinates of the X axis of the right posterior descending branch in the atrioventricular sulcus coordinate system on each equilibrium type sample coronary image are obtained through calculation, and the maximum coordinate and the minimum coordinate thereof are used as the first coordinate threshold range of the right branch; secondly, marking a left posterior descending branch on the left superior type sample coronary image, then calculating to obtain the coordinate of the X axis of the left posterior descending branch on each left superior type sample coronary image under the atrioventricular groove coordinate system, and taking the maximum coordinate and the minimum coordinate as a left second coordinate threshold range; thirdly, a right posterior descending branch can be marked on the right superior type sample coronary image, then the coordinates of the X axis of the right posterior descending branch on each right superior type sample coronary image under the atrioventricular groove coordinate system are calculated, and the maximum coordinate and the minimum coordinate of the coordinates are used as the threshold range of the right second coordinate.
Specifically, after obtaining the coordinates of at least one left fulcrum and the coordinates of at least one right fulcrum, the coordinates of the left fulcrum may be compared with the corresponding left-branch coordinate threshold range to obtain the number of left-branch target points within the left-branch coordinate threshold range, and simultaneously, the coordinates of the right fulcrum may also be compared with the corresponding right-branch coordinate threshold range to obtain the number of right-branch target points within the right-branch coordinate threshold range, and then the number of left-branch target points and the number of right-branch target points are compared with the number threshold, and according to the comparison result, the dominant analysis result is obtained.
When the number of the left branch target points and the number of the right branch target points are obtained and compared with the number threshold, the comparison can be carried out in three conditions, namely, the comparison is carried out with the coordinate threshold range and the number threshold of the balanced type, the comparison is carried out with the coordinate threshold range and the number threshold of the left side dominant type, and the comparison is carried out with the coordinate threshold range and the number threshold of the right side dominant type.
During specific comparison, because the balanced condition has two coordinate threshold ranges, correspondingly, there may also be two quantity thresholds, namely a left branch first quantity threshold and a right branch first quantity threshold, and the comparison process may be to match the coordinate of at least one left fulcrum with the left branch first coordinate threshold range to obtain the quantity of left branch first target points, and to match the coordinate of at least one right fulcrum with the right branch first coordinate threshold range to obtain the quantity of right branch first target points; comparing the number of the first target points of the left branch with a first threshold value of the left branch, and comparing the number of the first target points of the right branch with a first threshold value of the right branch; and if the number of the left first target points is greater than the left first number threshold and the number of the right first target points is greater than the right first number threshold, determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type.
In addition, the left branch second coordinate threshold range under the left dominant type condition may correspond to a quantity threshold and is recorded as a left branch second quantity threshold, the specific comparison process may be to match the coordinate of at least one left fulcrum with the left branch second coordinate threshold range to obtain the quantity of the left branch second target point, and compare the quantity of the left branch second target point with the left branch second quantity threshold, and if the quantity of the left branch second target point is greater than the left branch second quantity threshold, it is determined that the dominant type category of the coronary artery in the coronary artery image to be detected is the left dominant type.
In addition, the right branch second coordinate threshold range under the right dominant type condition may correspond to a quantity threshold and is recorded as a right branch second quantity threshold, the specific comparison process may be to match the coordinate of at least one right fulcrum with the right branch second coordinate threshold range to obtain the quantity of right branch second target points, and compare the quantity of right branch second target points with the right branch second quantity threshold, and if the quantity of right branch second target points is greater than the right branch second quantity threshold, it is determined that the dominant type category of the coronary artery in the coronary artery image to be detected is the right dominant type.
The size of the left branch first number threshold, the size of the right branch first number threshold, the size of the left branch second number threshold, and the size of the right branch second number threshold may all be determined according to actual situations, and for example, assuming that the number of the left fulcrums is 1000 and the number of the left branch target points is 800, the left branch first number threshold may be set to 700, 750, and the like.
In addition, the coordinates of the left fulcrum and the coordinates of the right fulcrum may be compared with the corresponding coordinate threshold range, and when the coordinates of the left fulcrum and/or the coordinates of the right fulcrum do not exceed the corresponding coordinate threshold range, the number of points in the coordinate threshold range may be compared with the corresponding number threshold.
For example, assume that the left-branch first coordinate threshold range is denoted as [ CL (μ -2 σ), CL (μ +2 σ) ], the corresponding left-branch first number threshold value is denoted as CL _ min, the right-branch first coordinate threshold range is denoted as [ CR (μ -2 σ), CR (μ +2 σ) ], the corresponding right-branch first number threshold value is denoted as CR _ min, the left-branch second coordinate threshold range is denoted as [ L (μ -2 σ), L (μ +2 σ) ], the corresponding left-branch second number threshold value is denoted as L _ min, the right-branch second coordinate threshold range is denoted as [ R (μ -2 σ), R (μ +2 σ) ], and the corresponding right-branch second number threshold value is denoted as R _ min.
Then when the dominant type is determined for the coronary image to be detected,
if 1) counting the number of points with the absolute value of the coordinate X of the central line point of the left branch of the coronary artery in the value range of [ CL (mu-2 sigma), CL (mu +2 sigma) ] and the number of the points is larger than a counted threshold value CL _ min, 2) counting the number of points with the absolute value of the coordinate X of the central line point of the right branch of the coronary artery in the value range of [ CR (mu-2 sigma), CR (mu +2 sigma) ] and the number of the points is larger than the counted threshold value CR _ min, judging the coronary artery to be in a balanced type;
if the number of the points of which the absolute value of the coordinate X of the central line point of the left branch of the coronary artery is in the value range of [ L (mu-2 sigma), L (mu +2 sigma) ] is obtained through statistics, and the number of the points is greater than a statistical threshold value L _ min, the left dominant type is judged;
and if the number of the points of which the absolute value of the coordinate X of the central line point of the right branch of the coronary artery is in the value range of [ R (mu-2 sigma), R (mu +2 sigma) ] is obtained through statistics, and the number of the points is greater than a statistical threshold value R _ min, determining the right dominant type.
It should be noted that, when the coordinates of the left fulcrum and the coordinates of the right fulcrum are compared with the corresponding coordinate threshold range, and the coordinates of the left fulcrum and/or the coordinates of the right fulcrum do not exceed the corresponding coordinate threshold range, there may be 6 comparison sequences, where the 6 comparison sequences are the same as the 6 matching sequences corresponding to the above atrioventricular intersection, but the matching parameters during matching are different, and this embodiment is not described herein again.
The image analysis method provided in this embodiment may calculate coordinates of each point on the centerline of the left coronary artery in the atrioventricular groove coordinate system to obtain coordinates of at least one left fulcrum, calculate coordinates of each point on the centerline of the right coronary artery in the atrioventricular groove coordinate system to obtain coordinates of at least one right fulcrum, and analyze the image of the coronary artery to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum, and a preset coordinate threshold range to obtain an analysis result. In this embodiment, the number of target points of coordinates of points on the central lines of the left and right coronary arteries within the coordinate threshold range can be counted, and the number of the target points and the threshold are compared to obtain the dominant type category.
In another embodiment, another image detection method is provided, and the embodiment relates to a specific process of how to segment the coronary artery in the coronary artery image to be detected. On the basis of the above embodiment, as shown in fig. 7a, the above S204 may include the following steps:
s702, inputting the coronary artery image to be detected into a preset segmentation model to obtain a probability map of left and right coronary arteries corresponding to the coronary artery image to be detected; the pixel values of the positions on the probability graph of the left and right coronary arteries are the probability that the pixel values of the corresponding positions on the coronary artery image to be detected belong to the left and right coronary arteries, the segmentation model is obtained by training based on a second sample coronary artery image and a gold standard image corresponding to the second sample coronary artery image, and the gold standard image corresponding to the second sample coronary artery image comprises left and right coronary artery marks corresponding to the second sample coronary artery image.
And S704, performing binarization processing on the probability maps of the left and right coronary arteries according to a preset first probability threshold to obtain segmentation images of the left and right coronary arteries corresponding to the probability maps of the left and right coronary arteries.
In this embodiment, when the segmentation model is used to segment the left and right coronary arteries in the coronary artery image to be detected, the segmentation model needs to be trained in advance, and when the segmentation model is trained, the training process may include the following three steps:
1) a training data set is generated. Using annotation software to respectively delineate (or mark) N coronary left branch coronary vessel interested areas (the pixel values in the areas are marked as V1, V1 belongs to natural numbers) and N coronary right branch coronary vessel interested areas (the pixel values in the areas are marked as V2, V2 belongs to natural numbers which are not equal to V1) in the N second sample coronary images, setting the pixel values outside the left and right branch vessel interested areas as 0, generating N delineated images of left and right branch coronary vessels by the processing, and pairing the delineated images with the N second sample coronary images to form a training data set, wherein the number of the delineated images is N; the modality of the second sample coronary image is the same as the modality of the coronary image to be detected, and the number of N may be determined according to actual circumstances, and may be 1000, 2000, or the like, for example. For example, refer to fig. 7b, which is a schematic diagram of left and right coronary vessels marked on a second sample coronary image, and the four diagrams are respectively left and right coronary vessels marked on a sagittal plane, a coronal plane, a horizontal plane, and a three-dimensional solid (it should be noted that these four diagrams are just an example, and do not affect the essence of the embodiment of the present application).
2) And establishing a segmentation model which can also be a convolution neural network model. Constructing a convolutional neural network model, and setting hyper-parameters of the convolutional neural network model, wherein an input channel of the network is 1, the input channel is a second sample coronary image, an output channel is 3, the left branch probability map and the right branch probability map respectively correspond to N left and right coronary vessel segmentation images, and the right branch probability map corresponds to a probability map of a background; in addition, the training data set in 1) can be divided into a training set X1 verification set X2 and a test set X3, wherein the training set, the verification set and the test set are independent of each other, the number of the training set, the verification set and the test set is N1, N2 and N3, the training set, the verification set and the test set are natural numbers, N1+ N2+ N3 is equal to N, and N1 is equal to or more than 1/2N; for example, assuming that the number of second sample coronary images is 1000, it may be n 1-500, n 2-200, and n 3-300.
3) And training a segmentation model. Training the segmentation model established in step 2) using the training data set generated in step 1). The training set X1 is used for training the segmentation model, the verification set X2 is used for evaluating the current performance of the model, and the test set X3 is used for checking the generalization performance of the model; in the training process, the training set is divided into a plurality of batches (for example, 100 rounds of training in 100 batches) and repeatedly input into the segmentation model to train the plurality of rounds, meanwhile, the difference between the output image and the gold standard image is calculated by using the cost function and is fed back to the segmentation model as a training error (here, the difference of pixel values between the output left and right branch images and the corresponding left and right branch gold standard images can be respectively calculated, and the sum or mean of the two differences is fed back to the segmentation model), and the model parameters are updated by a learning algorithm; and after the training of each batch is finished, performing performance test on the segmentation model by using the verification set, considering that the training of the segmentation model is finished when the performance test indexes tend to be stable, and storing the trained network model. In addition, the network structure of the segmentation model may be the same as the network structure of the detection model, and is not described herein again.
In addition, the above-mentioned segmentation model may be a deep convolutional neural network CNN, a generative countermeasure network GAN, convolutional neural networks U-Net and V-Net, or a recurrent neural network RNN, etc.; the hyper-parameters may include the number of network layers, convolution kernels, learning rate, parameter initialization, number of training rounds, and batch size. The cost function may be a set similarity metric function (Dice loss) or a focus loss function (Focal loss), and then one of a Stochastic Gradient Descent (SGD), an Adaptive Moment Estimation optimization algorithm (Adam), and a Momentum algorithm (Momentum) may also be used to minimize a training error to train the segmentation model.
After the segmentation model is trained, inputting the coronary image to be detected into the trained segmentation model to obtain a probability map of a left coronary blood vessel and a probability map of a right coronary blood vessel corresponding to the coronary image to be detected, wherein each probability value on the probability map of the left coronary blood vessel can represent the probability that the corresponding position belongs to the left coronary artery, and each probability value on the probability map of the right coronary blood vessel can represent the probability that the corresponding position belongs to the right coronary artery; then, comparing each probability value on the probability map of the left coronary artery blood vessel with the first probability threshold, setting the probability value smaller than the first probability threshold on the probability map as 0, and setting the probability value larger than or equal to the first probability threshold on the probability map as 1 to obtain a binary image which is marked as a segmentation image of the left coronary artery; similarly, the probability values on the probability map of the right coronary artery blood vessel may be compared with the first probability threshold, the probability value smaller than the first probability threshold on the probability map may be set to 0, and the probability value greater than or equal to the first probability threshold on the probability map may be set to 1, so as to obtain a binarized image, which is referred to as a right coronary artery segmentation image.
The image analysis method provided in this embodiment may input the coronary artery image to be detected to a preset segmentation model, obtain probability maps of the left and right coronary arteries corresponding to the coronary artery image to be detected, and perform binarization processing on the probability maps of the left and right coronary arteries according to a preset first probability threshold, so as to obtain segmentation images of the left and right coronary arteries corresponding to the probability maps of the left and right coronary arteries. In this embodiment, since the segmentation model is obtained by training based on the sample coronary artery image and the gold standard images labeled on the left and right coronary arteries, the trained segmentation model is relatively accurate, so that when the trained segmentation model is used to segment the left and right coronary arteries on the coronary artery image to be detected, the obtained segmentation images of the left and right coronary arteries are relatively accurate, and the subsequent coronary artery central lines extracted by using the segmentation images and the obtained dominant type analysis result are more accurate.
It should be understood that although the individual steps in the flowcharts of fig. 2, 3b, 4, 5a, 6, 7a are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2, 3b, 4, 5a, 6, and 7a may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of performing the steps or stages is not necessarily sequential, but may be performed alternately or alternatingly with other steps or at least some of the other steps.
In one embodiment, as shown in fig. 8, there is provided an image analysis apparatus including: detection module 10, segmentation module 11 and analysis module 12, wherein:
the detection module 10 is configured to perform detection processing on a coronary artery image to be detected to obtain reference information of a coronary artery in the coronary artery image to be detected; the reference information of the coronary artery is related to the position information of an atrium and a ventricle surrounded by the coronary artery in the coronary artery image to be detected;
the segmentation module 11 is configured to perform segmentation processing on the coronary artery image to be detected to obtain segmentation images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extract central lines of the left and right coronary arteries of the segmentation images of the left and right coronary arteries;
the analysis module 12 is configured to analyze the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
For specific limitations of the image analysis apparatus, reference may be made to the above limitations of the image analysis method, which are not described herein again.
In another embodiment, another image analysis apparatus is provided, based on the above embodiment, the detection module 10 is further configured to perform detection processing on the atrioventricular intersection point in the coronary artery image to be detected by using a preset detection model, to obtain position information of the atrioventricular intersection point in the coronary artery image to be detected, and to use the position information of the atrioventricular intersection point as reference information of the coronary artery; the detection model is obtained by training based on a first sample coronary image and a gold standard image corresponding to the first sample coronary image, the gold standard image corresponding to the first sample coronary image comprises a position mark of an atrioventricular intersection point corresponding to the first sample coronary image, and the atrioventricular intersection point is a point at a junction of a central vein between ventricles and a coronary sinus of the heart room.
Optionally, the detection module 10 may include a first detection unit, a first processing unit, an obtaining unit, and a determining unit, where:
the first detection unit is used for inputting the coronary artery image to be detected into the detection model to obtain a probability map of the atrioventricular intersection point corresponding to the coronary artery image to be detected; the pixel value of each position on the probability graph of the atrioventricular intersection point is the probability that the pixel value of the corresponding position on the coronary image to be detected belongs to the atrioventricular intersection point;
the first processing unit is used for carrying out binarization processing on the probability map of the chamber intersection point according to a preset second probability threshold value to obtain a binarization mask image corresponding to the probability map of the chamber intersection point;
the acquiring unit is used for marking the connected domains in the binary mask image and determining the maximum connected domain according to the marked connected domains;
and the determining unit is used for acquiring a weighted central point of the probability value corresponding to the maximum connected domain and determining the position information of the weighted central point as the position information of the chamber intersection point.
In another embodiment, another image analysis apparatus is provided, and on the basis of the above embodiment, the above analysis module 12 may include a calculation unit and an analysis unit, wherein:
the calculating unit is used for calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the left coronary artery to obtain a first distance; calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the right coronary artery to obtain a second distance;
and the analysis unit is used for analyzing the coronary artery image to be detected according to the first distance, the second distance and a preset distance threshold range to obtain an analysis result.
Optionally, the preset distance threshold range includes a first distance threshold range, a second distance threshold range, and a third distance threshold range, and the analysis unit is further configured to match the first distance, the second distance, and the first distance threshold range; if the first distance and the second distance do not exceed the first distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type according to the analysis result; or, matching the first distance with a second distance threshold range; if the first distance does not exceed the second distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a left dominant type; or matching the second distance with a third distance threshold range; and if the second distance does not exceed the third distance threshold range, determining that the dominant type of the coronary artery in the coronary artery image to be detected is the right dominant type.
Optionally, the first distance threshold range includes a left branch first distance threshold range and a right branch first distance threshold range, and the analysis unit is further configured to match the first distance with the left branch first distance threshold range, and match the second distance with the right branch first distance threshold range; and if the first distance does not exceed the first distance threshold range of the left branch and the second distance does not exceed the first distance threshold range of the right branch, determining that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type.
In another embodiment, another image analysis apparatus is provided, and on the basis of the above embodiment, the detection module 10 may include a second detection unit and a creation unit, wherein:
the second detection unit is used for carrying out segmentation processing on the coronary artery image to be detected to obtain a segmentation result of the heart chamber; the segmentation result of the heart chamber comprises left and right atria and left and right ventricles of the heart;
the establishing unit is used for establishing an atrioventricular groove coordinate system according to the segmentation result of the heart chamber and taking the atrioventricular groove coordinate system as the reference information of the coronary artery; the transverse axis direction of the atrioventricular groove coordinate system is the direction of the heart atrioventricular groove boundary line, and the longitudinal axis direction of the atrioventricular groove coordinate system is the direction of the heart ventricular groove boundary line.
In another embodiment, on the basis of the above embodiment, the calculating unit is further configured to calculate coordinates of each point on the centerline of the left coronary artery in the atrioventricular groove coordinate system to obtain coordinates of at least one left fulcrum, and calculate coordinates of each point on the centerline of the right coronary artery in the atrioventricular groove coordinate system to obtain coordinates of at least one right fulcrum;
the analysis unit is further configured to analyze the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and a preset coordinate threshold range, so as to obtain an analysis result.
Optionally, the analyzing unit is further configured to match the coordinate of the at least one left fulcrum and the coordinate of the at least one right fulcrum with a coordinate threshold range, to obtain the number of left branch target points and the number of right branch target points, compare the number of left branch target points and the number of right branch target points with a preset number threshold, and obtain an analysis result according to a comparison result; the left supporting target point is a point of which the coordinate in at least one left supporting point does not exceed the range of the coordinate threshold, and the right supporting target point is a point of which the coordinate in at least one right supporting point does not exceed the range of the coordinate threshold.
Optionally, if the coordinate threshold range includes a left first coordinate threshold range and a right first coordinate threshold range, the number threshold includes a left first number threshold and a right first number threshold; the analysis unit is further configured to match the coordinate of the at least one left pivot with the threshold range of the left first coordinate to obtain the number of left first target points, and match the coordinate of the at least one right pivot with the threshold range of the right first coordinate to obtain the number of right first target points; comparing the number of the first target points of the left branch with a first threshold value of the left branch, and comparing the number of the first target points of the right branch with a first threshold value of the right branch; and if the number of the left first target points is greater than the left first number threshold and the number of the right first target points is greater than the right first number threshold, determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type.
Optionally, if the coordinate threshold range includes a left second coordinate threshold range and a right second coordinate threshold range, the number threshold includes a left second number threshold and a right second number threshold; the analysis unit is further configured to match the coordinate of the at least one left fulcrum with a left branch second coordinate threshold range to obtain the number of left branch second target points, compare the number of the left branch second target points with the left branch second number threshold, and determine that the analysis result is that the dominant type category of the coronary artery in the coronary artery image to be detected is a left dominant type if the number of the left branch second target points is greater than the left branch second number threshold; or matching the coordinate of at least one right pivot with the threshold range of the right second coordinate to obtain the number of right second target points, comparing the number of the right second target points with the threshold of the right second number, and determining that the dominant type of the coronary artery in the coronary artery image to be detected is the right dominant type if the number of the right second target points is greater than the threshold of the right second number.
In another embodiment, another image analysis apparatus is provided, and on the basis of the above embodiment, the powder modules 11 may include a segmentation unit and a second processing unit, wherein:
the segmentation unit is used for inputting the coronary artery image to be detected into a preset segmentation model to obtain a probability map of left and right coronary arteries corresponding to the coronary artery image to be detected; the pixel values of the positions on the probability graph of the left and right coronary arteries are the probability that the pixel values of the corresponding positions on the coronary artery image to be detected belong to the left and right coronary arteries, the segmentation model is obtained by training based on a second sample coronary artery image and a gold standard image corresponding to the second sample coronary artery image, and the gold standard image corresponding to the second sample coronary artery image comprises left and right coronary artery marks corresponding to the second sample coronary artery image;
and the second processing unit is used for carrying out binarization processing on the probability maps of the left and right coronary arteries according to a preset first probability threshold value to obtain segmentation images of the left and right coronary arteries corresponding to the probability maps of the left and right coronary arteries.
For specific limitations of the image analysis apparatus, reference may be made to the above limitations of the image analysis method, which are not described herein again. The modules in the image analysis device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.
In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (14)

1. A method of image analysis, the method comprising:
detecting a coronary image to be detected to obtain reference information of a coronary in the coronary image to be detected; the reference information is related to position information of an atrium and a ventricle surrounded by a coronary artery in the coronary image to be detected;
performing segmentation processing on the coronary artery image to be detected to obtain segmentation images of left and right coronary arteries corresponding to the coronary artery image to be detected, and extracting central lines of the left and right coronary arteries of the segmentation images of the left and right coronary arteries;
analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result; and the analysis result is used for representing the dominant type category of the coronary artery in the coronary artery image to be detected.
2. The method according to claim 1, wherein the detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected comprises:
detecting atrioventricular intersection points in a coronary image to be detected by adopting a preset detection model to obtain position information of the atrioventricular intersection points in the coronary image to be detected, and taking the position information of the atrioventricular intersection points as reference information of the coronary;
the detection model is obtained by training based on a first sample coronary image and a gold standard image corresponding to the first sample coronary image, the gold standard image corresponding to the first sample coronary image comprises a position mark of an atrioventricular intersection point corresponding to the first sample coronary image, and the atrioventricular intersection point is a point at a junction of a central vein between ventricles of a septal surface of a heart and a coronary sinus of a cardiac room.
3. The method according to claim 2, wherein the central lines of the left and right coronary arteries include a central line of the left coronary artery and a central line of the right coronary artery, and the analyzing the image of the coronary artery to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result comprises:
calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the left coronary artery to obtain a first distance; calculating the shortest distance between the position information of the atrioventricular intersection point and the central line of the right coronary artery to obtain a second distance;
and analyzing the coronary artery image to be detected according to the first distance, the second distance and a preset distance threshold range to obtain the analysis result.
4. The method according to claim 3, wherein the preset distance threshold range includes a first distance threshold range, a second distance threshold range and a third distance threshold range, and the analyzing the coronary artery image to be detected according to the first distance, the second distance and the preset distance threshold range to obtain the analysis result includes:
matching the first distance and the second distance to the first distance threshold range; if the first distance and the second distance do not exceed the first distance threshold range, determining that the analysis result is that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type;
or, matching the first distance with the second distance threshold range; if the first distance does not exceed the second distance threshold range, determining that the analysis result is that the dominant type category of the coronary artery in the coronary artery image to be detected is a left dominant type;
or, matching the second distance with the third distance threshold range; and if the second distance does not exceed the third distance threshold range, determining that the analysis result is that the dominant type of the coronary artery in the coronary artery image to be detected is a right dominant type.
5. The method of claim 4, wherein the first distance threshold range comprises a left branch first distance threshold range and a right branch first distance threshold range, and wherein the matching the first distance and the second distance to the first distance threshold range; if neither the first distance nor the second distance exceeds the first distance threshold range, determining that the analysis result is that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type, including:
matching the first distance with the threshold range of the left first distance, and matching the second distance with the threshold range of the right first distance;
and if the first distance does not exceed the first distance threshold range of the left branch and the second distance does not exceed the first distance threshold range of the right branch, determining that the analysis result is that the dominant type of the coronary artery in the coronary artery image to be detected is a balanced type.
6. The method according to claim 1, wherein the detecting the coronary artery image to be detected to obtain the reference information of the coronary artery in the coronary artery image to be detected comprises:
carrying out segmentation processing on the coronary artery image to be detected to obtain a segmentation result of the heart chamber; the segmentation result of the heart chamber comprises left and right atria and left and right ventricles of the heart;
establishing an atrioventricular groove coordinate system according to the segmentation result of the heart chamber, and taking the atrioventricular groove coordinate system as reference information of the coronary artery; the transverse axis direction of the atrioventricular groove coordinate system is the direction of the heart atrioventricular groove boundary line, and the longitudinal axis direction of the atrioventricular groove coordinate system is the direction of the heart ventricular groove boundary line.
7. The method according to claim 6, wherein the central lines of the left and right coronary arteries include a central line of the left coronary artery and a central line of the right coronary artery, and the analyzing the coronary artery image to be detected according to the reference information of the coronary artery and the central lines of the left and right coronary arteries to obtain an analysis result comprises:
calculating the coordinates of each point on the midline of the left coronary artery under the atrioventricular sulcus coordinate system to obtain the coordinates of at least one left fulcrum, and calculating the coordinates of each point on the midline of the right coronary artery under the atrioventricular sulcus coordinate system to obtain the coordinates of at least one right fulcrum;
and analyzing the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and a preset coordinate threshold range to obtain an analysis result.
8. The method according to claim 7, wherein the analyzing the coronary artery image to be detected according to the coordinates of the at least one left fulcrum, the coordinates of the at least one right fulcrum and a preset threshold range of coordinates to obtain an analysis result comprises:
matching the coordinates of the at least one left fulcrum and the coordinates of the at least one right fulcrum with the coordinate threshold range respectively to obtain the number of left branch target points and the number of right branch target points, comparing the number of the left branch target points and the number of the right branch target points with a preset number threshold, and obtaining the analysis result according to the comparison result;
the left supporting target point is a point of the at least one left supporting point, the coordinate of which does not exceed the coordinate threshold range, and the right supporting target point is a point of the at least one right supporting point, the coordinate of which does not exceed the coordinate threshold range.
9. The method of claim 8, wherein if the threshold range of coordinates comprises a first threshold range of left coordinates and a first threshold range of right coordinates, the number threshold comprises a first number threshold of left coordinates and a first number threshold of right coordinates,
the matching the coordinates of the at least one left fulcrum and the coordinates of the at least one right fulcrum with the coordinate threshold range to obtain the number of left branch target points and the number of right branch target points, comparing the number of left branch target points and the number of right branch target points with a preset number threshold, and obtaining the analysis result according to the comparison result includes:
matching the coordinates of the at least one left fulcrum with the threshold range of the first coordinates of the left fulcrum to obtain the number of first target points of the left fulcrum, and matching the coordinates of the at least one right fulcrum with the threshold range of the first coordinates of the right fulcrum to obtain the number of first target points of the right fulcrum;
comparing the number of the left first target points with the left first number threshold, and comparing the number of the right first target points with the right first number threshold;
and if the number of the left branch first target points is greater than the left branch first number threshold and the number of the right branch first target points is greater than the right branch first number threshold, determining that the analysis result is that the dominant type category of the coronary artery in the coronary artery image to be detected is a balanced type.
10. The method of claim 8, wherein if the range of coordinate thresholds includes a left branch second range of coordinate thresholds and a right branch second range of coordinate thresholds, the number threshold includes a left branch second number threshold and a right branch second number threshold,
the matching the coordinates of the at least one left fulcrum and the coordinates of the at least one right fulcrum with the coordinate threshold range to obtain the number of left branch target points and the number of right branch target points, comparing the number of left branch target points and the number of right branch target points with a preset number threshold, and obtaining the analysis result according to the comparison result includes:
matching the coordinates of the at least one left fulcrum with the threshold range of the second left fulcrum to obtain the number of second left fulcrum target points, comparing the number of the second left fulcrum target points with the threshold of the second left fulcrum target points, and if the number of the second left fulcrum target points is greater than the threshold of the second left fulcrum target points, determining that the analysis result is that the dominant type category of the coronary artery in the coronary artery image to be detected is a left dominant type; alternatively, the first and second electrodes may be,
and matching the coordinate of the at least one right pivot with the threshold range of the second right pivot to obtain the number of second right target points, comparing the number of the second right target points with the threshold of the second right number, and determining that the dominant type category of the coronary artery in the coronary artery image to be detected is a right dominant type if the number of the second right target points is greater than the threshold of the second right number.
11. The method according to any one of claims 1 to 10, wherein the segmenting the coronary artery image to be detected to obtain segmented images of left and right coronary arteries corresponding to the coronary artery image to be detected comprises:
inputting the coronary artery image to be detected into a preset segmentation model to obtain a probability map of left and right coronary arteries corresponding to the coronary artery image to be detected; the pixel values of the positions on the probability maps of the left and right coronary arteries are the probability that the pixel values of the corresponding positions on the coronary artery image to be detected belong to the left and right coronary arteries, the segmentation model is obtained by training based on a second sample coronary artery image and a gold standard image corresponding to the second sample coronary artery image, and the gold standard image corresponding to the second sample coronary artery image comprises left and right coronary artery marks corresponding to the second sample coronary artery image;
and carrying out binarization processing on the probability maps of the left and right coronary arteries according to a preset first probability threshold to obtain segmentation images of the left and right coronary arteries corresponding to the probability maps of the left and right coronary arteries.
12. The method according to any one of claims 2 to 5, wherein the detecting the atrioventricular intersection point in the coronary image to be detected by using the preset detection model to obtain the position information of the atrioventricular intersection point in the coronary image to be detected comprises:
inputting the coronary artery image to be detected into the detection model to obtain a probability map of the atrioventricular intersection point corresponding to the coronary artery image to be detected; the pixel value of each position on the probability graph of the atrioventricular intersection point is the probability that the pixel value of the corresponding position on the coronary image to be detected belongs to the atrioventricular intersection point;
carrying out binarization processing on the probability map of the chamber intersection point according to a preset second probability threshold value to obtain a binarization mask image corresponding to the probability map of the chamber intersection point;
marking connected domains in the binary mask image, and determining the maximum connected domain according to the marked connected domains;
and acquiring a weighted central point of the probability value corresponding to the maximum connected domain, and determining the position information of the weighted central point as the position information of the chamber intersection point.
13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method according to any of claims 1 to 12.
14. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 12.
CN202010137683.0A 2020-03-02 2020-03-02 Image analysis method, apparatus and storage medium Active CN111369525B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010137683.0A CN111369525B (en) 2020-03-02 2020-03-02 Image analysis method, apparatus and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010137683.0A CN111369525B (en) 2020-03-02 2020-03-02 Image analysis method, apparatus and storage medium

Publications (2)

Publication Number Publication Date
CN111369525A true CN111369525A (en) 2020-07-03
CN111369525B CN111369525B (en) 2023-05-30

Family

ID=71210384

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010137683.0A Active CN111369525B (en) 2020-03-02 2020-03-02 Image analysis method, apparatus and storage medium

Country Status (1)

Country Link
CN (1) CN111369525B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112164027A (en) * 2020-09-01 2021-01-01 联影智能医疗科技(北京)有限公司 Spinal dural sac compression detection method, device and storage medium
CN112336361A (en) * 2020-09-29 2021-02-09 查云飞 Coronary artery detection method, device and storage medium
CN112801999A (en) * 2021-02-05 2021-05-14 深圳睿心智能医疗科技有限公司 Method and device for determining heart coronary artery dominant type
CN112907563A (en) * 2021-03-17 2021-06-04 复旦大学附属华山医院 CTA full-automatic collateral circulation scoring method and system based on deep learning
CN113139961A (en) * 2021-05-28 2021-07-20 沈阳先进医疗设备技术孵化中心有限公司 Method and device for determining dominant coronary artery type
CN113409323A (en) * 2021-06-18 2021-09-17 推想医疗科技股份有限公司 Coronary artery segmentation method and device, computer-readable storage medium and electronic equipment
CN113450322A (en) * 2021-06-22 2021-09-28 沈阳先进医疗设备技术孵化中心有限公司 Method and device for judging origin abnormality of coronary artery
CN114078139A (en) * 2021-11-25 2022-02-22 四川长虹电器股份有限公司 Image post-processing method based on portrait segmentation model generation result
CN114418947A (en) * 2021-12-16 2022-04-29 深圳市阅影科技有限公司 Naming method and device for heart coronary artery blood vessel
CN116030326A (en) * 2023-01-18 2023-04-28 山东省人工智能研究院 Multi-view fused coronary plaque identification method
CN113450322B (en) * 2021-06-22 2024-05-24 东软医疗系统股份有限公司 Method and device for judging provenance abnormality of coronary artery

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120184840A1 (en) * 2009-04-07 2012-07-19 Kayvan Najarian Automated Measurement of Brain Injury Indices Using Brain CT Images, Injury Data, and Machine Learning
DE102012215035A1 (en) * 2012-08-23 2014-02-27 Siemens Aktiengesellschaft Method for simulation of virtual or actual stress of coronary stent for open-holding of e.g. trachea of human stent-angioplasty patient after dilation during cancer treatment, involves representing segment superimposed with visualization
CN104978725A (en) * 2014-04-03 2015-10-14 上海联影医疗科技有限公司 Method and device for dividing coronary artery
CN105741310A (en) * 2016-03-21 2016-07-06 东北大学 Heart's left ventricle image segmentation system and method
US20170004618A1 (en) * 2014-03-19 2017-01-05 Koninklijke Philips N.V. Method of scan geometry planning for determining wall thickness of an anatomic detail using magnetic resonance imaging
CN110443808A (en) * 2019-07-04 2019-11-12 杭州深睿博联科技有限公司 Medical image processing method and device, equipment, storage medium for the detection of brain middle line

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120184840A1 (en) * 2009-04-07 2012-07-19 Kayvan Najarian Automated Measurement of Brain Injury Indices Using Brain CT Images, Injury Data, and Machine Learning
DE102012215035A1 (en) * 2012-08-23 2014-02-27 Siemens Aktiengesellschaft Method for simulation of virtual or actual stress of coronary stent for open-holding of e.g. trachea of human stent-angioplasty patient after dilation during cancer treatment, involves representing segment superimposed with visualization
US20170004618A1 (en) * 2014-03-19 2017-01-05 Koninklijke Philips N.V. Method of scan geometry planning for determining wall thickness of an anatomic detail using magnetic resonance imaging
CN104978725A (en) * 2014-04-03 2015-10-14 上海联影医疗科技有限公司 Method and device for dividing coronary artery
CN105741310A (en) * 2016-03-21 2016-07-06 东北大学 Heart's left ventricle image segmentation system and method
CN110443808A (en) * 2019-07-04 2019-11-12 杭州深睿博联科技有限公司 Medical image processing method and device, equipment, storage medium for the detection of brain middle line

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
ERIC D.MORRIS等: "Cardiac substructure segmentation with deep learning for improved cardiac sparing", PMC *
JELMER M.WOLTERINK等: "Graph Convolutional Networks for Coronary Artery Segmentation in Cardiac CT Angiography", ELECTRICAL ENGINEERING AND SYSTEMS SCIENCE *
YO-CHUAN CHEN等: "Coronary Artery Segmentation in Cardiac CT Angiography Using 3D Multi-Channel U-net", IMAGE AND VIDEO PROCESSING *
徐礼胜等: "心脏核磁共振图像左心室底层组织分割方法", 东北大学学报(自然科学版) *
郭笑妍等: "形状约束下活动轮廓模型冠脉血管图像多尺度分割", 中国图象图形学报 *
钱家新;: "64排螺旋CT对冠状动脉解剖形态显示的影像评价", 现代医用影像学 *

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112164027B (en) * 2020-09-01 2023-10-27 联影智能医疗科技(北京)有限公司 Spinal dural sac compression detection method, equipment and storage medium
CN112164027A (en) * 2020-09-01 2021-01-01 联影智能医疗科技(北京)有限公司 Spinal dural sac compression detection method, device and storage medium
CN112336361A (en) * 2020-09-29 2021-02-09 查云飞 Coronary artery detection method, device and storage medium
CN112336361B (en) * 2020-09-29 2023-10-31 查云飞 Coronary artery detection method, device and storage medium
CN112801999A (en) * 2021-02-05 2021-05-14 深圳睿心智能医疗科技有限公司 Method and device for determining heart coronary artery dominant type
CN112907563A (en) * 2021-03-17 2021-06-04 复旦大学附属华山医院 CTA full-automatic collateral circulation scoring method and system based on deep learning
CN112907563B (en) * 2021-03-17 2023-12-01 复旦大学附属华山医院 CTA full-automatic collateral circulation scoring method and system based on deep learning
CN113139961A (en) * 2021-05-28 2021-07-20 沈阳先进医疗设备技术孵化中心有限公司 Method and device for determining dominant coronary artery type
CN113139961B (en) * 2021-05-28 2024-04-23 东软医疗系统股份有限公司 Method and device for determining dominant type of coronary artery
CN113409323B (en) * 2021-06-18 2022-03-08 推想医疗科技股份有限公司 Coronary artery segmentation method and device, computer-readable storage medium and electronic equipment
CN113409323A (en) * 2021-06-18 2021-09-17 推想医疗科技股份有限公司 Coronary artery segmentation method and device, computer-readable storage medium and electronic equipment
CN113450322A (en) * 2021-06-22 2021-09-28 沈阳先进医疗设备技术孵化中心有限公司 Method and device for judging origin abnormality of coronary artery
CN113450322B (en) * 2021-06-22 2024-05-24 东软医疗系统股份有限公司 Method and device for judging provenance abnormality of coronary artery
CN114078139A (en) * 2021-11-25 2022-02-22 四川长虹电器股份有限公司 Image post-processing method based on portrait segmentation model generation result
CN114078139B (en) * 2021-11-25 2024-04-16 四川长虹电器股份有限公司 Image post-processing method based on human image segmentation model generation result
CN114418947A (en) * 2021-12-16 2022-04-29 深圳市阅影科技有限公司 Naming method and device for heart coronary artery blood vessel
CN114418947B (en) * 2021-12-16 2023-11-03 深圳市阅影科技有限公司 Naming method and device for heart coronary blood vessel
CN116030326A (en) * 2023-01-18 2023-04-28 山东省人工智能研究院 Multi-view fused coronary plaque identification method
CN116030326B (en) * 2023-01-18 2023-09-08 山东省人工智能研究院 Multi-view fused coronary plaque identification method

Also Published As

Publication number Publication date
CN111369525B (en) 2023-05-30

Similar Documents

Publication Publication Date Title
CN111369525B (en) Image analysis method, apparatus and storage medium
CN110599451B (en) Medical image focus detection and positioning method, device, equipment and storage medium
CN110782474B (en) Deep learning-based method for predicting morphological change of liver tumor after ablation
CN109934812B (en) Image processing method, image processing apparatus, server, and storage medium
CN111383259B (en) Image analysis method, computer device, and storage medium
CN111709485B (en) Medical image processing method, device and computer equipment
JP2017532092A (en) System and method for segmenting medical images based on anatomical landmark-based features
KR102458324B1 (en) Data processing method using a learning model
CN107909622A (en) Model generating method, the scanning planing method of medical imaging and medical image system
CN111932552B (en) Aorta modeling method and device
CN111583199B (en) Sample image labeling method, device, computer equipment and storage medium
CN113538471B (en) Plaque segmentation method, plaque segmentation device, computer equipment and storage medium
Bisla et al. Skin lesion segmentation and classification with deep learning system
CN108597589B (en) Model generation method, target detection method and medical imaging system
CN111568451A (en) Exposure dose adjusting method and system
CN116091466A (en) Image analysis method, computer device, and storage medium
KR102545906B1 (en) Data processing method using a learning model
US11830622B2 (en) Processing multimodal images of tissue for medical evaluation
CN108205805B (en) Automatic establishment method for dense correspondence of cone beam CT image intermediate pixels
CN115880358A (en) Construction method of positioning model, positioning method of image mark points and electronic equipment
CN114693671A (en) Lung nodule semi-automatic segmentation method, device, equipment and medium based on deep learning
CN111815569B (en) Image segmentation method, device, equipment and storage medium based on deep learning
Rahman et al. Deep Learning-Based Left Ventricular Ejection Fraction Estimation from Echocardiographic Videos
CN115578400A (en) Image processing method, and training method and device of image segmentation network
CN113222872A (en) Image processing method, image processing apparatus, electronic device, and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant