CN115049642A - Carotid artery blood vessel intima-media measurement and plaque detection method - Google Patents
Carotid artery blood vessel intima-media measurement and plaque detection method Download PDFInfo
- Publication number
- CN115049642A CN115049642A CN202210960871.2A CN202210960871A CN115049642A CN 115049642 A CN115049642 A CN 115049642A CN 202210960871 A CN202210960871 A CN 202210960871A CN 115049642 A CN115049642 A CN 115049642A
- Authority
- CN
- China
- Prior art keywords
- intima
- media
- blood vessel
- plaque detection
- original
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 74
- 210000001715 carotid artery Anatomy 0.000 title claims abstract description 53
- 238000005259 measurement Methods 0.000 title claims abstract description 40
- 230000011218 segmentation Effects 0.000 claims abstract description 100
- 210000004204 blood vessel Anatomy 0.000 claims abstract description 84
- 239000012528 membrane Substances 0.000 claims abstract description 45
- 238000012805 post-processing Methods 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 14
- 238000002604 ultrasonography Methods 0.000 claims description 19
- 238000000034 method Methods 0.000 claims description 18
- 210000004231 tunica media Anatomy 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 4
- 201000001320 Atherosclerosis Diseases 0.000 description 3
- 230000003902 lesion Effects 0.000 description 3
- 208000010125 myocardial infarction Diseases 0.000 description 3
- 208000024172 Cardiovascular disease Diseases 0.000 description 2
- 206010067116 Carotid arteriosclerosis Diseases 0.000 description 2
- 208000032382 Ischaemic stroke Diseases 0.000 description 2
- 230000003143 atherosclerotic effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 208000037260 Atherosclerotic Plaque Diseases 0.000 description 1
- 208000014882 Carotid artery disease Diseases 0.000 description 1
- 102000008186 Collagen Human genes 0.000 description 1
- 108010035532 Collagen Proteins 0.000 description 1
- 208000032843 Hemorrhage Diseases 0.000 description 1
- 241000282414 Homo sapiens Species 0.000 description 1
- 208000031481 Pathologic Constriction Diseases 0.000 description 1
- 208000006011 Stroke Diseases 0.000 description 1
- 208000007536 Thrombosis Diseases 0.000 description 1
- 208000025865 Ulcer Diseases 0.000 description 1
- 206010072810 Vascular wall hypertrophy Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 208000037876 carotid Atherosclerosis Diseases 0.000 description 1
- 208000006170 carotid stenosis Diseases 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 230000007213 cerebrovascular event Effects 0.000 description 1
- 229920001436 collagen Polymers 0.000 description 1
- 210000002808 connective tissue Anatomy 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000005314 correlation function Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002526 effect on cardiovascular system Effects 0.000 description 1
- 230000000004 hemodynamic effect Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000000302 ischemic effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000007505 plaque formation Effects 0.000 description 1
- 230000036262 stenosis Effects 0.000 description 1
- 208000037804 stenosis Diseases 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- 230000008719 thickening Effects 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/08—Clinical applications
- A61B8/0833—Clinical applications involving detecting or locating foreign bodies or organic structures
- A61B8/085—Clinical applications involving detecting or locating foreign bodies or organic structures for locating body or organic structures, e.g. tumours, calculi, blood vessels, nodules
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/08—Clinical applications
- A61B8/0891—Clinical applications for diagnosis of blood vessels
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B8/52—Devices using data or image processing specially adapted for diagnosis using ultrasonic, sonic or infrasonic waves
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T7/13—Edge detection
-
- 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
-
- 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
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10132—Ultrasound image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Biophysics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Radiology & Medical Imaging (AREA)
- Artificial Intelligence (AREA)
- Animal Behavior & Ethology (AREA)
- Computing Systems (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Software Systems (AREA)
- Heart & Thoracic Surgery (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Vascular Medicine (AREA)
- Geometry (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The invention discloses a carotid artery intima-media measurement and plaque detection method, which relates to the technical field of image processing, and aims at a carotid artery ultrasonic scanning image to respectively obtain an original blood vessel segmentation image, an original intima-media segmentation image and an original plaque detection frame in the carotid artery ultrasonic scanning image; performing blood vessel segmentation post-processing on the original blood vessel segmentation map to obtain an optimized blood vessel segmentation map; performing plaque detection frame post-processing on the original plaque detection frame based on the optimized blood vessel segmentation map to obtain an optimized plaque detection frame; performing inner and middle membrane segmentation post-processing on the original inner and middle membrane segmentation map based on the optimized blood vessel segmentation map to obtain an optimized inner and middle membrane segmentation map; and carrying out intima-media thickness measurement on the intima-media area in the optimized intima-media segmentation chart. Based on the characteristic that intima-media and plaque exist in blood vessels, the plaque detection frame and the intima-media segmentation region which are overlapped with the blood vessels too little or not are filtered, false positive and false alarm of the model are reduced, and the measurement efficiency and the measurement accuracy are improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a carotid artery blood vessel intima-media measurement and plaque detection method.
Background
Cardiovascular and cerebrovascular diseases occurring on the basis of atherosclerotic lesions are one of the major killers of modern human beings. Atherosclerosis is a long-term, 30-40 years hidden process of development. Carotid atherosclerosis, is the manifestation of systemic atherosclerosis in the carotid arteries, and usually occurs in adolescents and becomes progressively worse with age. It is currently believed that carotid arteriosclerosis is closely related to the occurrence of ischemic stroke in the elderly. The carotid arteriosclerosis is firstly shown as intima-media thickening in the early stage, then an atherosclerotic plaque is gradually formed, and then, the plaque internal hemorrhage, plaque rupture and detachment, mural thrombosis, secondary vessel stenosis and the like are caused on the basis, so that the corresponding hemodynamics is changed, and the occurrence of ischemic cerebrovascular events is caused. Recent studies suggest that: the intima-media thickness in the carotid artery has obvious correlation with the occurrence of myocardial infarction; the risk of myocardial infarction increases by 11% for every 0.1mm increase in intima-media thickness.
B-mode ultrasound is the only instrument currently available that can dynamically observe and measure the intima-media thickness in carotid arteries in real time and without damage. On ultrasound images, normal carotid and femoral wall structures appear as typical "double-line features". The near-lumen side echo line is formed by the blood-intima interface, the near-wall side echo line is formed by the interface between the media and the adventitia, and the distance between the two echo lines is the intima-media thickness (IMT). Normally, the intima is thin and difficult to measure, while atherosclerosis occurs in the intima with small changes in media thickness, so the intima-media thickness is often measured to observe atherosclerotic lesions. The Intima-Media Thickness (IMT) of the carotid artery blood vessel wall in ultrasonic imaging can be used as an important index for evaluating the early lesion degree of cardiovascular diseases, and has important value for diagnosing and preventing sudden myocardial infarction and stroke.
In the conventional method, the Intima-Media thickness of the carotid artery ultrasound image is usually obtained by manual measurement, and a measurer manually delineates a Lumen-Intima Interface (LII) and a Media-Adventitia Interface (MAI) in the image, and then obtains the IMT by calculating the distance between the two boundaries. Normal IMT should be less than 1.0mm, IMT is intimal thickening between 1.0-1.2 mm, plaque formation between 1.2-1.4 mm, IMT greater than 1.4mm is carotid stenosis. However, such manual measurement is labor intensive, time consuming, and the final results obtained are influenced by the training, personal experience, and subjective judgment of the tester.
In addition, under ultrasound, according to morphological and echogenic features, plaques can be classified as: 1. hypoechoic fatty soft plaques; 2. moderate echogenic fibrous flat plaques rich in collagen tissue; 3. calcified hard plaque with hyperechoic sound shadow; 4. ulcerative mixed plaques with unequal echo intensities. Among them, soft plaque, flat plaque and mixed plaque are unstable plaque, and are one of the important causes of ischemic stroke.
For the case of plaque and intima-media simultaneously, the current approach is to choose to avoid plaque for intima-media measurements.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a carotid artery intima-media measurement and plaque detection method, which reduces false positive of a model and improves the measurement accuracy and the measurement efficiency.
In order to achieve the purpose, the invention adopts the following technical scheme that:
a carotid artery intima-media measurement and plaque detection method comprises the following steps:
and S1, aiming at the carotid artery ultrasonic scanning image, respectively acquiring an original blood vessel segmentation image, an original intima-media segmentation image and an original plaque detection frame in the carotid artery ultrasonic scanning image.
S2, performing blood vessel segmentation post-processing on the original blood vessel segmentation map acquired in the step S1 to obtain an optimized blood vessel segmentation map;
s3, performing plaque detection frame post-processing on the original plaque detection frame obtained in the step S1 based on the optimized blood vessel segmentation map to obtain an optimized plaque detection frame;
s4, based on the optimized blood vessel segmentation map, carrying out intra-intima-media segmentation post-processing on the original intra-intima-media segmentation map obtained in the step S1 to obtain an optimized intra-media segmentation map;
and S5, performing intima-media thickness measurement on the intima-media area in the optimized intima-media segmentation map.
Preferably, in step S4, the inner membrane division post-processing mode is:
s41, extracting the outer contour of each intima-media region in the original intima-media segmentation graph, respectively calculating the outer contour area of each intima-media region, and eliminating the intima-media region with the outer contour area smaller than a set threshold value intima _ area _ thr;
s42, aiming at the remaining non-removed intima-media region in the step S41, respectively calculating the intersection area of the intima-media region and each blood vessel region in the optimized blood vessel segmentation graph, respectively calculating the ratio of each intersection area to the intima-media region, and removing all intima-media regions with the ratio smaller than a set threshold value intima _ thr;
s43, the remaining regions of the inner and middle membranes which are not removed in the step S42 form an optimized inner and middle membrane segmentation map.
Preferably, in step S3, the post-processing method of the blob detection frame is as follows: for each original plaque detection frame, respectively calculating the intersection area of each blood vessel region in the original plaque detection frame and the optimized blood vessel segmentation image, respectively calculating the ratio of each intersection area to the area of the original plaque detection frame, removing all original plaque detection frames with the ratio smaller than a set threshold value plain _ thr, and taking the remaining original plaque detection frames which are not removed as the optimized plaque detection frames.
Preferably, in step S2, the blood vessel segmentation post-processing method is: extracting the outer contour of each blood vessel region in the original blood vessel segmentation graph, respectively calculating the outer contour area of each blood vessel region, eliminating the blood vessel regions with the outer contour area smaller than a set threshold value vessel _ area _ thr, and forming the optimized blood vessel segmentation graph by remaining the blood vessel regions which are not eliminated.
Preferably, in step S5, the specific manner of measuring the thickness of the intima-media film is as follows:
s51, extracting the outline of the inner tunica media area to obtain the outline pixel point coordinates on the outline outside the inner tunica media area; wherein, the horizontal axis of the coordinate, namely the X-axis direction, is consistent with the extension direction of the inner mesolamella;
s52, extracting the maximum value Xmax and the minimum value Xmin of the horizontal axis coordinate from all the contour pixel point coordinates;
s53, traversing contour pixel points on each horizontal axis coordinate from the minimum value Xmin to the maximum value Xmax, if two or more contour pixel points have the same horizontal axis coordinate, taking the contour pixel point with the maximum vertical axis coordinate as an upper contour pixel point, and taking the contour pixel point with the minimum vertical axis coordinate as a lower contour pixel point;
s54, for each upper contour pixel point, respectively calculating the distance from the upper contour pixel point to each lower contour pixel point, selecting the minimum distance as the downward distance of the upper contour pixel point, and the lower contour pixel point corresponding to the minimum distance is the lower measuring point of the upper contour pixel point;
s55, selecting the largest downward distance as the inner and middle membrane thickness of the inner and middle membrane area according to the downward distance of each upper contour pixel point; the upper contour pixel point corresponding to the maximum downward distance is an upper measuring point of the thickness of the inner and middle membranes, and the lower measuring point of the upper contour pixel point is a lower measuring point of the thickness of the inner and middle membranes.
Preferably, the carotid artery ultrasound scanning image is a frame image of a carotid artery ultrasound scanning video.
Preferably, the carotid artery ultrasonic scanning image is processed by utilizing a multitask network, the carotid artery ultrasonic scanning image is input into the multitask network, the multitask network respectively obtains an original plaque detection frame, an original blood vessel segmentation image and an original intima-media segmentation image in the carotid artery ultrasonic scanning image, and the optimized blood vessel segmentation image, the optimized intima-media segmentation image, the optimized plaque detection frame and the intima-media thickness of the output intima-media area are obtained through processing.
Preferably, the multitask network is generated by training sample data of carotid artery ultrasonic scanning images; sample data of the carotid artery ultrasound scanning image comprises: carotid artery ultrasonic scanning images, and plaque detection frames, blood vessel segmentation maps, intima-media segmentation maps and intima-media thicknesses of intima-media areas of the carotid artery ultrasonic scanning images.
The invention has the advantages that:
(1) based on the characteristic that the intima-media and the plaque exist in the blood vessel, the plaque detection frame and the intima-media segmentation region which are overlapped with the blood vessel too little or not overlapped with the blood vessel are filtered, false positive false alarm of the model is reduced, and the efficiency and the accuracy of subsequent intima-media measurement are improved.
(2) The invention can rapidly measure the thickness of the intima-media membrane based on the intima-media membrane segmentation region, saves the time for a doctor to operate an ultrasonic machine to measure the thickness of the intima-media membrane, improves the measurement efficiency and the measurement accuracy, and completely meets the real-time requirement.
(3) According to the rapid measuring method for the thickness of the intima-media film, the main direction of the intima-media film is basically the horizontal direction, so that the upper and lower contours of the intima-media film can be split based on the abscissa by utilizing the characteristic, and the position with the thickest thickness of the intima-media film can be rapidly calculated based on the distance relation between the upper and lower contour points.
(4) The invention is based on the intelligent detection of dynamic carotid artery ultrasonic scanning video, can automatically detect plaque, segment blood vessels, segment intima-media and measure intima-media thickness in real time while scanning patients, and can meet the real-time requirement.
(5) In the prior art, carotid artery blood vessels are started from a single task of plaque detection or intima-media measurement, the invention uses a multi-task network which can simultaneously carry out plaque detection, blood vessel segmentation and intima-media segmentation, and simultaneously solves two tasks of plaque detection and intima-media measurement. In addition, due to the fact that the multi-task network simultaneously learns a plurality of different tasks, the different tasks share the same backbone network, learning information is richer, learning limitation is stricter, the tasks have correlation, and compared with a single-task network, the model accuracy is relatively higher, so that false positive false alarm is lower, and meanwhile storage and calculation costs brought by multiple models are greatly reduced due to part of shared model parameters. In addition, false positive false reports which are not in a blood vessel region are filtered by means of the position relation between the blood vessel segmentation and the intima-media and plaque, and false reports which possibly appear in the model cannot be further filtered by the aid of the characteristics of the blood vessel by the aid of a single-task network model.
(6) In the prior art, the traditional digital image processing technology or the traditional digital image processing and deep learning combined technology is used for extracting the lumen-intima boundary and the media-adventitia boundary to achieve the purpose of calculating the thickness of the intima-media membrane. The method utilizes massive labeled data to directly learn the inner and middle membrane segmentation areas, and a data-driven deep learning method has better generalization compared with the traditional method and can continuously improve the performance along with the increase of the data.
Drawings
FIG. 1 is a flow chart of a carotid intima-media measurement and plaque detection method of the present invention.
Figure 2 is a schematic view of an ultrasound scan of carotid vessels, plaque and intima.
FIG. 3 is a process diagram of the present invention multitasking network for endomembrane measurement and plaque detection.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1 and 3, a carotid artery intima-media measurement and plaque detection method includes the following steps:
and S1, reading the frame image of the carotid artery ultrasound scanning video, inputting the frame image of the carotid artery ultrasound scanning video into a multitask network, and respectively acquiring the original plaque detection frame, the original blood vessel segmentation image and the original intima-media segmentation image by the multitask network.
The multitask network is generated by training sample data of carotid artery ultrasonic scanning images; sample data of the carotid artery ultrasound scanning image comprises: carotid artery ultrasonic scanning images, and plaque detection frames, blood vessel segmentation maps, intima-media segmentation maps and intima-media thicknesses of intima-media areas of the carotid artery ultrasonic scanning images.
S2, the multitask network carries out blood vessel segmentation post-processing on the original blood vessel segmentation map acquired in the step S1 to obtain an optimized blood vessel segmentation map.
The blood vessel segmentation post-processing mainly uses a correlation function in an opencv open source digital image processing tool to process an original blood vessel segmentation image, and specifically comprises the following steps: firstly, carrying out binarization processing on an original blood vessel segmentation graph by using a threshold function to obtain a binary image, then extracting the outline of a blood vessel region from the binary image by using a findContours function, then calculating the outline area of the blood vessel region by using a contourArea function, sequencing the blood vessel regions according to the outline areas, eliminating the blood vessel region of which the outline area is less than a set threshold value vessel _ area _ thr, and forming an optimized blood vessel segmentation graph by using the remaining blood vessel region which is not eliminated.
S3, the multitask network carries out patch detection frame post-processing on the original patch detection frame obtained in the step S1 to obtain an optimized patch detection frame.
The post-processing of the plaque detection frame specifically comprises the following steps: for each original plaque detection frame, respectively calculating the intersection area of the original plaque detection frame and each blood vessel region in the optimized blood vessel segmentation image obtained in the step S2, respectively calculating the ratio of each intersection area to the area of the original plaque detection frame, removing all original plaque detection frames with the ratio smaller than a set threshold value plain _ thr, and taking the remaining original plaque detection frames which are not removed as the optimized plaque detection frames.
S4, the multitask network carries out inner and middle membrane segmentation post-processing on the original inner and middle membrane segmentation map acquired in the step S1 to obtain an optimized inner and middle membrane segmentation map.
The internal and middle membrane segmentation post-processing is also to process an original internal and middle membrane segmentation graph by using a function in an opencv open source digital image processing tool, and specifically comprises the following steps:
s41, firstly, carrying out binarization processing on an original inner and middle membrane segmentation graph by using a threshold function to obtain a binarized image, then extracting the outer contour of an inner and middle membrane region from the binarized image by using a findContours function, then calculating the outer contour area of the inner and middle membrane region by using a contourArea function, sequencing the inner and middle membrane region according to the contour area, and removing the inner and middle membrane region of which the contour area is smaller than a set threshold value intima _ area _ thr;
s42, aiming at the remaining non-removed intima-media region in the step S41, respectively calculating the intersection area of the intima-media region and each blood vessel region in the optimized blood vessel segmentation graph, respectively calculating the ratio of each intersection area to the intima-media region, and removing all intima-media regions with the ratio smaller than a set threshold value intima _ thr;
s43, the remaining regions of the inner and middle membranes which are not removed in the step S42 form an optimized inner and middle membrane segmentation map.
And S5, the multitask network obtains an optimized inner and middle membrane segmentation map according to the step S4, and inner and middle membrane thickness measurement is carried out on each inner and middle membrane area in the optimized inner and middle membrane segmentation map.
The specific mode of measuring the thickness of the inner tunica media is as follows:
s51, extracting the outer contour of the intima-media area, wherein the contour is approximately horizontal when scanning carotid artery blood vessels, and the main direction of the intima-media area, namely the extension direction, is also approximately horizontal as shown in figure 2, so that the coordinate horizontal axis, namely the X-axis direction, is consistent with the extension direction of the intima-media area; obtaining the coordinates of contour pixel points on the outer contour of the inner mesolamella region;
s52, extracting the maximum value Xmax and the minimum value Xmin of the horizontal axis coordinate from all the contour pixel point coordinates; the outline pixel point with the horizontal axis coordinate as the maximum value Xmax is the outline right end point, and the outline pixel point with the horizontal axis coordinate as the minimum value Xmin is the outline left end point;
s53, splitting the outline into an upper outline and a lower outline according to the left end point and the right end point of the outline outside the inner tunica media area obtained in the step S52;
traversing contour pixel points on each horizontal axis coordinate from the minimum value Xmin to the maximum value Xmax, and if two or more contour pixel points have the same horizontal axis coordinate, taking the contour pixel point with the maximum vertical axis coordinate as an upper contour pixel point and taking the contour pixel point with the minimum vertical axis coordinate as a lower contour pixel point;
s54, respectively calculating the distance from each upper contour pixel point to each lower contour pixel point, selecting the minimum distance as the downward distance of the upper contour pixel point, and the lower contour pixel point corresponding to the minimum distance is the lower measuring point of the upper contour pixel point;
s55, selecting the largest downward distance as the inner and middle membrane thickness of the inner and middle membrane area according to the downward distance of each upper contour pixel point; the upper contour pixel point corresponding to the maximum downward distance is an upper measuring point of the thickness of the inner and middle membranes, and the lower measuring point of the upper contour pixel point is a lower measuring point of the thickness of the inner and middle membranes.
The invention can automatically detect plaque, cut blood vessel, cut intima-media and measure intima-media thickness in real time while scanning a patient, filters out a plaque detection frame and an intima-media segmentation area which are not overlapped with the blood vessel based on the characteristic that the plaque and the intima-media exist in the blood vessel, reduces false positive false alarm of a model, saves time for a doctor to operate an ultrasonic machine to measure the intima-media thickness based on rapid measurement of the intima-media segmentation area, improves efficiency and completely meets the real-time requirement.
The carotid artery ultrasonic scanning image is processed by utilizing a multitask network, the carotid artery ultrasonic scanning image is input into the multitask network, the multitask network respectively obtains an original plaque detection frame, an original blood vessel segmentation image and an original intima-media segmentation image in the carotid artery ultrasonic scanning image, and the optimized blood vessel segmentation image, the optimized intima-media segmentation image, the optimized plaque detection frame and the intima-media thickness of an output intima-media area are obtained through processing. The multitask network is generated by training sample data of carotid artery ultrasonic scanning images; sample data of the carotid artery ultrasound scanning image comprises: carotid artery ultrasonic scanning images, and plaque detection frames, blood vessel segmentation maps, intima-media segmentation maps and intima-media thicknesses of intima-media areas of the carotid artery ultrasonic scanning images.
In the prior art, carotid artery blood vessels are started from a single task of plaque detection or intima-media measurement, the invention uses a multi-task network which can simultaneously carry out plaque detection, blood vessel segmentation and intima-media segmentation, and simultaneously solves two tasks of plaque detection and intima-media measurement. In addition, due to the fact that the multi-task network simultaneously learns a plurality of different tasks, the different tasks share the same backbone network, learning information is richer, learning limitation is stricter, correlation exists among the tasks, and model accuracy is higher relative to a single-task network, false positive false alarm is lower, and storage and calculation costs brought by multiple models are greatly reduced due to the fact that part of shared model parameters. In addition, false positive false reports which are not in a blood vessel region are filtered by means of the position relation between the blood vessel segmentation and the intima-media and plaque, and false reports which possibly appear in the model cannot be further filtered by the aid of the characteristics of the blood vessel by the aid of a single-task network model.
In the prior art, the traditional digital image processing technology or the combination technology of traditional digital image processing and deep learning is used for extracting the lumen-intima boundary and the media-adventitia boundary to achieve the purpose of calculating the thickness of the inner media and the media. The method utilizes massive labeled data to directly learn the inner and middle membrane segmentation areas, and a data-driven deep learning method has better generalization compared with the traditional method and can continuously improve the performance along with the increase of the data.
The invention is not to be considered as limited to the specific embodiments shown and described, but is to be understood to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the appended claims.
Claims (8)
1. A carotid artery intima-media measurement and plaque detection method is characterized by comprising the following steps:
s1, aiming at the carotid artery ultrasonic scanning image, respectively acquiring an original blood vessel segmentation image, an original intima-media segmentation image and an original plaque detection frame in the carotid artery ultrasonic scanning image;
s2, performing blood vessel segmentation post-processing on the original blood vessel segmentation map acquired in the step S1 to obtain an optimized blood vessel segmentation map;
s3, performing plaque detection frame post-processing on the original plaque detection frame obtained in the step S1 based on the optimized blood vessel segmentation map to obtain an optimized plaque detection frame;
s4, based on the optimized blood vessel segmentation map, performing inner and middle membrane segmentation post-processing on the original inner and middle membrane segmentation map obtained in the step S1 to obtain an optimized inner and middle membrane segmentation map;
and S5, performing intima-media thickness measurement on the intima-media area in the optimized intima-media segmentation map.
2. The method for carotid intima-media measurement and plaque detection according to claim 1, wherein in step S4, the intima-media segmentation post-processing method is as follows:
s41, extracting the outer contour of each intima-media region in the original intima-media segmentation graph, respectively calculating the outer contour area of each intima-media region, and eliminating the intima-media region with the outer contour area smaller than a set threshold value intima _ area _ thr;
s42, aiming at the remaining non-removed intima-media region in the step S41, respectively calculating the intersection area of the intima-media region and each blood vessel region in the optimized blood vessel segmentation graph, respectively calculating the ratio of each intersection area to the intima-media region, and removing all intima-media regions with the ratio smaller than a set threshold value intima _ thr;
s43, the remaining regions of the inner and middle membranes which are not removed in the step S42 form an optimized inner and middle membrane segmentation map.
3. The carotid intima-media measurement and plaque detection method according to claim 1, wherein in step S3, the plaque detection frame is post-processed by: for each original plaque detection frame, respectively calculating the intersection area of each blood vessel region in the original plaque detection frame and the optimized blood vessel segmentation image, respectively calculating the ratio of each intersection area to the area of the original plaque detection frame, removing all original plaque detection frames with the ratio smaller than a set threshold value plain _ thr, and taking the remaining original plaque detection frames which are not removed as the optimized plaque detection frames.
4. The carotid intima-media measurement and plaque detection method according to claim 1, 2 or 3, characterized in that in step S2, the post-segmentation processing manner of the blood vessel is: extracting the outer contour of each blood vessel region in the original blood vessel segmentation graph, respectively calculating the outer contour area of each blood vessel region, eliminating the blood vessel regions with the outer contour area smaller than a set threshold value vessel _ area _ thr, and forming the optimized blood vessel segmentation graph by remaining the blood vessel regions which are not eliminated.
5. The method for carotid intima-media measurement and plaque detection according to claim 1, wherein in step S5, the intima-media thickness measurement is specifically performed by:
s51, extracting the outer contour of the inner tunica media area to obtain the contour pixel point coordinates on the outer contour of the inner tunica media area; wherein, the horizontal axis of the coordinate, namely the X-axis direction, is consistent with the extension direction of the inner mesolamella;
s52, extracting the maximum value Xmax and the minimum value Xmin of the horizontal axis coordinate from all the contour pixel point coordinates;
s53, traversing contour pixel points on each horizontal axis coordinate from the minimum value Xmin to the maximum value Xmax, if two or more contour pixel points have the same horizontal axis coordinate, taking the contour pixel point with the maximum vertical axis coordinate as an upper contour pixel point, and taking the contour pixel point with the minimum vertical axis coordinate as a lower contour pixel point;
s54, respectively calculating the distance from each upper contour pixel point to each lower contour pixel point, selecting the minimum distance as the downward distance of the upper contour pixel point, and the lower contour pixel point corresponding to the minimum distance is the lower measuring point of the upper contour pixel point;
s55, selecting the largest downward distance as the inner and middle membrane thickness of the inner and middle membrane area according to the downward distance of each upper contour pixel point; the upper contour pixel point corresponding to the maximum downward distance is the upper measurement point of the thickness of the inner-middle membrane, and the lower measurement point of the upper contour pixel point is the lower measurement point of the thickness of the inner-middle membrane.
6. The method for carotid intima-media measurement and plaque detection according to claim 1, wherein the carotid ultrasound scanning image is a frame image of a carotid ultrasound scanning video.
7. The carotid intima-media measurement and plaque detection method according to claim 1, characterized in that a multitask network is used to process carotid artery ultrasound scanning images, the carotid artery ultrasound scanning images are input into the multitask network, the multitask network respectively obtains an original plaque detection frame, an original blood vessel segmentation map and an original intima-media segmentation map in the carotid artery ultrasound scanning images, and the optimized blood vessel segmentation map, the optimized intima-media segmentation map, the optimized plaque detection frame and the intima-media thickness of the output intima-media region are obtained through processing.
8. The carotid artery intima-media measurement and plaque detection method according to claim 7, wherein the multitask network is generated by training with sample data of carotid artery ultrasound scanning images; sample data of the carotid artery ultrasound scanning image comprises: carotid artery ultrasonic scanning images, and plaque detection frames, blood vessel segmentation maps, intima-media segmentation maps and intima-media thicknesses of intima-media areas of the carotid artery ultrasonic scanning images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210960871.2A CN115049642A (en) | 2022-08-11 | 2022-08-11 | Carotid artery blood vessel intima-media measurement and plaque detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210960871.2A CN115049642A (en) | 2022-08-11 | 2022-08-11 | Carotid artery blood vessel intima-media measurement and plaque detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115049642A true CN115049642A (en) | 2022-09-13 |
Family
ID=83168162
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210960871.2A Pending CN115049642A (en) | 2022-08-11 | 2022-08-11 | Carotid artery blood vessel intima-media measurement and plaque detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115049642A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116898483A (en) * | 2023-07-17 | 2023-10-20 | 逸超医疗科技(北京)有限公司 | Accurate ultrasonic carotid intima-media thickness measurement method |
| CN117078695A (en) * | 2023-08-18 | 2023-11-17 | 内蒙古工业大学 | A deep learning-based carotid artery plaque ultrasound image recognition and segmentation method |
| CN117173101A (en) * | 2023-08-01 | 2023-12-05 | 逸超医疗科技(北京)有限公司 | A method for automatic identification and elimination of liver blood vessels in acoustic attenuation imaging |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101872425A (en) * | 2010-07-29 | 2010-10-27 | 哈尔滨工业大学 | Obtaining Image Features and Measuring Corresponding Physical Parameters Based on Empirical Mode Decomposition |
| CN102800088A (en) * | 2012-06-28 | 2012-11-28 | 华中科技大学 | Automatic dividing method of ultrasound carotid artery plaque |
| CN104700422A (en) * | 2015-03-27 | 2015-06-10 | 深圳市美侨医疗科技有限公司 | Method for automatically segmenting bonded red blood cells and white blood cells in urinary sediment image |
| CN107316077A (en) * | 2017-06-21 | 2017-11-03 | 上海交通大学 | A kind of fat cell automatic counting method based on image segmentation and rim detection |
| CN110570350A (en) * | 2019-09-11 | 2019-12-13 | 深圳开立生物医疗科技股份有限公司 | two-dimensional follicle detection method and device, ultrasonic equipment and readable storage medium |
| CN111192251A (en) * | 2019-12-30 | 2020-05-22 | 上海交通大学医学院附属国际和平妇幼保健院 | Follicle ultrasonic processing method and system based on level set image segmentation |
| CN111476794A (en) * | 2019-01-24 | 2020-07-31 | 武汉兰丁医学高科技有限公司 | UNET-based cervical pathological tissue segmentation method |
| CN112967277A (en) * | 2021-03-31 | 2021-06-15 | 成都思多科医疗科技有限公司 | Carotid artery ultrasound image blood vessel and intima positioning method based on deep learning network |
| CN114052794A (en) * | 2021-10-13 | 2022-02-18 | 山东大学 | A carotid ultrasound report generation system based on multimodal information |
| CN114240846A (en) * | 2021-11-23 | 2022-03-25 | 复旦大学附属华山医院 | System and method for reducing false positive rate of medical image lesion segmentation results |
| CN114764761A (en) * | 2020-12-30 | 2022-07-19 | 无锡科美达医疗科技有限公司 | Automatic carotid artery ultrasound image intima-media measurement method |
-
2022
- 2022-08-11 CN CN202210960871.2A patent/CN115049642A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101872425A (en) * | 2010-07-29 | 2010-10-27 | 哈尔滨工业大学 | Obtaining Image Features and Measuring Corresponding Physical Parameters Based on Empirical Mode Decomposition |
| CN102800088A (en) * | 2012-06-28 | 2012-11-28 | 华中科技大学 | Automatic dividing method of ultrasound carotid artery plaque |
| CN104700422A (en) * | 2015-03-27 | 2015-06-10 | 深圳市美侨医疗科技有限公司 | Method for automatically segmenting bonded red blood cells and white blood cells in urinary sediment image |
| CN107316077A (en) * | 2017-06-21 | 2017-11-03 | 上海交通大学 | A kind of fat cell automatic counting method based on image segmentation and rim detection |
| CN111476794A (en) * | 2019-01-24 | 2020-07-31 | 武汉兰丁医学高科技有限公司 | UNET-based cervical pathological tissue segmentation method |
| CN110570350A (en) * | 2019-09-11 | 2019-12-13 | 深圳开立生物医疗科技股份有限公司 | two-dimensional follicle detection method and device, ultrasonic equipment and readable storage medium |
| CN111192251A (en) * | 2019-12-30 | 2020-05-22 | 上海交通大学医学院附属国际和平妇幼保健院 | Follicle ultrasonic processing method and system based on level set image segmentation |
| CN114764761A (en) * | 2020-12-30 | 2022-07-19 | 无锡科美达医疗科技有限公司 | Automatic carotid artery ultrasound image intima-media measurement method |
| CN112967277A (en) * | 2021-03-31 | 2021-06-15 | 成都思多科医疗科技有限公司 | Carotid artery ultrasound image blood vessel and intima positioning method based on deep learning network |
| CN114052794A (en) * | 2021-10-13 | 2022-02-18 | 山东大学 | A carotid ultrasound report generation system based on multimodal information |
| CN114240846A (en) * | 2021-11-23 | 2022-03-25 | 复旦大学附属华山医院 | System and method for reducing false positive rate of medical image lesion segmentation results |
Non-Patent Citations (2)
| Title |
|---|
| XIANGJING AN ET AL.: "Faster R-CNN for Detection of Carotid Plaque on Ultrasound Images", 《2019 COMPUTING, COMMUNICATIONS AND IOT APPLICATIONS (COMCOMAP)》 * |
| 陈余航等: "颈动脉超声影像内中膜智能分割和斑块识别", 《机电一体化》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116898483A (en) * | 2023-07-17 | 2023-10-20 | 逸超医疗科技(北京)有限公司 | Accurate ultrasonic carotid intima-media thickness measurement method |
| CN116898483B (en) * | 2023-07-17 | 2024-10-18 | 逸超医疗科技(北京)有限公司 | Accurate ultrasonic carotid intima-media thickness measurement method |
| CN117173101A (en) * | 2023-08-01 | 2023-12-05 | 逸超医疗科技(北京)有限公司 | A method for automatic identification and elimination of liver blood vessels in acoustic attenuation imaging |
| CN117078695A (en) * | 2023-08-18 | 2023-11-17 | 内蒙古工业大学 | A deep learning-based carotid artery plaque ultrasound image recognition and segmentation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115049642A (en) | Carotid artery blood vessel intima-media measurement and plaque detection method | |
| Loizou et al. | An integrated system for the segmentation of atherosclerotic carotid plaque | |
| Loizou et al. | Manual and automated media and intima thickness measurements of the common carotid artery | |
| CN102163326B (en) | Method for automatic computerized segmentation and analysis on thickness uniformity of intima media of carotid artery blood wall in sonographic image | |
| JP6106190B2 (en) | Visualization method of blood and blood likelihood in blood vessel image | |
| US12026886B2 (en) | Method and system for automatically estimating a hepatorenal index from ultrasound images | |
| JP2020018694A (en) | Ultrasonic diagnostic apparatus and ultrasonic image processing method | |
| CN109003280A (en) | Inner membrance dividing method in a kind of blood vessel of binary channels intravascular ultrasound image | |
| CN112971844A (en) | Ultrasonic image acquisition quality evaluation method and ultrasonic imaging equipment | |
| JP2002269539A (en) | Image processor, image processing method, and computer- readable storage medium with image processing program stored therein, and diagnosis support system using them | |
| CN111476790A (en) | Method and device for enhancing display of puncture needle in ultrasonic puncture | |
| EP4076208B1 (en) | Systems and methods for assessing a placenta | |
| Loizou et al. | Atherosclerotic carotid plaque segmentation | |
| CN107169978A (en) | Ultrasonoscopy edge detection method and system | |
| Ilea et al. | An automatic 2D CAD algorithm for the segmentation of the IMT in ultrasound carotid artery images | |
| CN111434311A (en) | Ultrasonic imaging device and image processing method | |
| JP3662835B2 (en) | Ultrasonic diagnostic equipment | |
| CN112294361B (en) | Ultrasonic imaging device and method for generating cross-sectional images of pelvic floor | |
| Peng et al. | A multiscale morphological approach to local contrast enhancement for ultrasound images | |
| JP2023051175A (en) | Computer program, information processing method, and information processing apparatus | |
| CN115517709A (en) | Ultrasound imaging method and ultrasound imaging system | |
| Menchón-Lara et al. | Automatic evaluation of carotid intima-media thickness in ultrasounds using machine learning | |
| EP4014884A1 (en) | Apparatus for use in analysing an ultrasound image of a subject | |
| CN119138923A (en) | Ultrasonic imaging apparatus and ultrasonic imaging method | |
| Zahnd et al. | A new user-independent in vivo method for 2D motion estimation of the carotid wall by ultrasound imaging for early detection of pathological behavior |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220913 |
|
| RJ01 | Rejection of invention patent application after publication |