CN117745641A - Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image - Google Patents
Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image Download PDFInfo
- Publication number
- CN117745641A CN117745641A CN202311477076.9A CN202311477076A CN117745641A CN 117745641 A CN117745641 A CN 117745641A CN 202311477076 A CN202311477076 A CN 202311477076A CN 117745641 A CN117745641 A CN 117745641A
- Authority
- CN
- China
- Prior art keywords
- calcification
- calcified plaque
- optimized
- plaque
- calcified
- 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
- 238000007917 intracranial administration Methods 0.000 title claims abstract description 98
- 210000001367 artery Anatomy 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 63
- 230000002308 calcification Effects 0.000 claims abstract description 241
- 230000011218 segmentation Effects 0.000 claims abstract description 112
- 238000004445 quantitative analysis Methods 0.000 claims abstract description 67
- 238000013528 artificial neural network Methods 0.000 claims abstract description 31
- 238000001514 detection method Methods 0.000 claims abstract description 29
- 239000000835 fiber Substances 0.000 claims description 32
- 238000004590 computer program Methods 0.000 claims description 13
- 238000003384 imaging method Methods 0.000 claims description 13
- 230000007797 corrosion Effects 0.000 claims description 12
- 238000005260 corrosion Methods 0.000 claims description 12
- 230000000306 recurrent effect Effects 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 10
- 230000015654 memory Effects 0.000 claims description 9
- OYPRJOBELJOOCE-UHFFFAOYSA-N Calcium Chemical compound [Ca] OYPRJOBELJOOCE-UHFFFAOYSA-N 0.000 claims description 8
- 229910052791 calcium Inorganic materials 0.000 claims description 8
- 239000011575 calcium Substances 0.000 claims description 8
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 238000011002 quantification Methods 0.000 claims 2
- 238000004458 analytical method Methods 0.000 abstract description 13
- 208000037803 restenosis Diseases 0.000 abstract description 12
- 230000002980 postoperative effect Effects 0.000 abstract description 8
- 238000011156 evaluation Methods 0.000 abstract description 5
- 208000004434 Calcinosis Diseases 0.000 description 189
- 238000012014 optical coherence tomography Methods 0.000 description 18
- 239000000523 sample Substances 0.000 description 15
- 238000002372 labelling Methods 0.000 description 14
- 210000004204 blood vessel Anatomy 0.000 description 9
- 238000005259 measurement Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000002513 implantation Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 210000001715 carotid artery Anatomy 0.000 description 4
- 230000007213 cerebrovascular event Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 210000001519 tissue Anatomy 0.000 description 4
- 230000002792 vascular Effects 0.000 description 4
- 239000011800 void material Substances 0.000 description 4
- 244000208734 Pisonia aculeata Species 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000001356 surgical procedure Methods 0.000 description 3
- 208000037260 Atherosclerotic Plaque Diseases 0.000 description 2
- 206010050403 Carotid artery dissection Diseases 0.000 description 2
- 238000005481 NMR spectroscopy Methods 0.000 description 2
- 230000003143 atherosclerotic effect Effects 0.000 description 2
- 159000000007 calcium salts Chemical class 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000003759 clinical diagnosis Methods 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000002059 diagnostic imaging Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 239000002207 metabolite Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000004393 prognosis Methods 0.000 description 2
- 238000002601 radiography Methods 0.000 description 2
- 206010002383 Angina Pectoris Diseases 0.000 description 1
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 208000012902 Nervous system disease Diseases 0.000 description 1
- 208000031481 Pathologic Constriction Diseases 0.000 description 1
- 208000006011 Stroke Diseases 0.000 description 1
- 210000000709 aorta Anatomy 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 210000001175 cerebrospinal fluid Anatomy 0.000 description 1
- 208000026106 cerebrovascular disease Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 210000004351 coronary vessel Anatomy 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000000004 hemodynamic effect Effects 0.000 description 1
- 201000005851 intracranial arteriosclerosis Diseases 0.000 description 1
- 239000012528 membrane Substances 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
- 210000003205 muscle Anatomy 0.000 description 1
- 208000010125 myocardial infarction Diseases 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000036262 stenosis Effects 0.000 description 1
- 208000037804 stenosis Diseases 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000010408 sweeping Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention provides a method for detecting and quantitatively analyzing calcified plaque based on intracranial artery images, which comprises the following steps: acquiring an intracranial artery image; inputting the intracranial artery image into the segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model; inputting the fibrous plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fibrous plaque area and an optimized calcified plaque area which are output by the circulating neural network; determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region; and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result. The quantitative analysis result comprises preoperative calcification detection and property evaluation and postoperative restenosis risk prediction, so that the analysis result has more reference value for experts, and the accuracy and reliability of calcified plaque quantitative analysis are improved.
Description
Technical Field
The invention relates to the technical field of calcified plaque detection and quantitative analysis, in particular to a method for detecting and quantitatively analyzing calcified plaque based on intracranial artery images.
Background
Calcified plaque is a metabolite in biological tissue, which refers to the area of the atherosclerotic plaque tissue where calcium salts are deposited, and is usually detected and displayed in medical imaging by X-ray, CT scanning, nuclear magnetic resonance, etc. imaging techniques.
For an expert, calcified plaques may be used to predict the risk of occurrence of a cerebrovascular event. By analyzing the number, distribution and density of calcified plaques, the likelihood of a patient suffering from a cerebrovascular event can be assessed, thereby taking appropriate precautions. The presence and extent of calcified plaque may affect the choice of surgical procedure, such as analysis of carotid plaque, and if plaque calcification is of a higher hardness and greater area may affect stent placement, the operator may choose carotid artery dissection rather than carotid artery stent implantation. In addition, different parameters of calcified plaque (angle, thickness, depth, etc.) can affect stent placement, stent expansion, and long-term prognosis.
However, in clinical applications, it is often necessary to rely on an expert to manually find calcified regions from a large number of sequential images of the intracranial artery OCT (Optical Coherence Tomography) and manually measure important parameters using auxiliary tools, which can create significant time and labor costs depending on the length of OCT images generated by a single pullback and the number of calcified plaques actually present. Meanwhile, the accuracy of parameter measurement and plaque characterization depends on the expertise of an expert.
Disclosure of Invention
The invention provides a method for detecting and quantitatively analyzing calcified plaques based on intracranial artery images, which is used for solving the defect that the length of an OCT image generated by single pullback and the number of calcified plaques actually existing in the detection of the calcified plaques of the intracranial artery images in the prior art probably generate huge time and labor cost, and meanwhile, the accuracy of parameter measurement and plaque qualitative dependence on the professional degree of an expert.
The invention provides a method for detecting and quantitatively analyzing calcified plaque based on intracranial artery images, which comprises the following steps:
acquiring an intracranial artery image;
inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model;
inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network;
determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region;
and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the calcified plaque angle, the calcified depth, the calcified thickness and the calcified length of the calcified plaque are determined based on the optimized fibrous plaque area and the optimized calcified plaque area, and the method comprises the following steps:
extracting a lumen wall from the optimized fibrous plaque region;
determining a blank area lumen wall of the lumen wall, and determining a lumen center based on the lumen wall and the blank area lumen wall;
the calcification angle, the calcification depth, the calcification thickness and the calcification length are determined based on the lumen center and the optimized calcified plaque region.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the method for determining the calcified angle, the calcified depth, the calcified thickness and the calcified length based on the lumen center and the optimized calcified plaque area comprises the following steps:
scanning the optimized calcified plaque area by taking the center of the lumen as a polar coordinate center to obtain a first scanning coordinate and a last scanning coordinate of the optimized calcified plaque area, and obtaining the number of first pixel points corresponding to each depth of the optimized calcified plaque area and the number of second pixel points corresponding to each thickness of the optimized calcified plaque area;
Determining the calcification angle based on the first scan coordinates and the last scan coordinates;
determining the calcification depth based on the first number of pixels;
determining the calcification thickness based on the second number of pixels;
determining the frame quantity and the frame interval of the intracranial artery image corresponding to each plaque in the optimized calcified plaque area;
the length of the calcification is determined based on the number of frames and the frame interval.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the training steps of the segmentation model comprise the following steps:
determining an initial segmentation model, wherein the initial segmentation model is of a multi-layer structure, and the multi-layer structure of the initial segmentation model is different in corresponding characteristic scale;
acquiring a sample intracranial artery image and a label mask of the sample intracranial artery image;
inputting the sample intracranial artery image into an initial segmentation model to obtain each layer of image characteristics output by the initial segmentation model;
downsampling the label masks to feature scales corresponding to the image features of each layer one by one to obtain downsampled features of each label mask;
and determining total loss based on the difference between the downsampling characteristics of each label mask and the image characteristics of each layer, and carrying out parameter iteration on the initial segmentation model based on the total loss to obtain the segmentation model.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the encoder of the initial segmentation model comprises an attention mechanism.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the quantitative analysis is performed based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result, and the method comprises the following steps:
obtaining a calcification score and a calcification type based on the calcification angle, the calcification depth, the calcification thickness and the calcification length;
and obtaining the quantitative analysis result based on the calcium score and the calcification type.
According to the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image, the fiber plaque area and the calcified plaque area are input into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area output by the circulating neural network, and then the method further comprises the following steps:
sequentially performing expansion corrosion and cavity removal on the optimized fiber plaque area to obtain a treated fiber plaque area;
And sequentially performing expansion corrosion and cavity removal on the optimized calcified plaque area to obtain a treated calcified plaque area.
The invention also provides a device for detecting and quantitatively analyzing calcified plaque based on intracranial artery images, which comprises:
an acquisition unit for acquiring an intracranial artery image;
the input unit is used for inputting the intracranial artery image into the segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model;
the optimizing unit is used for inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network;
a calcification parameter determining unit for determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region;
and the quantitative analysis unit is used for carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor is used for realizing the method for detecting and quantitatively analyzing calcified plaque based on intracranial artery images when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method of calcified plaque detection and quantitative analysis based on intracranial artery images as described in any of the foregoing.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a method of calcified plaque detection and quantitative analysis based on intracranial arterial images as described in any of the foregoing.
The method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image further considers the problem of unclear calcified plaque boundary caused by image imaging quality, pertinently improves a segmentation model, enables the optimized segmentation model to generate a more accurate calcified plaque mask, improves the accuracy of a measurement result, increases the credibility of the whole intelligent detection and quantitative analysis in clinical application, and simply scores the plaque by using relevant standards after obtaining the position and basic parameters of the calcified plaque in the prior art, thereby obtaining the risk degree of the calcified plaque. The method is characterized in that a more comprehensive calcified plaque analysis mode is provided based on the clinical requirements, namely quantitative analysis is carried out based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain quantitative analysis results, and the quantitative analysis results comprise preoperative calcification detection and property evaluation and postoperative restenosis risk prediction, so that the analysis results have more reference value for experts, and the accuracy and reliability of calcified plaque quantitative analysis are improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for detection and quantitative analysis of calcified plaque based on intracranial artery images provided by the invention;
FIG. 2 is a schematic diagram of a segmentation module framework for fusing continuous frame information provided by the present invention;
FIG. 3 is a flow chart of step 140 provided by the present invention;
FIG. 4 is a schematic diagram of a model-assisted annotation mode provided by the present invention;
FIG. 5 is a schematic flow chart of the quantitative analysis provided by the invention;
FIG. 6 is a schematic diagram of the device for detecting and quantitatively analyzing calcified plaque based on intracranial artery images;
fig. 7 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The terms first, second and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate, such that embodiments of the present application may be capable of being practiced in sequences other than those illustrated and described herein, and that "first," "second," etc. are typically of the same type.
In the related art, calcified plaque is a metabolite in biological tissues, and refers to a region where calcium salt is deposited in the atherosclerotic plaque tissues, and the calcified plaque is usually detected and displayed in medical imaging by an imaging technology such as X-ray, CT scan, nuclear magnetic resonance, and the like. Atherosclerotic calcified plaque can lead to arterial lumen stenosis and hemodynamic disturbances, potentially leading to cardiovascular disease such as angina, myocardial infarction, stroke, etc. At the same time, atherosclerotic calcified plaque is also associated with restenosis within the stent. On the other hand, compared with the aorta, carotid artery and the like, the research on intracranial atherosclerosis calcified plaque has more direct nervous system health association, can deeply discuss the mechanism of cerebrovascular diseases, guides clinical treatment, and is expected to provide more effective methods and strategies for preventing and treating nervous system diseases.
For an expert, calcified plaques may be used to predict the risk of occurrence of a cerebrovascular event. By analyzing the number, distribution and density of calcified plaques, the likelihood of a patient suffering from a cerebrovascular event can be assessed, thereby taking appropriate precautions. The presence and extent of calcified plaque may affect the choice of surgical procedure, such as analysis of carotid plaque, and if plaque calcification is of a higher hardness and greater area may affect stent placement, the operator may choose carotid artery dissection rather than carotid artery stent implantation. In addition, different parameters of calcified plaque (angle, thickness, depth, etc.) can affect stent placement, stent expansion, and long-term prognosis. Of course, since different parameters have an effect on stent placement and expansion, there is also some suggestion for stent design. However, in clinical applications, it is often necessary to rely on an expert to manually find calcified regions from a large number of continuous images of intracranial artery OCT and manually measure important parameters using auxiliary tools, which can create significant time and labor costs depending on the length of OCT images generated by a single pullback and the number of calcified plaques actually present. Meanwhile, the accuracy of parameter measurement and plaque characterization depends on the expertise of an expert.
Other existing intelligent detection and analysis techniques for vascular calcified plaque generally accomplish the following tasks:
1) Detection of vascular calcified plaque. The detection process is usually implemented based on a segmentation technique, i.e. each pixel of the original OCT image is classified, the probability that it belongs to the background, calcified plaque and other targets is obtained, and finally the complete calcified plaque morphology is restored;
2) Scoring of vascular calcified plaque. After the calcified plaque is detected, the form of the calcified plaque can be directly measured according to the segmentation result to obtain a series of parameter indexes (angle, thickness, depth and the like), and further the calcified plaque is quantitatively scored.
In the related art, the method mainly focuses on the parts such as coronary arteries and carotid arteries, and the data used are intravascular OCT images or heart-sweeping CT images of the parts, but the search work for intracranial artery OCT is not seen. The environment of the intracranial artery is filled with cerebrospinal fluid, so that the blood vessel is extremely small, the structure is free of an external elastic membrane, and the muscle layer is thin, so that although the blood vessel OCT is used as the blood vessel OCT, the imaging, pathology and treatment of the intracranial artery OCT have certain difference with blood vessels at other parts.
Based on the above-mentioned problems, the present invention provides a method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image, fig. 1 is a flow chart of the method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image provided by the present invention, as shown in fig. 1, the method comprises:
Step 110, an intracranial artery image is acquired.
Specifically, an intracranial artery image may be obtained, where the intracranial artery image is an image that needs to be subjected to calcified plaque detection and quantitative analysis, and the intracranial artery image may be an intracranial artery OCT image, where the intracranial artery image may be acquired in advance by an image acquisition device, may be acquired in real time, may be acquired by capturing an image in real time, or may be downloaded or scanned through the internet, and the embodiment of the present invention is not limited in particular.
Here, the intracranial artery image contains a plurality of successive frame images, each frame image representing a cross section of a blood vessel at a different position, so that adjacent frame images emphasize a spatial interval of the illustrated cross section in a three-dimensional entity, and the spatial interval of adjacent frame images is constant.
And 120, inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model.
Specifically, after acquiring the intracranial artery image, the intracranial artery image may be input into the segmentation model, resulting in a fibrous plaque region and a calcified plaque region output by the segmentation model.
Here, the segmentation model may be a U-Net model or the like, which is not particularly limited in the embodiment of the present invention.
The fiber plaque is a rich and uniform high-signal area, has the characteristics of high reflection and low attenuation, has higher degree of distinction from surrounding areas, and has lower segmentation difficulty; in contrast, calcified plaque has low reflection and low attenuation characteristics, is unevenly distributed, and has boundary definition and imaging quality which are closely related, so that the calcified plaque has higher segmentation difficulty.
In the above-mentioned segmentation task, the purpose of segmenting the fibrous plaque is to determine the lumen wall area, although the existing method works on the lumen wall and mostly adopts the direct segmentation mode, because of the particularity of the intravascular OCT acquisition mode, the imaging of the lumen wall is incomplete, and there is a blank area of the lumen wall. On the other hand, in the case of lumen effusion without drainage, abnormal texture of the lumen in OCT imaging also affects the recognition of this region by the segmentation model.
In contrast, the imaging of the fiber plaque is more stable, the lumen contour can be accurately revealed, and the blank area can be more conveniently complemented by using the existing contour information. The second purpose of the calcified plaque segmentation is positioning and measuring respectively, the problem of unclear calcified plaque boundaries caused by image imaging quality is further considered, and the segmentation model is improved pertinently, so that the optimized segmentation model can generate a more accurate calcified plaque mask, the accuracy of a measurement result is improved, and the credibility of the whole intelligent system in clinical application is improved.
In the embodiment of the invention, the following modules are introduced on the basis of a segmentation model (U-Net model):
a) Attention mechanisms. Although calcified plaque has a clear outline under normal radiography, there is also a situation that the boundary is not clear, and in addition, part of calcified plaque presents a smaller form, and is easily ignored by the model when the texture is not obvious. Aiming at the problems, the embodiment of the invention adds the attention module in the segmentation model, and the module can strengthen the region which is worth focusing through a mask in the training process, so that the interference of the irrelevant region to the region of interest in the segmentation process is reduced.
b) And (5) multi-scale supervision. In the process of data labeling, because labeling personnel cannot perfectly outline a target area, inherent errors necessarily exist in the label. The multi-scale supervision is significant in that the error of manual labeling on the outline boundary is relieved by downsampling different scales of the supervision mask and the prediction mask, so that the interference of the pixels in the irrelevant area on model training is reduced.
And 130, inputting the fibrous plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fibrous plaque area and an optimized calcified plaque area which are output by the circulating neural network.
Specifically, fig. 2 is a schematic diagram of a segmentation module framework for fusing continuous frame information, as shown in fig. 2, after obtaining a fibrous plaque area and a calcified plaque area, the fibrous plaque area and the calcified plaque area in the continuous frame may be input into a recurrent neural network (Recurrent Neural Network, RNN) to obtain an optimized fibrous plaque area and an optimized calcified plaque area output by the recurrent neural network.
Therefore, the optimized fiber plaque area and the optimized calcified plaque area of a plurality of continuous frames which are input simultaneously are further input into the cyclic neural network, and the network can effectively fuse time sequence information contained between the continuous frames, so that the segmentation result of each frame is not obtained only through single-frame image information.
Here, the recurrent neural network may be LSTM (Long Short-Term Memory network), bi-RNN (Bidirectional Recurrent Neural Network, bi-directional recurrent neural network), or the like, which is not particularly limited in the embodiment of the present invention.
Step 140, determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region.
Specifically, after the optimized fibrous plaque region and the optimized calcified plaque region are obtained, the calcification angle, the calcification depth, the calcification thickness, and the calcification length of the calcified plaque may be determined based on the optimized fibrous plaque region and the optimized calcified plaque region.
Here, the calcification angle of the calcified plaque refers to the maximum expansion angle of the calcified plaque with respect to the center of the lumen in the cross-sectional direction.
The depth of calcification refers to the location of calcified plaque within the vessel wall, typically expressed in terms of distance from the luminal surface of the vessel.
Calcification thickness refers to the maximum value of the thickness of the calcified plaque at each position along the direction perpendicular to the wall of the lumen.
The length of calcification refers to the length of a calcified plaque within a blood vessel, and is generally expressed as the distance the calcified plaque extends in the direction of the blood vessel.
And step 150, quantitatively analyzing based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
Specifically, after the calcification angle, the calcification depth, the calcification thickness, and the calcification length are determined, quantitative analysis may be performed based on the calcification angle, the calcification depth, the calcification thickness, and the calcification length, resulting in quantitative analysis results.
Here, the calcification score and the calcification type may be obtained based on the calcification angle, the calcification depth, the calcification thickness, and the calcification length, and the quantitative analysis result may be obtained based on the calcification score and the calcification type.
That is, the parameters such as the calcification length, the calcification angle, the calcification thickness and the like are classified in a plurality of sections and scored, and if the score is higher than a certain threshold value, the risk of implanting the stent is higher. Although calcium scores can actually provide effective suggestions for clinical diagnosis, clinical requirements are not limited to the calcium scores, and the embodiment of the invention provides a more comprehensive analysis method based on the clinical requirements.
In particular, embodiments of the present invention are directed to providing guidance for two phases in clinical applications by analyzing calcified plaques: preoperative calcification detection and property assessment and postoperative restenosis risk prediction.
The method provided by the embodiment of the invention further considers the problem of unclear calcified plaque boundaries caused by image imaging quality, and pertinently improves the segmentation model, so that the optimized segmentation model can generate a more accurate calcified plaque mask, thereby improving the accuracy of a measurement result, increasing the credibility of the whole intelligent detection and quantitative analysis in clinical application, and the prior art only simply scores the calcified plaque by using relevant standards after obtaining the position and basic parameters of the calcified plaque, thereby obtaining the risk degree of the calcified plaque. The method is characterized in that a more comprehensive calcified plaque analysis mode is provided based on the clinical requirements, namely quantitative analysis is carried out based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain quantitative analysis results, and the quantitative analysis results comprise preoperative calcification detection and property evaluation and postoperative restenosis risk prediction, so that the analysis results have more reference value for experts, and the accuracy and reliability of calcified plaque quantitative analysis are improved.
Based on the above embodiment, fig. 3 is a schematic flow chart of step 140 provided in the present invention, as shown in fig. 3, step 140 includes:
step 141, extracting a lumen wall from the optimized fibrous plaque region;
step 142, determining a blank area lumen wall of the lumen wall, and determining a lumen center based on the lumen wall and the blank area lumen wall;
step 143, determining the calcification angle, the calcification depth, the calcification thickness and the calcification length based on the lumen center and the optimized calcified plaque region.
Specifically, the lumen wall can be extracted from the optimized fibrous plaque region, and the lumen wall can be obtained as long as the inner wall of the optimized fibrous region remains, since the inner wall of the optimized fibrous region substantially coincides with the lumen wall. The end point of the fiber boundary near the blank area is specially treated, and the tangential point of the optimized fiber area relative to the probe center can be taken as an end point by taking the probe center as a reference.
Because of the blank area inherently present in the intracranial arterial OCT image, there is no image information available in this area, and therefore the lumen wall cannot be directly acquired, and only the area can be simulated from the two endpoints obtained in the previous step. The theoretical basis of simulation generation is that the vessel wall presents regular circular arcs, so that the arc can be drawn by taking the center of the probe as the center of a circle and taking two end points as the starting points, so as to simulate the complete lumen wall, namely the lumen wall in a blank area.
After obtaining the lumen wall and the empty region lumen wall, a lumen center may be determined based on the lumen wall and the empty region lumen wall. The position of the lumen center is well determined, and only the average value of the mask pixel point coordinates of all lumen walls obtained by the lumen walls and the lumen walls in the blank area is needed to be calculated, so that the preparation is made for the accurate parameter measurement later, and the expert takes the lumen center as a reference point when manually measuring the parameters in clinical diagnosis.
Finally, after determining the lumen center, the calcification angle, calcification depth, calcification thickness, and calcification length may be determined based on the lumen center and optimizing the calcified plaque region.
Based on the above embodiment, step 143 includes:
step 1431, scanning the optimized calcified plaque region by taking the center of the lumen as a polar coordinate center, obtaining a first scanning coordinate and a last scanning coordinate of the optimized calcified plaque region, and obtaining the number of first pixels corresponding to each depth of the optimized calcified plaque region and the number of second pixels corresponding to each thickness of the optimized calcified plaque region;
step 1432, determining the calcification angle based on the first scan coordinates and the last scan coordinates;
Step 1433, determining the calcification depth based on the first number of pixels;
step 1434, determining the calcification thickness based on the second number of pixels;
step 1435, determining the frame number and frame interval of the intracranial artery image corresponding to each plaque in the optimized calcified plaque area;
step 1436, determining the length of the calcification based on the number of frames and the frame interval.
Specifically, in order to facilitate parameter measurement, the embodiment of the invention performs polar coordinate conversion on the original image, that is, expands the image along the circumferential direction by taking the center of the lumen as the center of the circle.
That is, the original image can be unfolded by taking the center of the lumen as the center of the polar coordinates, and the optimized calcified plaque region is scanned in the unfolded image to obtain the first scanning coordinate y of the optimized calcified plaque region 1 And last scan coordinate y 2 And obtaining the first pixel point number n corresponding to each depth of the optimized calcified plaque region d A second number of pixels n corresponding to each thickness of the optimized calcified plaque region t 。
The calcification angle may be determined based on the first scan coordinates and the last scan coordinates as follows:
wherein arc represents the angle of calcification, y 1 Representing the first scan coordinates, y 2 And the last scanning coordinate is represented, and l represents the total number of pixels of the image along the circumferential direction under the polar coordinate.
The depth of calcification may be determined based on the first number of pixels, as follows:
depth＝n d ×r xy
wherein depth represents the calcification depth, n d Representing the number of first pixel points, r xy The voxel size corresponding to the pixel point is indicated.
The calcification thickness may be determined based on the second number of pixels, as follows:
thickness＝n t ×r xy
wherein, the thickness of calcification is represented by thickness, n t Representing the number of second pixel points, r xy The voxel size corresponding to the pixel point is indicated.
Strictly speaking, the calcified regions in each frame of intracranial artery image are multiple sections of the complete calcified plaque. Therefore, after the calcification position and the morphological parameters in each frame of image are obtained, all calcified plaques need to be distributed and reconstructed in a three-dimensional space by combining adjacent frames.
Two plaque sections with overlapping areas in two adjacent frames are known to belong to the same calcified plaque in the three-dimensional structure, and the calcified plaque sections in all frames can be numbered based on the principle, so that the effect of three-dimensional reconstruction is realized.
In this process, the number of frames n of the intracranial artery image corresponding to each plaque in the calcified plaque region can be determined l And frame interval r z 。
After determining the number of frames and the frame interval, the length of the calcification may be determined based on the number of frames and the frame interval, as follows:
length＝n l ×r z
wherein length represents the length of calcification, n l Representing the number of frames, r z Representing the frame interval.
Based on the above embodiment, the training step of the segmentation model includes:
step 210, determining an initial segmentation model, wherein the initial segmentation model is of a multi-layer structure, and the multi-layer structure of the initial segmentation model is different in corresponding feature scale;
step 220, acquiring a sample intracranial artery image and a label mask of the sample intracranial artery image;
step 230, inputting the sample intracranial artery image into an initial segmentation model to obtain each layer of image characteristics output by the initial segmentation model;
step 240, downsampling the label mask to a feature scale corresponding to each layer of image features one by one to obtain downsampled features of each label mask;
step 250, determining a total loss based on the difference between the downsampled feature of each label mask and the image feature of each layer, and performing parameter iteration on the initial segmentation model based on the total loss to obtain the segmentation model.
Specifically, in order to obtain a segmented model better, it is necessary to acquire the segmented model by:
the initial segmentation model may be predetermined, where the initial segmentation model is a multi-layer structure, and the multi-layer structure of the initial segmentation model corresponds to different feature scales, and the initial segmentation model may be a U-Net model, which is not specifically limited in the embodiment of the present invention.
Sample intracranial artery images can also be pre-collected, as well as a label mask for the sample intracranial artery images.
In order to train the segmentation model used in the embodiments of the present invention and verify performance, it is necessary to acquire and label enough images of the patient's intracranial artery OCT.
In the image acquisition stage, the embodiment of the invention acquires the dicom of 48 patients in total, and selects 779 total images of 31 image stacks containing calcified areas from the dicom for training and testing the model, and in order to ensure the segmentation performance of the model, the image stacks also comprise healthy areas without calcified plaques.
In the image labeling stage, considering the difficulty of medical image interpretation, pure human labeling has considerable human and time costs, and in addition, the segmentation quality of the final labeling result can be influenced by long-time repeated operation.
FIG. 4 is a schematic diagram of the auxiliary labeling mode of the model provided by the invention, as shown in FIG. 4, which can be used because of the characteristics of two segmentation targets, and fiber regions exist in each frame, which have long and thin shapes and different shapes, and are easy to recognize but take longer time for labeling; the difficulty in labeling calcified plaque is mainly identification and interpretation, and once the calcified plaque is characterized by expert, the outline of the calcified plaque is relatively easy to manually outline, and the time is short. On the other hand, the segmentation model training of the fiber region can be completed by only a small number of images. Therefore, the embodiment of the invention is expected to realize the labeling work of most fiber areas by using the model, and can also give a few typical calcified reference masks, so that an expert is more focused on interpretation rather than repeated mechanical work in the labeling process.
Specifically, 100 images can be selected from the OCT images of all the patient's intracranial arteries (not limited to the 31 stacks selected), which need to be guaranteed not to have spatial adjacency while possessing a distinct, easily identifiable outline calcified plaque; then, the images are delivered to an expert for accurate manual labeling, and a rough auxiliary segmentation model is obtained by training the labeled images; and finally, inputting all the images to be marked into the model to obtain a rough segmentation result, and transmitting the rough segmentation result to an expert for correction to finally finish the marking process. It is worth mentioning that the fibrous region in the result has been better segmented, but the calcified plaque segmentation effect is not ideal, so that the expert only needs to concentrate on the interpretation and correction of the calcified plaque in the image, and only needs to simply optimize the fibrous region, thereby greatly saving the labeling time and having higher labeling result compared with the quality of pure manual.
After the initial segmentation model is obtained, the initial segmentation model can be trained by applying a sample intracranial artery image collected in advance and a label mask of the sample intracranial artery image:
firstly, a sample intracranial artery image is input into an initial segmentation model, and a convolution layer can be externally connected behind each layer of the initial segmentation model to obtain image characteristics of each layer output by the initial segmentation model.
Then, the label mask can be downsampled to the feature scale corresponding to the image features of each layer one by one, so as to obtain downsampled features of each label mask.
After the label mask downsampling features and the image features of each layer are obtained, a total loss can be determined based on the difference between the label mask downsampling features and the image features of each layer, with the following formula:
wherein L is multi-scal Representing the total loss, i representing the layer index of the initial segmentation model, N representing the total layer number of the initial segmentation model, f i (x) Output image features representing the ith layer of the initial segmentation model, f 0 (x) For final output image features of the initial segmentation model, DOWN i (y) means i downsampling the label mask.
It can be appreciated that the greater the difference between the downsampling features of each label mask and the image features of each layer, the greater the total loss; the smaller the difference between the downsampled feature of each label mask and the image feature of each layer, the smaller the total loss.
The segmentation model after parameter iteration has the same model structure as the initial segmentation model, namely the segmentation model is also a multi-layer structure, and the feature scales corresponding to the multi-layer structure of the segmentation model are different.
Here, the segmentation loss L seg A cross entropy loss function (Cross Entropy Loss Function) may be used, or a mean square error loss function (Mean Squared Error, MSE) may be used, as embodiments of the present invention are not particularly limited.
Based on the above embodiment, the encoder of the initial segmentation model includes an attention mechanism.
In particular, it is considered that although calcified plaque has a clear outline in the case of normal radiography, there is also a case where the boundary is unclear, and furthermore, a part of calcified plaque exhibits a smaller morphology, which is easily ignored by the model when the texture is not clear
In view of the above problems, in the embodiment of the present invention, an attention mechanism is added to an initial segmentation model, where the attention mechanism can strengthen a region of interest through a mask during a training process, so as to reduce interference of an irrelevant region on the region of interest during segmentation.
Based on the above embodiment, fig. 5 is a schematic flow chart of the quantitative analysis provided by the present invention, as shown in fig. 5, step 150 includes:
Step 151, obtaining a calcification score and a calcification type based on the calcification angle, the calcification depth, the calcification thickness and the calcification length;
and step 152, obtaining the quantitative analysis result based on the calcification score and the calcification type.
Specifically, the calcium score is a common index for predicting the risk of abnormal expansion of the stent after implantation in the operation, and table 1 shows a general calculation method of the calcium score.
TABLE 1 calculation of Calif. reference Table
/>
When the integral reaches 4 minutes, there is a greater risk of the problem of poor stent expansion, while when the integral is below 4 minutes, the risk of poor stent expansion is lower. This result can assist the expert in formulating an appropriate surgical plan, selecting a stent with greater support as appropriate, or combining post-dilation to reduce the incidence of restenosis.
The calcification type assessment is derived from multi-dimensional multi-parameter detection of calcification properties before operation, and can provide a large reference for clinical stent treatment and prediction of postoperative restenosis risk. Calcified plaques can be classified into giant calcification, punctiform calcification and microcalcification by taking the length and angle of calcification as the standard, and the specific classification modes are shown in Table 2.
TABLE 2 type evaluation reference Table for calcified plaque
Wherein, the huge calcification is a calcification type needing to pay attention to, the huge calcification has a larger association with vascular restenosis, and if the huge calcified plaque is found, the postoperative follow-up strategy needs to be adjusted according to the situation. On the other hand, the calcification type can also assist in predicting the risk of stent poor expansion, i.e. for plaques with a calcification score of less than 4, if a large calcification is determined, the negative impact on the stent implantation procedure still needs to be considered.
It should be noted that, for the calcification type, if the measured parameter features do not satisfy any of the three calcifications, the calcification is marked as other calcifications and is submitted to an expert for secondary interpretation.
Clinical questions may be entered. The clinical problems are different in the use modes of indexes, and the diagnosis and treatment suggestions required to be obtained are different, so that a user is required to select specific clinical problems according to the current actual situation before further analysis. Specifically, when preparing a stent implantation surgical plan, the expert is concerned about whether a problem of poor stent expansion occurs after implantation of the stent; while in the preparation of a post-operative follow-up regimen, the expert is concerned with whether there is a risk of restenosis in the vessel.
Then, interpretation can be performed based on the index and diagnosis and treatment advice can be given. When predicting the expansion condition of the stent after the stent is implanted in the operation, if the calcium score is equal to 4 minutes, directly judging that the risk of poor expansion of the large stent exists; if the score of calcification is less than 4 points, further analysis in combination with the type of calcification is also required, and even if the score of calcification is less than 4 points, the conclusion of poor stent expansion needs to be output. In predicting the risk of restenosis after surgery, only the type of calcified plaque in a blood vessel segment needs to be judged, and if the blood vessel segment has huge calcification, the risk of restenosis is also large, and special attention of an expert is needed.
Based on the above embodiment, step 130 further includes:
step 131, sequentially performing expansion corrosion and cavity removal on the optimized fiber plaque area to obtain a treated fiber plaque area;
and step 132, sequentially performing expansion corrosion and cavity removal on the optimized calcified plaque region to obtain a treated calcified plaque region.
Specifically, in order to further optimize the mask, after obtaining the optimized fibrous plaque area and the optimized calcified plaque area, a post-processing operation is further performed, mainly by two means: expansion corrosion and void removal.
The expansion corrosion can effectively smooth the boundary, and eliminate abnormal fracture and partial small cavities in the over-narrow area; the cavity is removed, so that the problem that cavities appear in the mask caused by uneven textures is solved, and the integrity of the plaque is ensured.
According to the method provided by the embodiment of the invention, the optimized fibrous plaque area is sequentially subjected to the expansion corrosion and the cavity removal to obtain the treated fibrous plaque area, the optimized calcified plaque area is sequentially subjected to the expansion corrosion and the cavity removal to obtain the treated calcified plaque area, and the optimization is performed on the basis of the treated fibrous plaque area and the treated calcified plaque area, so that the segmentation accuracy and the segmentation reliability of the fibrous plaque area and the calcified plaque area are improved.
The device for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image provided by the invention is described below, and the device for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image and the method for detecting and quantitatively analyzing the calcified plaque based on the intracranial artery image described above can be correspondingly referred to each other.
Based on any of the above embodiments, the present invention provides an apparatus for detecting and quantitatively analyzing calcified plaque based on intracranial artery image, and fig. 6 is a schematic structural diagram of the apparatus for detecting and quantitatively analyzing calcified plaque based on intracranial artery image, as shown in fig. 6, the apparatus includes:
An acquisition unit 610 for acquiring an intracranial artery image;
an input unit 620, configured to input the intracranial artery image into a segmentation model, and obtain a fibrous plaque region and a calcified plaque region output by the segmentation model;
an optimizing unit 630, configured to input the fibrous plaque area and the calcified plaque area into a recurrent neural network, and obtain an optimized fibrous plaque area and an optimized calcified plaque area output by the recurrent neural network;
a calcification parameter determining unit 640 for determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region;
and a quantitative analysis unit 650 for performing quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length, and obtaining a quantitative analysis result.
The device provided by the embodiment of the invention further considers the problem of unclear calcified plaque boundaries caused by image imaging quality, and pertinently improves the segmentation model, so that the optimized segmentation model can generate a more accurate calcified plaque mask, thereby improving the accuracy of a measurement result, increasing the credibility of the whole intelligent detection and quantitative analysis in clinical application, and the prior art simply scores the calcified plaque by using relevant standards after obtaining the position and basic parameters of the calcified plaque, thereby obtaining the risk degree of the calcified plaque. The method is characterized in that a more comprehensive calcified plaque analysis mode is provided based on the clinical requirements, namely quantitative analysis is carried out based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain quantitative analysis results, and the quantitative analysis results comprise preoperative calcification detection and property evaluation and postoperative restenosis risk prediction, so that the analysis results have more reference value for experts, and the accuracy and reliability of calcified plaque quantitative analysis are improved.
Based on any of the above embodiments, the determining calcification parameter unit 640 specifically includes:
an extraction lumen wall unit for extracting a lumen wall from the optimized fibrous plaque region;
a void area lumen wall unit for determining a void area lumen wall of the lumen wall and determining a lumen center based on the lumen wall and the void area lumen wall;
a calcification parameter determination subunit for determining the calcification angle, the calcification depth, the calcification thickness and the calcification length based on the lumen center and the optimized calcified plaque region.
Based on any of the above embodiments, the determining calcification parameter subunit is specifically configured to:
scanning the optimized calcified plaque area by taking the center of the lumen as a polar coordinate center to obtain a first scanning coordinate and a last scanning coordinate of the optimized calcified plaque area, and obtaining the number of first pixel points corresponding to each depth of the optimized calcified plaque area and the number of second pixel points corresponding to each thickness of the optimized calcified plaque area;
determining the calcification angle based on the first scan coordinates and the last scan coordinates;
determining the calcification depth based on the first number of pixels;
Determining the calcification thickness based on the second number of pixels;
determining the frame quantity and the frame interval of the intracranial artery image corresponding to each plaque in the optimized calcified plaque area;
the length of the calcification is determined based on the number of frames and the frame interval.
Based on any of the above embodiments, the training step of the segmentation model includes:
determining an initial segmentation model, wherein the initial segmentation model is of a multi-layer structure, and the multi-layer structure of the initial segmentation model is different in corresponding characteristic scale;
acquiring a sample intracranial artery image and a label mask of the sample intracranial artery image;
inputting the sample intracranial artery image into an initial segmentation model to obtain each layer of image characteristics output by the initial segmentation model;
downsampling the label masks to feature scales corresponding to the image features of each layer one by one to obtain downsampled features of each label mask;
and determining total loss based on the difference between the downsampling characteristics of each label mask and the image characteristics of each layer, and carrying out parameter iteration on the initial segmentation model based on the total loss to obtain the segmentation model.
Based on any of the above embodiments, an attention mechanism is included in the encoder of the initial segmentation model.
Based on any of the above embodiments, the quantitative analysis unit 650 specifically functions to:
obtaining a calcification score and a calcification type based on the calcification angle, the calcification depth, the calcification thickness and the calcification length;
and obtaining the quantitative analysis result based on the calcium score and the calcification type.
Based on any of the foregoing embodiments, the apparatus further includes a processing unit, where the processing unit is specifically configured to:
sequentially performing expansion corrosion and cavity removal on the optimized fiber plaque area to obtain a treated fiber plaque area;
and sequentially performing expansion corrosion and cavity removal on the optimized calcified plaque area to obtain a treated calcified plaque area.
Fig. 7 illustrates a physical schematic diagram of an electronic device, as shown in fig. 7, which may include: processor 710, communication interface (Communications Interface) 720, memory 730, and communication bus 740, wherein processor 710, communication interface 720, memory 730 communicate with each other via communication bus 740. Processor 710 may invoke logic instructions in memory 730 to perform a method for detection and quantitative analysis of calcified plaque based on intracranial arterial images, the method comprising: acquiring an intracranial artery image; inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model; inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network; determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region; and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
Further, the logic instructions in the memory 730 described above may be implemented in the form of software functional units and may be stored in a computer readable storage medium when sold or used as a stand alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing the method for detecting and quantitatively analyzing calcified plaque based on intracranial artery images provided by the above methods, the method comprising: acquiring an intracranial artery image; inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model; inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network; determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region; and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method for intracranial arterial image-based calcified plaque detection and quantitative analysis provided by the above methods, the method comprising: acquiring an intracranial artery image; inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model; inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network; determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region; and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for detection and quantitative analysis of calcified plaque based on intracranial arterial imaging, comprising:
acquiring an intracranial artery image;
inputting the intracranial artery image into a segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model;
inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network;
determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region;
and carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
2. The method of intracranial artery image-based calcified plaque detection and quantification as set forth in claim 1, wherein the determining the calcified plaque's calcification angle, calcification depth, calcification thickness and calcification length based on the optimized fibrous plaque region and the optimized calcified plaque region includes:
Extracting a lumen wall from the optimized fibrous plaque region;
determining a blank area lumen wall of the lumen wall, and determining a lumen center based on the lumen wall and the blank area lumen wall;
the calcification angle, the calcification depth, the calcification thickness and the calcification length are determined based on the lumen center and the optimized calcified plaque region.
3. The method of intracranial artery image-based calcified plaque detection and quantification as recited in claim 2, wherein the determining the calcification angle, the calcification depth, the calcification thickness and the calcification length based on the lumen center and the optimized calcified plaque region comprises:
scanning the optimized calcified plaque area by taking the center of the lumen as a polar coordinate center to obtain a first scanning coordinate and a last scanning coordinate of the optimized calcified plaque area, and obtaining the number of first pixel points corresponding to each depth of the optimized calcified plaque area and the number of second pixel points corresponding to each thickness of the optimized calcified plaque area;
determining the calcification angle based on the first scan coordinates and the last scan coordinates;
Determining the calcification depth based on the first number of pixels;
determining the calcification thickness based on the second number of pixels;
determining the frame quantity and the frame interval of the intracranial artery image corresponding to each plaque in the optimized calcified plaque area;
the length of the calcification is determined based on the number of frames and the frame interval.
4. The method of intracranial artery image-based calcified plaque detection and quantitative analysis of claim 1, wherein the training step of the segmentation model comprises:
determining an initial segmentation model, wherein the initial segmentation model is of a multi-layer structure, and the multi-layer structure of the initial segmentation model is different in corresponding characteristic scale;
acquiring a sample intracranial artery image and a label mask of the sample intracranial artery image;
inputting the sample intracranial artery image into an initial segmentation model to obtain each layer of image characteristics output by the initial segmentation model;
downsampling the label masks to feature scales corresponding to the image features of each layer one by one to obtain downsampled features of each label mask;
and determining total loss based on the difference between the downsampling characteristics of each label mask and the image characteristics of each layer, and carrying out parameter iteration on the initial segmentation model based on the total loss to obtain the segmentation model.
5. The method of intracranial arterial image-based calcified plaque detection and quantitative analysis of claim 4, wherein the initial segmentation model encoder includes a attentional mechanism.
6. The method for detection and quantitative analysis of calcified plaque based on intracranial artery image as recited in any one of claims 1 to 5, wherein the quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length, results in quantitative analysis, comprises:
obtaining a calcification score and a calcification type based on the calcification angle, the calcification depth, the calcification thickness and the calcification length;
and obtaining the quantitative analysis result based on the calcium score and the calcification type.
7. The method of any one of claims 1 to 5, wherein the inputting the fibrous plaque region and the calcified plaque region into a recurrent neural network results in an optimized fibrous plaque region and an optimized calcified plaque region output by the recurrent neural network, and further comprising:
sequentially performing expansion corrosion and cavity removal on the optimized fiber plaque area to obtain a treated fiber plaque area;
And sequentially performing expansion corrosion and cavity removal on the optimized calcified plaque area to obtain a treated calcified plaque area.
8. An apparatus for detection and quantitative analysis of calcified plaque based on intracranial arterial imaging, comprising:
an acquisition unit for acquiring an intracranial artery image;
the input unit is used for inputting the intracranial artery image into the segmentation model to obtain a fibrous plaque area and a calcified plaque area which are output by the segmentation model;
the optimizing unit is used for inputting the fiber plaque area and the calcified plaque area into a circulating neural network to obtain an optimized fiber plaque area and an optimized calcified plaque area which are output by the circulating neural network;
a calcification parameter determining unit for determining a calcification angle, a calcification depth, a calcification thickness and a calcification length of the calcified plaque based on the optimized fibrous plaque region and the optimized calcified plaque region;
and the quantitative analysis unit is used for carrying out quantitative analysis based on the calcification angle, the calcification depth, the calcification thickness and the calcification length to obtain a quantitative analysis result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of intracranial artery image-based calcified plaque detection and quantitative analysis as recited in any one of claims 1 to 7 when the program is executed by the processor.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the method of intracranial artery image-based calcified plaque detection and quantitative analysis as recited in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311477076.9A CN117745641A (en) | 2023-11-07 | 2023-11-07 | Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311477076.9A CN117745641A (en) | 2023-11-07 | 2023-11-07 | Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117745641A true CN117745641A (en) | 2024-03-22 |
Family
ID=90251616
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311477076.9A Pending CN117745641A (en) | 2023-11-07 | 2023-11-07 | Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117745641A (en) |
-
2023
- 2023-11-07 CN CN202311477076.9A patent/CN117745641A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11615894B2 (en) | Diagnostically useful results in real time | |
| US11406337B2 (en) | Calculating a fractional flow reserve | |
| US10470730B2 (en) | Creating a vascular tree model | |
| EP2946321B1 (en) | Diagnostically useful results in real time | |
| CN114424290B (en) | Apparatus and method for providing a longitudinal display of coronary calcium loading | |
| JP2021140820A (en) | Detection and validation of shadows in intravascular images | |
| CN117745641A (en) | Method for detecting and quantitatively analyzing calcified plaque based on intracranial artery image | |
| TWI490790B (en) | Dynamic cardiac imaging analysis and cardiac function assessment system | |
| Trzupek | Semantic interpretation of heart vessel structures based on graph grammars | |
| CN115147320A (en) | Coronary artery CT image subtraction method and device, electronic equipment and storage medium | |
| Kolluru et al. | Yazan Gharaibeh, Pengfei Dong, David Prabhu |
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 |