CN110969633B - Automatic optimal phase identification method for cardiac CT imaging - Google Patents
Automatic optimal phase identification method for cardiac CT imaging Download PDFInfo
- Publication number
- CN110969633B CN110969633B CN201911191975.6A CN201911191975A CN110969633B CN 110969633 B CN110969633 B CN 110969633B CN 201911191975 A CN201911191975 A CN 201911191975A CN 110969633 B CN110969633 B CN 110969633B
- Authority
- CN
- China
- Prior art keywords
- network
- phase
- images
- cardiac
- training
- 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.)
- Active
Links
- 230000000747 cardiac effect Effects 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013170 computed tomography imaging Methods 0.000 title claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 46
- 238000012360 testing method Methods 0.000 claims abstract description 21
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 238000013135 deep learning Methods 0.000 claims abstract description 6
- 230000011218 segmentation Effects 0.000 claims description 22
- 210000003516 pericardium Anatomy 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 11
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 239000013598 vector Substances 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000004807 localization Effects 0.000 claims description 2
- 230000001360 synchronised effect Effects 0.000 claims description 2
- 230000033001 locomotion Effects 0.000 abstract description 10
- 238000002591 computed tomography Methods 0.000 abstract description 4
- 238000010801 machine learning Methods 0.000 abstract description 3
- 230000008859 change Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 210000004351 coronary vessel Anatomy 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 201000000057 Coronary Stenosis Diseases 0.000 description 1
- 206010011089 Coronary artery stenosis Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 230000002308 calcification Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 230000010247 heart contraction Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000036244 malformation Effects 0.000 description 1
- 208000031225 myocardial ischemia Diseases 0.000 description 1
- 230000000241 respiratory effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/30048—Heart; Cardiac
Abstract
The invention discloses an automatic optimal phase identification method for cardiac CT imaging, which comprises the following steps: collecting projection data of N patient cardiac scans, reconstructing N CT images of different phases of each patient, wherein each phase corresponds to a label value; establishing a phase estimation network based on a deep learning network model; inputting the training set and the testing set data into a network model, and training to obtain network parameters of a phase estimation network; CT images of different phases of the same patient are randomly selected from the data set to serve as test images, and the phase corresponding to the optimal label value is determined according to the label value of each CT image. The invention adopts the technology based on machine learning and neural network to construct the phase network model, can quickly find the optimal CT scanning phase, reduces the motion artifact and improves the image quality.
Description
Technical Field
The invention relates to the technical field of medical imaging, in particular to an automatic optimal phase identification method for cardiac CT imaging based on machine learning and neural network technology.
Background
The CT-based human heart and the scanning of coronary arteries thereof are always important detection means in aspects of coronary artery stenosis, coronary calcification score, coronary malformation variation, myocardial ischemia and the like; in view of the specificity of rapid beating of the heart, scanning of the human heart and its coronary arteries requires a high imaging time resolution.
The application of the real-time electrocardiogram relieves the problems to a certain extent, the characteristic that the diastole movement in the heart beating period is relatively gentle is utilized, the data source (as shown in figure 1) in the period is acquired, the three-dimensional reconstruction data of relatively stable heart scanning is acquired, and the imaging quality of CT scanning heart is greatly improved. Since the heart is stationary for only a short period of time throughout the cardiac cycle, the exposure interval for scanning is set according to each cycle of the electrocardiographic signal, and usually two parameters, one is the length of exposure and the other is the center phase of exposure, in order to be able to effectively obtain cardiac images. These two parameters are typically chosen based on empirical values, the length of the exposure is typically the minimum reconstruction interval required for the image plus a redundant time, and the center phase of the exposure is typically 70%. After scanning, the segment of exposure data is utilized to reconstruct an image according to the default exposure center phase, so that a required CT image can be obtained.
The current cardiac scan is typically based on a preselected phase, and then image reconstruction is performed on the cardiac image based on scan data in the vicinity of the corresponding phase. However, this default phase does not necessarily provide the best image quality, and it is often the case that after image reconstruction, if a doctor finds that motion artifacts in an image are serious, the doctor will change the reconstruction phase to acquire images of other scan intervals according to the motion artifact. And then until a more appropriate reconstructed phase is found as the final clinical diagnostic image.
Because of the complexity of motion artifacts, conventional methods do not have a good way to quantitatively identify the magnitude of the motion artifact, and therefore there is no automatic method to automatically identify the optimal phase. In addition, the traditional manual method needs to traverse different phases, so that a lot of reconstruction and processing time is needed, and the efficiency is difficult to effectively improve.
Disclosure of Invention
The technical purpose is that: aiming at the technical problems, the invention provides an automatic optimal phase identification method for cardiac CT imaging, which is based on machine learning and neural network technology, builds a phase network model, can realize rapid finding of an optimal phase, reduces motion pseudo influence and improves image quality.
The technical scheme is as follows: in order to realize the technical scheme, the invention adopts the following technical scheme:
an automatic optimal phase identification method for cardiac CT imaging is characterized by comprising the following steps:
a1, data set preparation: collecting projection data of heart scanning of M patients, reconstructing M CT images with different phases of each patient, wherein each phase corresponds to a label value representing image quality, and randomly dividing the obtained CT image data into a training set, a test set and a verification set, wherein M is larger than or equal to 30, and M is larger than or equal to 20;
a2, designing a neural network: constructing a network model as a phase estimation network, wherein the input of the phase estimation network is three-dimensional volume data obtained by reconstructing N different phase points, and the three-dimensional volume data is output as vectors representing image quality corresponding to each phase;
a3, network training: inputting the training set and the testing set data into a network model, and training to obtain network parameters of a phase estimation network;
a4, optimal phase estimation: and selecting CT images of N phases of the same patient from the data set as test images, inputting the test images into a phase estimation network to obtain the label values of the N CT images and the N phases, and determining the phase corresponding to the optimal label value according to the label value of each CT image.
Preferably, the phase estimation network in the step A2 adopts a fully connected network, or a convlstm network structure, and the network loss function takes the mean square error MSE as an objective function.
Preferably, in the step A4, a phase corresponding to the optimal label value is searched according to the label value of each CT image; or fitting a straight line or curve through the tag value, wherein y=fun (x) is used for representing, x is a measured tag value, y is a phase value, and the optimal center phase is finally obtained through the fitting result; or directly outputting a phase-image quality vector of K points, and obtaining the optimal phase according to the optimal label value.
Preferably, a heart segmentation network is added before the phase estimation network is established in step A2.
Preferably, the step of establishing the cardiac segmentation network in the step A2 is:
b1, constructing a data set of a heart segmentation network: collecting projection data of K patient cardiac scans, reconstructing K CT images with different phases for each patient, and marking a pericardial area in the obtained CT images, wherein K is larger than or equal to 30 and K is larger than or equal to 20;
b2, designing a neural network: constructing a deep learning network structure of U-Net to automatically divide the pericardium;
b3: training a network: training a network, namely inputting training set and test set data into the network, and training to obtain network parameters;
b4: pericardium segmentation: the reconstructed CT images are input into a network to obtain segmented CT images, and the images after pericardium segmentation correspond to different phase labels respectively.
Preferably, in step A2, a heart region-based attention network is added before the phase estimation network is established.
Preferably, the step of establishing the attention network based on the heart region in the step A2 is:
c1, constructing a data set of the attention network based on the heart region: collecting projection data of K patient cardiac scans, reconstructing K CT images with different phases for each patient, marking a pericardial area in the obtained CT images, and performing Gaussian blur on masks of the CT images, wherein K is larger than or equal to 30 and K is larger than or equal to 20;
c2, designing a neural network, namely constructing a deep learning network structure of the U-Net, and positioning the position of the pericardium based on a hetmap regression;
and C3, network training: training a network, namely inputting training set and test set data into the network, and training to obtain network parameters;
c4. cardiac localization: the reconstructed CT image is input into a network to obtain the approximate position of the heart, and a region with a fixed size is cut around the center of the predicted boundary frame to obtain the heart region.
Preferably, in the step A3, the loss function training is optimized through a gradient descent method to construct a network model.
Preferably, the projection data of the cardiac scan is obtained in an axial scan mode, a step scan mode or a helical scan mode synchronized according to the cardiac phase.
The beneficial effects are that: the invention has the following advantages:
(1) According to the automatic phase identification method for cardiac CT imaging, provided by the invention, by reconstructing a few phases and utilizing an artificial intelligence technology, the optimal phase can be estimated through training of a certain amount of cardiac data, so that the optimal phase reconstruction is realized, and the motion artifact is reduced;
(2) On the basis of the prior art, the automatic phase identification method for cardiac CT imaging processes images through an algorithm, improves the quality of cardiac images, has wider practicability compared with the prior art, can solve the problems of motion artifacts and the like which cannot be solved by hardware technology, and can effectively improve the quality of images;
(3) Compared with the traditional method, the method has higher efficiency, does not need to traverse all phase images, and then carries out manual screening; the optimal phase information can be estimated by only automatically processing the images of a plurality of phases; in the training process, the blood vessels in the heart are not required to be manually segmented by a complex segmentation algorithm, and only the whole image quality is required to be scored, so that the marking workload is much smaller, and the calculation cost is lower.
Drawings
FIG. 1 is a schematic illustration of a real-time electrocardiogram-based cardiac scan;
FIG. 2 is a schematic diagram of a VGG+LSTM network;
FIG. 3 is a reconstructed image of a cardiac CT scan taken from several different phases;
FIG. 4 is a schematic diagram of tag values corresponding to different phase images according to the first embodiment;
fig. 5 is a schematic diagram of a structure of automatic phase identification for CT cardiac imaging according to the second embodiment.
Detailed Description
The invention will be further described with reference to the accompanying drawings.
The invention provides an automatic phase identification method for cardiac CT imaging, which comprises the following steps:
data set preparation: collecting projection data of heart scanning of M patients, reconstructing M CT images with different phases of each patient, and randomly dividing the obtained CT image data into a training set, a test set and a verification set;
designing a neural network: each phase corresponds to a label value, and a phase estimation network is established;
training a network: optimizing a loss function training network by a gradient descent method, inputting training set and testing set data into a network model, and training to obtain network parameters;
phase estimation: CT images of different phases of the same patient are randomly selected from the data set to serve as test images, and the phase corresponding to the optimal label value is determined according to the label value of each CT image.
Embodiment one:
the invention relates to an automatic phase identification method for cardiac CT imaging, which is characterized in that the optimal phase estimation is carried out on the whole reconstructed image, and the method comprises the following specific steps:
step A1, data preparation: projection data of 30 cardiac scans of patients are collected, the phases are between 0 and 1, CT images of different phases are reconstructed, and 20 data of different phases are reconstructed for each patient at a step interval of 0.05, as shown in FIG. 3, the reconstructed images of several different phases are selected, and in FIG. 3, the five images are CT scan images of 0.55,0.65,0.75,0.85 and 0.95 phases respectively. The different phase images correspond to their own phase tag values, respectively. Taking the labels set to be linearly distributed as an example, as shown in fig. 4, by observing that the optimal phase is at the position of 0.75, the label value corresponding to the 0.75 phase is set to be 0, and the label values of other phases centered at 0.75 are distributed between-0.5 and 0.5. The above data is randomly divided into training sets, test sets and validation sets.
Step A2, designing a neural network: taking a convlstm type network model as an example, a VGG+LSTM network is used for predicting the optimal center phase, the LSTM is added to a training network for learning information which is dependent on a long term, the input image of the network is regarded as a time sequence, the VGG+LSTM is used, not only can the time sequence relation (the relation between different phases) be obtained, but also the spatial characteristics can be extracted like a convolution layer, and the state-to-state switching is changed into convolution calculation. In the scheme, CT images (for example, 0.55,0.65,0.75,0.85,0.95 phases are selected) of 5 different phase positions of the same patient are randomly selected as input images of a network, the sizes of the input images are 5 multiplied by 1 multiplied by 512, the output is a deviation phase value from an optimal phase, the scheme uses a phase deviation from 0.75 as an output value, 0 represents an optimal center phase, and a network loss function uses a Mean Square Error (MSE) as an objective function.
Step A3, network training: and optimizing the loss function training network by a gradient descent method, inputting the training set and the testing set data into the network, and training to obtain network parameters.
Step A4, phase estimation: and randomly selecting 5 CT images with different phases of the same patient from the data set as test images, predicting 20 label values, and searching for the phase corresponding to the optimal label, or fitting a straight line or curve through the label values, wherein y=fun (x) is used for representing, x is a measured label value, and y is a phase value. And finally obtaining the optimal center phase corresponding to the tag 0 through the fitting result, and reminding an operator of carrying out corresponding change.
From the above steps, there are two implementation manners of this solution, one implementation manner is: the input of the network is three-dimensional volume data obtained by reconstructing N (such as N=5) different phase points, the N volume data are respectively input into the network with the same phase estimation, the N phases can be obtained by outputting an image quality index of the current volume data, and then the change of the image quality along with the phase is estimated according to the N phases to estimate the optimal phase point; in another implementation, N individual data are sequentially input into a convlstm-based network, a sequence of features of an image index and the phases are output, and then the features are input into a subsequent network to give a curve of the phase change with time, and an optimal phase point is estimated.
Embodiment two:
a main difference from the embodiment is that a heart segmentation network is added before the phase estimation network is constructed in step 2, or a heart region-based attention network is added, and the advantage of adopting this structure is that the network can pay more attention to the image quality of the heart region and is not easily interfered by the images of other regions. Shown in fig. 5. The heart segmentation network and the classification network are trained separately. The specific process of constructing the heart segmentation network is as follows:
step B1, segment CNN data preparation: projection data of 30 patient cardiac scans are collected, 30000 CT images of different phases (such as 55%, 65%, 75%, 85%, 95%) are reconstructed, and pericardial areas are manually marked; if the attention network is adopted, the same mask can be used for training after Gaussian blur.
Step B2: designing a neural network: taking a deep learning network structure of U-Net as an example to carry out automatic pericardium segmentation;
step B3: training a network: optimizing a loss function training network by a gradient descent method, inputting training set and test set data into the network, and training to obtain network parameters;
step B4: pericardium segmentation: inputting the reconstructed CT image into a network to obtain a segmented CT image; the images after pericardium segmentation correspond to different phase labels respectively,
classification CNN data preparation: the images after pericardium segmentation correspond to different phase labels, respectively, and are trained and tested by the method mentioned in scheme one.
The present embodiment further considers that there are differences in cardiac CT reconstructed images of different phases due to respiratory motion and the like, and that it is difficult for other tissue structures to be consistent, as well as differences in cardiac regions. A heart segmentation network is constructed, and the heart region is segmented, so that the optimal heart phase can be judged more accurately. The heart region segmentation method can be a conventional method in addition to the neural network method.
The foregoing is only a preferred embodiment of the invention, it being noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the invention.
Claims (7)
1. An automatic optimal phase identification method for cardiac CT imaging is characterized by comprising the following steps:
a1, data set preparation: collecting projection data of heart scanning of M patients, reconstructing M CT images with different phases of each patient, wherein each phase corresponds to a label value representing image quality, and randomly dividing the obtained CT image data into a training set, a test set and a verification set, wherein M is larger than or equal to 30, and M is larger than or equal to 20;
a2, designing a neural network: constructing a network model as a phase estimation network, wherein the input of the phase estimation network is three-dimensional volume data obtained by reconstructing N different phase points, and the three-dimensional volume data is output as vectors representing image quality corresponding to the N phases;
a3, network training: inputting the training set and the testing set data into a network model, and training to obtain network parameters of a phase estimation network;
a4, optimal phase estimation: selecting CT images of N phases of the same patient from the data set as test images, inputting the test images into a phase estimation network to obtain the label values of the N CT images and the N phases, and determining the phase corresponding to the optimal label value according to the label value of each CT image;
in the step A2, a heart segmentation network is added before the phase estimation network is built, and the building steps of the heart segmentation network are as follows:
b1, constructing a data set of a heart segmentation network: collecting projection data of K patient cardiac scans, reconstructing K CT images with different phases for each patient, and marking a pericardial area in the obtained CT images, wherein K is larger than or equal to 30 and K is larger than or equal to 20;
b2, designing a neural network: constructing a deep learning network structure of U-Net to automatically divide the pericardium;
b3: training a network: training a network, namely inputting training set and test set data into the network, and training to obtain network parameters of a heart segmentation network;
b4: pericardium segmentation: the reconstructed CT images are input into a network to obtain segmented CT images, and the images after pericardium segmentation correspond to different phase labels respectively.
2. The automatic optimal phase identification method for cardiac CT imaging of claim 1, wherein: the phase estimation network in the step A2 adopts a convolutional neural network, and is followed by a fully-connected network or a convlstm network, and the network loss function takes a mean square error MSE as an objective function.
3. The automatic optimal phase identification method for cardiac CT imaging of claim 1, wherein: in the step A4, a phase corresponding to the optimal label value is searched according to the label value of each CT image; or fitting a straight line or curve through the tag values, wherein the fitting result is expressed by y=fun (x), x is a measured tag value, y is a phase value, and the optimal center phase is finally obtained through the fitting result; or directly outputting a vector of phase-image quality of K points, obtaining the optimal phase according to the optimal label value, wherein K is the CT images of K different phases reconstructed for each patient when the heart segmentation network is established in the step A2.
4. The automatic optimal phase identification method for cardiac CT imaging of claim 1, wherein: the step A2 is to add a heart region based attention network before the phase estimation network is established.
5. The automatic optimal phase identification method for cardiac CT imaging according to claim 4, wherein the step of establishing the attention network based on the cardiac region in step A2 comprises the steps of:
c1, constructing a data set of the attention network based on the heart region: collecting projection data of K patient cardiac scans, reconstructing K CT images with different phases for each patient, marking a pericardial area in the obtained CT images, and performing Gaussian blur on masks of the CT images, wherein K is larger than or equal to 30 and K is larger than or equal to 20;
c2, designing a neural network, namely constructing a deep learning network structure of the U-Net, and positioning the position of the pericardium based on a hetmap regression;
and C3, network training: training a network, namely inputting training set and test set data into the network, and training to obtain network parameters of the attention network based on the heart region;
c4. cardiac localization: the reconstructed CT image is input into a network to obtain the approximate position of the heart, and a region with a fixed size is cut around the center of the predicted boundary frame to obtain the heart region.
6. The automatic optimal phase identification method for cardiac CT imaging of claim 1, wherein: and in the step A3, optimizing the network model built by the loss function training through a gradient descent method.
7. The automatic optimal phase identification method for cardiac CT imaging of claim 1, wherein: the projection data of the cardiac scan is obtained by adopting an axial scan mode, a step scan mode or a spiral scan mode which are synchronous according to the heartbeat phase.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911191975.6A CN110969633B (en) | 2019-11-28 | 2019-11-28 | Automatic optimal phase identification method for cardiac CT imaging |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911191975.6A CN110969633B (en) | 2019-11-28 | 2019-11-28 | Automatic optimal phase identification method for cardiac CT imaging |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110969633A CN110969633A (en) | 2020-04-07 |
CN110969633B true CN110969633B (en) | 2024-02-27 |
Family
ID=70032238
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911191975.6A Active CN110969633B (en) | 2019-11-28 | 2019-11-28 | Automatic optimal phase identification method for cardiac CT imaging |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110969633B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111462020B (en) * | 2020-04-24 | 2023-11-14 | 上海联影医疗科技股份有限公司 | Method, system, storage medium and apparatus for motion artifact correction of cardiac images |
CN112288752B (en) * | 2020-10-29 | 2021-08-27 | 中国医学科学院北京协和医院 | Full-automatic coronary calcified focus segmentation method based on chest flat scan CT |
CN113256529B (en) * | 2021-06-09 | 2021-10-15 | 腾讯科技(深圳)有限公司 | Image processing method, image processing device, computer equipment and storage medium |
CN117115577B (en) * | 2023-10-23 | 2023-12-26 | 南京安科医疗科技有限公司 | Cardiac CT projection domain optimal phase identification method, equipment and medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103198299A (en) * | 2013-03-27 | 2013-07-10 | 西安电子科技大学 | Face recognition method based on combination of multi-direction dimensions and Gabor phase projection characteristics |
CN108230338A (en) * | 2018-01-11 | 2018-06-29 | 温州大学 | A kind of stereo-picture dividing method based on convolutional neural networks |
CN108280814A (en) * | 2018-02-08 | 2018-07-13 | 重庆邮电大学 | Light field image angle super-resolution rate method for reconstructing based on perception loss |
CN109993809A (en) * | 2019-03-18 | 2019-07-09 | 杭州电子科技大学 | Rapid magnetic resonance imaging method based on residual error U-net convolutional neural networks |
CN110309910A (en) * | 2019-07-03 | 2019-10-08 | 清华大学 | The adaptive micro imaging method of optimization and device based on machine learning |
CN110455747A (en) * | 2019-07-19 | 2019-11-15 | 浙江师范大学 | It is a kind of based on deep learning without halo effect white light phase imaging method and system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10997413B2 (en) * | 2018-03-23 | 2021-05-04 | NthGen Software Inc. | Method and system for obtaining vehicle target views from a video stream |
-
2019
- 2019-11-28 CN CN201911191975.6A patent/CN110969633B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103198299A (en) * | 2013-03-27 | 2013-07-10 | 西安电子科技大学 | Face recognition method based on combination of multi-direction dimensions and Gabor phase projection characteristics |
CN108230338A (en) * | 2018-01-11 | 2018-06-29 | 温州大学 | A kind of stereo-picture dividing method based on convolutional neural networks |
CN108280814A (en) * | 2018-02-08 | 2018-07-13 | 重庆邮电大学 | Light field image angle super-resolution rate method for reconstructing based on perception loss |
CN109993809A (en) * | 2019-03-18 | 2019-07-09 | 杭州电子科技大学 | Rapid magnetic resonance imaging method based on residual error U-net convolutional neural networks |
CN110309910A (en) * | 2019-07-03 | 2019-10-08 | 清华大学 | The adaptive micro imaging method of optimization and device based on machine learning |
CN110455747A (en) * | 2019-07-19 | 2019-11-15 | 浙江师范大学 | It is a kind of based on deep learning without halo effect white light phase imaging method and system |
Non-Patent Citations (1)
Title |
---|
李远 ; 解学乾 ; 张皓 ; 王毅 ; 李念云 ; 孟捷 ; 张贵祥.前瞻性心电门控单心动周期256排CT成像图像质量影响因素.《中国介入影像与治疗学》.2018, * |
Also Published As
Publication number | Publication date |
---|---|
CN110969633A (en) | 2020-04-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110969633B (en) | Automatic optimal phase identification method for cardiac CT imaging | |
Chen et al. | Improving the generalizability of convolutional neural network-based segmentation on CMR images | |
US20210104313A1 (en) | Medical image processing apparatus, medical image processing method and computer-readable medium | |
US20220028085A1 (en) | Method and system for providing an at least 3-dimensional medical image segmentation of a structure of an internal organ | |
US20210390696A1 (en) | Medical image processing apparatus, medical image processing method and computer-readable storage medium | |
US11922601B2 (en) | Medical image processing apparatus, medical image processing method and computer-readable medium | |
EP3624056B1 (en) | Processing image frames of a sequence of cardiac images | |
JP2004532703A (en) | Method for reconstructing high resolution 3D images | |
JP2006075601A (en) | Segmentation method of anatomical structure | |
Ferreira et al. | Automating in vivo cardiac diffusion tensor postprocessing with deep learning–based segmentation | |
CN113935976A (en) | Method and system for automatically segmenting blood vessels in internal organs by enhancing CT (computed tomography) image | |
Salahuddin et al. | Multi-resolution 3d convolutional neural networks for automatic coronary centerline extraction in cardiac CT angiography scans | |
CN113192069A (en) | Semantic segmentation method and device for tree structure in three-dimensional tomography image | |
US20220092771A1 (en) | Technique for quantifying a cardiac function from CMR images | |
CN111093506A (en) | Motion compensated heart valve reconstruction | |
WO2022096867A1 (en) | Image processing of intravascular ultrasound images | |
Pereira et al. | Fully automated quantification of cardiac indices from cine mri using a combination of convolution neural networks | |
CN113744215A (en) | Method and device for extracting center line of tree-shaped lumen structure in three-dimensional tomography image | |
KR20180008130A (en) | A method for correcting a motion in a coronary image using a convolutional neural network | |
EP3667618A1 (en) | Deep partial-angle coronary restoration | |
Regehr et al. | Machine learning and graph based approach to automatic right atrial segmentation from magnetic resonance imaging | |
Schirra et al. | Improvement of cardiac CT reconstruction using local motion vector fields | |
CN113223104B (en) | Cardiac MR image interpolation method and system based on causal relationship | |
US20230252636A1 (en) | Method and system for the automated determination of examination results in an image sequence | |
Ogiela et al. | Graph-based semantic description in medical knowledge representation and 3D coronary vessels recognition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |