CN104899904B - A kind of output method carrying out blood flow parameter image under low radiation dose - Google Patents
A kind of output method carrying out blood flow parameter image under low radiation dose Download PDFInfo
- Publication number
- CN104899904B CN104899904B CN201510297565.5A CN201510297565A CN104899904B CN 104899904 B CN104899904 B CN 104899904B CN 201510297565 A CN201510297565 A CN 201510297565A CN 104899904 B CN104899904 B CN 104899904B
- Authority
- CN
- China
- Prior art keywords
- image
- blood flow
- flow parameter
- cerebral
- radiation dose
- 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.)
- Expired - Fee Related
Links
Abstract
A kind of output method carrying out blood flow parameter image under low radiation dose, includes the following steps：The CT cerebral apoplexy image datas obtained under low radiation dose；The CT cerebral apoplexies image data is pre-processed；Obtain image feature in pretreated CT cerebral apoplexies image data；In medical large database concept, the image feature is matched, matching result is obtained；According to the matching result, blood flow parameter image is exported.The present invention efficiently uses complementary multi-modal medical data；The X-ray radiation dosage that patient receives in CT perfusion medical images is greatly lowered through the invention, improves the use value of the safety and CT of medical treatment for acute apoplexy.
Description
Technical field
It is of the present invention to realize medical method technical field using computer, and in particular to a kind of to reduce existing radiation
Dosage is to the method for obtaining same high quality blood flow dynamic parametric image in the case of 1/10.
Background technology
CT perfusions are to solve the most-often used medical image method of cranial vascular disease, especially cerebral apoplexy patient for diagnosing
And treatment means, however the sampling of continuous brain CT images is carried out to obtain blood flow multidate information in the prior art, this is adopted
When sample, the big of common thoracic cavity X scannings or static state is equivalent to using the dose of radiation of the cerebral apoplexy image scheme including CTP
8-9 times of the dose of radiation of brain, abdominal cavity CT.Cerebral apoplexy patient needs Multiple-Scan CTP, the dose of radiation of accumulation that can destroy health
Cell increases the risk of cancer.
The radiation agent quantifier elimination for reducing CT perfusions attracted attention since 2000.It reduces dose of radiation and drop may be used
Improvement on the sample modes such as low tube current, tube voltage, sample frequency, and it is most directly effectively and common wherein to reduce tube current
Reduction dose of radiation method because tube current is directly proportional to dose of radiation, however reduction tube current also can not avoid ground
Increase the noise and artifact in data, reduces the quality of image.
There are two bases to blood flow dynamic parametric image is calculated by the sinogram obtained from image documentation equipment in CT perfusions
This step.The first step：The reconstruction of image from sinogram to CT；Second step：From CT reconstruction images sequence to blood flow parameter image
Calculating, i.e. the process of deconvolution.
In order to reduce the noise and artifact in blood flow dynamic image, the accurate fixed of blood flow parameter is improved, initial method is
Including reduction mainly using the filtering in room and time dimension, scanning slice thickness is improved, but can be to reduce spatial resolution
For cost.The algorithm of space filtering was widely used in a variety of applications in recent years, including low-frequency filter, holding edge filter device
(edge-preserving filters), such as anisotropic filtering (anisotropic diffusion) (Saito et
Al.2008), two-sided filter (bilateral filtering) (Mendrik et al.2011), non-local average (non-
Local means) (Ma et al.2012), total variance constraint (total variation regularization) (Tian
Et al.2011), spatio-temporal filtering, the methods of such as high constraint rear-projection (HYPR), multiband filtering (MBF), can be used for reducing from
Noise in the Time-space serial CT perfusion images rebuild in sinogram and artifact.However only reduce the noise of reconstruction image simultaneously
Stability Calculation blood flow dynamic parametric image can not be solved the problems, such as from principle.
The process of blood flow parameter image is calculated from the Time-space serial of CTP, most widely used method is deconvolution
(deconvolution).However deconvolution is the ill-posed problem in an image, that is, solution that is unique and stablizing is not present,
And the dynamic blood flow parameter image being calculated by deconvolution in the Time-space serial image that low radiation dose collects is special
It is unstable and inaccurate.The Deconvolution Method of Model Independent is can relatively accurately to calculate different vascular morphologies, especially
Be Eigenvalues Decomposition method (Singular Value Decomposition, SVD) (Weisskoff,et
al.1996；Sorensen, et al.1996) and its extension, as cycle specificity value decomposes (block-
circulant SVD)(Wu et al.2003).However these methods are independently calculated both for each voxel (voxel),
Therefore unstability is stronger, and calculating application condition is big, and it is more stable to provide can not to effectively utilize the contextual information of image
Blood flow parameter.
Invention content
The object of the present invention is to provide a kind of under low radiation dose carries out the output method of blood flow parameter image, has simultaneously
Effect ground uses the non-comparison CT images of high quality and the high quality blood flow of angiography and other patients of cerebral apoplexy patient itself
Dynamics parametric image improves the blood flow dynamic parameter under low radiation dose in CT perfusions and calculates, is effectively reduced the high spoke of tradition
The risk of cancer present in the CT perfusion influence mode of dosage is penetrated, to overcome deficiencies of the prior art.
The purpose of the present invention is achieved through the following technical solutions：One kind carrying out blood flow parameter under low radiation dose
The output method of image, includes the following steps：
S1, the CT cerebral apoplexy image datas obtained under low radiation dose；
S2 pre-processes the CT cerebral apoplexies image data；
S3 obtains image feature in pretreated CT cerebral apoplexies image data；
S5 matches the image feature, obtains matching result in medical large database concept；
S6 exports blood flow parameter image according to the matching result.
Preferably, in S1, the CT cerebral apoplexies image data include the non-comparison CT of static state before and after developing CT perfusion images,
CT angiographies and CT perfusion image sequences.
Preferably, S2, it includes image registration, benchmark static state to carry out pretreated method to the CT cerebral apoplexies image data
Image calculating, dynamic image calculating, brain Mask calculating, space filtering, time filtering, tissue segmentation, non-cerebral tissue stripping,
Sample frequency interpolation, hematocrit value correction, artery and venous locations are chosen, partial volume effect correction and time point intercept.
Preferably, S3, pretreated CT cerebral apoplexies image data include for static state CT, angiography and perfusion image；
The image feature includes institutional framework, blood vessel structure and blood flow multidate information.
Preferably, the institutional framework is the anatomical structure information extracted in the static state CT；The blood vessel structure is
Structure and the position of blood vessel are extracted in the angiography；The blood flow multidate information is each of perfusion image extraction
Blood flow performance graph in voxel.
Preferably, in medical large database concept, the image feature is matched specially：
CTP images are obtained in medical large database concept according to the image feature；
Extraction and identical institutional framework in the image feature in CTP images, as the first blood flow parameter；In CTP
Extraction and identical blood vessel structure in the image feature in image, as the second blood flow parameter；Extraction and institute in CTP images
Identical blood flow multidate information in image feature is stated, as third blood flow parameter；
First blood flow parameter, the second blood flow parameter, third blood flow parameter are combined, matching result is obtained.
CTP:Computed Tomography Perfusion CT scans are perfused
Deconvolution：The inverse process of convolution eliminates the processing method of previously filtered effect.
Beneficial effects of the present invention are：
The present invention is swept by learning hospital and the existing a large amount of high dose of radiation data of medical centre and patient in cerebral apoplexy
The structural information in the medical image without comparison development CT images and CT angiographies in scheme is retouched, it is freshly harvested low to improve
In dose of radiation CTP data, and accuracy and the picture quality of blood flow dynamic image is obtained by calculation.Efficiently use complementation
Multi-modal medical data：By the medicine shadow of other patients in the plurality of medical imaging modality data and data with existing of patient itself
As the message complementary sense combination in data, the information of a variety of data shapes and different patients is preferably comprehensively utilized.Through the invention
The X-ray radiation dosage that patient receives in CT perfusion medical images is greatly lowered, the safety and CT for improving medical treatment are used for suddenly
The use value of property cerebral apoplexy.
Description of the drawings
Fig. 1 is the flow diagram of the present invention.
Specific implementation mode
In the prior art, closest the technical scheme is that：A. it is based on study high dose blood flow dynamics Parameter Map
The sparse warp area method (Fang et al.2013) of picture.The method is mainly by the method for dictionary learning and sparse coding, from big
Measure high dose of radiation, high quality blood flow dynamics Parameter Map in study have characteristic dictionary, then use sparse coding
Method rebuild the image segments in low radiation dose.The method that Gai methods use machine learning and data mining, from existing
Medical image in learn structural information, for reducing the dose of radiation of CTP, improve the accuracy of blood flow parameter figure.(Fang
Et al.2013) for learning structural information from the blood flow parameter figure of the high dose of radiation of different patients, and have ignored brain soldier
The non-comparison development CT (Non-contrast CT) and CT angiographies (CT of this patient oneself in middle diagnosis imaging scheme
Angiography, CTA) in include patient oneself brain structure and functional information.B. retained using pre-development image
Brain CT perfusion iterative image reconstruction (Ma et al.2012) the method for the priori at edge is by using developer
The non-development static state CT images of a high quality are first obtained before brain CT perfusion images to provide priori, and combine non-
The average method in ground come improve CTP be in Time-space serial each single frames rebuild quality.This method uses the same patient's
Non- development still image enhances the accuracy of blood flow parameter image.For the Problems of Reconstruction rather than Parameter Estimation Problem of CTP,
The instability of blood flow dynamics parameter computation model is not solved fundamentally.
Existing method also exists as follows defect：A. (Fang et al.2013) is only from the cerebral blood flow of different patients
Learn structural information in dynamics parametric image, and has ignored the patient of low radiation dose image itself in cerebral apoplexy image scheme
In the non-comparison CT that has to scan through and angiography CTA.B. (Ma et al.2012) is directed to the reconstruction of each frame of CTP images,
Rather than it improves the stability for the deconvolution process that blood flow parameter image calculates and accurately determines.Even if the Time-space serial figure in CT perfusions
As having obtained preferable centre, instability and uncertain problem existing for blood flow dynamics parametric image itself are calculated still
The error that blood flow can be caused to calculate.
A kind of output method carrying out blood flow parameter image under low radiation dose as shown in Figure 1, includes the following steps：
S1, the CT cerebral apoplexy image datas obtained under low radiation dose；CT cerebral apoplexy image datas include that development CT is filled
Non- comparison CT, CT angiography of static state before and after note image and CT perfusion image sequences；
S2 pre-processes CT cerebral apoplexy image datas；
S3 obtains image feature in pretreated CT cerebral apoplexies image data；Pretreated CT cerebral apoplexies image number
According to including for static CT, angiography and perfusion image；Image feature includes that institutional framework, blood vessel structure and blood flow dynamic are believed
Breath；
S5 matches image feature in medical large database concept, obtains matching result；
S6 exports blood flow parameter image according to matching result.
In the present embodiment, S2, it includes that image registration, benchmark are quiet to carry out pretreated method to CT cerebral apoplexy image datas
The calculating of state image, dynamic image calculating, brain Mask calculating, space filtering, time filtering, tissue segmentation, the stripping of non-cerebral tissue
From, the correction of sample frequency interpolation, hematocrit value, artery and venous locations are chosen, partial volume effect correction and time point are cut
It takes.
In the present embodiment, S3, static CT are to be pre-processed by the non-comparison CT of static state in S2, and angiography is
It is pre-processed by the CT angiographies in S2, perfusion image is to pre-process to obtain by the CT perfusion image sequences in S2
's；Institutional framework is the anatomical structure information extracted in static state CT；Blood vessel structure is the structure that blood vessel is extracted in angiography
The position and；Blood flow multidate information is the blood flow performance graph in each voxel of perfusion image extraction.
In the present embodiment, in medical large database concept, image feature is matched specially：
CTP images are obtained in medical large database concept according to image feature；
Extraction and identical institutional framework in image feature in CTP images, as the first blood flow parameter；In CTP images
Middle extraction and identical blood vessel structure in image feature, as the second blood flow parameter；In CTP images in extraction and image feature
Identical blood flow multidate information, as third blood flow parameter；
First blood flow parameter, the second blood flow parameter, third blood flow parameter are combined, matching result is obtained.
The blood flow dynamics Parameter Map (blood flow velocity, blood flow, mean transit time etc.) of CTP is radiologist to disease
The state of an illness of people makes the key images information of diagnosis and quality-determining scheme.Current CTP scanning devices and software are to each
Picture point individually calculates blood flow dynamics parameter value, the Parameter Map obtained in the case of low radiation dose, high noisy and artifact
As having prodigious error and noise.This patent is proposed the structure of blood flow dynamic model and big data excavation in physiology
Habit is combined, and by the pharmacokinetic model of the CTP image datas of low radiation dose patient and is based on a unified model
Dictionary learning is combined with the machine learning model of sparse coding, and is proposed iteration optimization algorithms and can comparatively fast be obtained optimal solution.
The present invention is described in detail above by specific and preferred embodiment, but those skilled in the art should be bright
In vain, the invention is not limited in embodiment described above, all within the spirits and principles of the present invention, made by it is any modification,
Equivalent replacement etc., should all be included in the protection scope of the present invention.
Claims (2)
1. a kind of output method carrying out blood flow parameter image under low radiation dose, which is characterized in that include the following steps：
S1, the CT cerebral apoplexy image datas obtained under low radiation dose；Wherein, the CT cerebral apoplexies image data includes development
Non- comparison CT, CT angiography of static state before and after CT perfusion images and CT perfusion image sequences；
S2 pre-processes the CT cerebral apoplexies image data；Wherein, the CT cerebral apoplexies image data is pre-processed
Method include image registration, the calculating of benchmark still image, dynamic image calculating, brain Mask calculating, space filtering, the time filter
Wave, tissue segmentation, the stripping of non-cerebral tissue, the correction of sample frequency interpolation, hematocrit value, artery and venous locations selection, portion
Partial volume effect is corrected and time point interception；
S3 obtains image feature in pretreated CT cerebral apoplexies image data；Wherein, pretreated CT cerebral apoplexies image number
According to including for static CT, angiography and perfusion image；The image feature includes institutional framework, blood vessel structure and blood flow dynamic
Information；
S4 matches the image feature, obtains matching result in medical large database concept；Wherein to image spy
Sign is matched specially：CTP images are obtained in medical large database concept according to the image feature；It is extracted in CTP images
The identical institutional framework with the image feature, as the first blood flow parameter；Extraction and the image feature in CTP images
In identical blood vessel structure, as the second blood flow parameter；Extraction and identical blood flow in the image feature in CTP images
State information, as third blood flow parameter；First blood flow parameter, the second blood flow parameter, third blood flow parameter are combined, obtained
Matching result；
S5 exports blood flow parameter image according to the matching result.
2. carrying out the output method of blood flow parameter image under low radiation dose according to claim 1, it is characterised in that：Institute
It is the anatomical structure information extracted in the static state CT to state institutional framework；The blood vessel structure is to be extracted in the angiography
The structure of blood vessel and position；The blood flow multidate information is that the blood flow dynamic in each voxel of perfusion image extraction is bent
Line.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510297565.5A CN104899904B (en) | 2015-06-03 | 2015-06-03 | A kind of output method carrying out blood flow parameter image under low radiation dose |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510297565.5A CN104899904B (en) | 2015-06-03 | 2015-06-03 | A kind of output method carrying out blood flow parameter image under low radiation dose |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104899904A CN104899904A (en) | 2015-09-09 |
| CN104899904B true CN104899904B (en) | 2018-09-28 |
Family
ID=54032548
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510297565.5A Expired - Fee Related CN104899904B (en) | 2015-06-03 | 2015-06-03 | A kind of output method carrying out blood flow parameter image under low radiation dose |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104899904B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108461144A (en) * | 2017-12-21 | 2018-08-28 | 深圳大学 | Chinese population low-level radiation suffers from the appraisal procedure and system of cancer risk |
| CN110313925B (en) * | 2018-03-30 | 2023-07-04 | 萨摩亚商亚茂医疗科技有限公司 | Developing system and developing method applied to brain radiography |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103356206A (en) * | 2012-03-28 | 2013-10-23 | 通用电气公司 | System and method for contrast agent estimation in X-ray imaging |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6934353B2 (en) * | 2003-10-01 | 2005-08-23 | Ge Medical Systems Global Technology Company, Llc | Methods, apparatus, and computer readable mediums for performing perfusion studies |
-
2015
- 2015-06-03 CN CN201510297565.5A patent/CN104899904B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103356206A (en) * | 2012-03-28 | 2013-10-23 | 通用电气公司 | System and method for contrast agent estimation in X-ray imaging |
Non-Patent Citations (6)
| Title |
|---|
| Tissue-Specific Sparse Deconvolution for Low-Dose CT Perfusion;Ruogu Fang 等;《MICCAI 2013》;20131231;第8149卷;全文 * |
| Towards robust deconvolution of low-dose perfusion CT: Sparse perfusion deconvolution using online dictionary learning;Ruogu Fang 等;《Medical Image Analysis》;20131231;第17卷(第4期);摘要，正文第6节 * |
| 医学图像分析的现状与展望;陈武凡 等;《中国生物医学工程学报》;20080430;第27卷(第2期);全文 * |
| 应用网格和水印技术对基于内容的肝脏CT图像检索的研究;蒋历军;《中国博士学位论文全文数据库 信息科技辑》;20110415;第2011年卷(第4期);正文第1、12页 * |
| 正常成人低剂量多层螺旋CT灌注成像脑血流动力学定量研究;方明 等;《中国医刊》;20141231;第49卷(第10期);全文 * |
| 采用CT灌注成像血流和血容量测定区分急性缺血性卒中的半影和梗死区;Murphy B.D.等;《世界核心医学期刊文摘(神经病学分册)》;20061231(第12期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104899904A (en) | 2015-09-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10839567B2 (en) | Systems and methods for correcting mismatch induced by respiratory motion in positron emission tomography image reconstruction | |
| Tian et al. | Low-dose CT reconstruction via edge-preserving total variation regularization | |
| US9754390B2 (en) | Reconstruction of time-varying data | |
| US11399779B2 (en) | System-independent quantitative perfusion imaging | |
| Wu et al. | Self-supervised dynamic CT perfusion image denoising with deep neural networks | |
| Jia et al. | 4D computed tomography reconstruction from few-projection data via temporal non-local regularization | |
| EP3455824B1 (en) | Noise reduction in image data | |
| US20230222709A1 (en) | Systems and methods for correcting mismatch induced by respiratory motion in positron emission tomography image reconstruction | |
| Jung et al. | Deep learning cross-phase style transfer for motion artifact correction in coronary computed tomography angiography | |
| WO2022257959A1 (en) | Multi-modality and multi-scale feature aggregation for synthesizing spect image from fast spect scan and ct image | |
| JP6985056B2 (en) | Medical image processing equipment and medical image processing program | |
| CN104899904B (en) | A kind of output method carrying out blood flow parameter image under low radiation dose | |
| Manhart et al. | Denoising and artefact reduction in dynamic flat detector CT perfusion imaging using high speed acquisition: first experimental and clinical results | |
| US20240104705A1 (en) | Systems and methods for image correction | |
| CN107886554B (en) | Reconstruction of stream data | |
| Li et al. | Radiation dose reduction in CT myocardial perfusion imaging using SMART-RECON | |
| US20190188885A1 (en) | Model regularized motion compensated medical image reconstruction | |
| Khodajou-Chokami et al. | A deep learning method for high-quality ultra-fast CT image reconstruction from sparsely sampled projections | |
| Fang et al. | Tensor total-variation regularized deconvolution for efficient low-dose CT perfusion | |
| Gu et al. | Promote quantitative ischemia imaging via myocardial perfusion CT iterative reconstruction with tensor total generalized variation regularization | |
| Petibon et al. | Direct parametric reconstruction in dynamic PET myocardial perfusion imaging: in vivo studies | |
| JP7362460B2 (en) | Medical image processing device, method and storage medium | |
| Hu et al. | CROSS: Cross-domain residual-optimization based structure strengthening reconstruction for limited-angle CT | |
| Hofmann et al. | Removing blooming artifacts with binarized deconvolution in cardiac CT | |
| Giordano et al. | Spatially regularized region-based perfusion estimation in peripherals using angiographic C-arm systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180928 Termination date: 20190603 |