CN104992457A - CT image reconstruction method and system - Google Patents
CT image reconstruction method and system Download PDFInfo
- Publication number
- CN104992457A CN104992457A CN201510346492.4A CN201510346492A CN104992457A CN 104992457 A CN104992457 A CN 104992457A CN 201510346492 A CN201510346492 A CN 201510346492A CN 104992457 A CN104992457 A CN 104992457A
- Authority
- CN
- China
- Prior art keywords
- image
- component
- part diagram
- projection
- target image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The invention relates to a CT image reconstruction method and a CT image reconstruction method system. The method comprises obtaining projection data of CT scanning, performing iteration processing according to the projection data to obtain a target image, performing nonnegative processing on the target image to obtain a nonnegative image, performing linear decomposition on the nonnegative image to obtaining a main component image and a secondary component image, performing sparsification processing on the main component image and/or the secondary component image to obtain an optimized sparse solution, and obtaining a CT reconstruction image according to the optimized sparse solution. The CT image reconstruction method and the CT image reconstruction method system can improve the speed of CT image reconstruction, shorten the CT scanning time, and then reduce an X-ray radiation dose on a human body.
Description
Technical field
The present invention relates to technical field of medical image processing, particularly relate to a kind of CT image rebuilding method and system.
Background technology
Computer tomography (CT) is the important imaging means of one being obtained internal structure of body information by lossless manner, it has the inferior many merits of high resolving power, high sensitivity and multilayer, be one of maximum medical imaging diagnostic device of China's installation amount, be widely used in each medicinal and check field.But owing to needing to use X ray in CT scan process, therefore in CT scan, the radiation dose problem of X ray is more and more subject to people's attention.Reasonable employment low dosage (As Low As Reasonably Achievable, ALARA) principles and requirements, under the prerequisite meeting clinical diagnosis, reduces the radiation dose to patient as far as possible.At present, conventional CT scan and formation method, consuming time longer, therefore radiation hazard to a certain degree is still existed to human body.
Summary of the invention
Based on this, be necessary for above-mentioned technical matters, a kind of CT image rebuilding method and system are provided.It can improve the speed of CT image reconstruction, shortens the CT scan time, thus reduces X ray to the radiation dose of human body.
A kind of CT image rebuilding method, described method comprises:
Obtain the data for projection of CT scan;
Iterative processing is carried out, to obtain target image according to described data for projection;
Non-negative process is carried out to described target image, obtains the non-negative image of described target image;
Linear decomposition is carried out to described non-negative image, obtains major component image and time component-part diagram picture;
LS-SVM sparseness is carried out to described major component image and/or described component-part diagram picture, obtains the optimization sparse solution meeting objective function;
Obtain CT according to described optimization sparse solution and rebuild image.
Wherein in an embodiment, described step of carrying out iterative processing according to described data for projection comprises:
Based on the imaging model of CT image, obtain the iterative model calculating target image according to described data for projection, the formula of described iterative model is expressed as:
Wherein, k is iterations, and λ is relaxation factor, and its span is that 0 < λ < 2, i, j is respectively positive integer, and 1≤i≤N, 1≤j≤M, w
ijfor the element in above-mentioned matrix of coefficients, I
θit is the projection index set under a projection angle;
The initial value of described target image is set, and utilize described iterative model to carry out iteration renewal to each pixel in described target image according to the iterations pre-set, obtain final target image, the current grayvalue of the pixel in described iterative model and the gray-scale value Uniform approximat of previous iteration.
Wherein in an embodiment, described the step that described target image carries out non-negative process to be comprised: gray-scale value in described target image is less than the pixel zero setting of 0.
Wherein in an embodiment, before the step of the data for projection of described acquisition CT scan, described method also comprises:
Obtain the projection image sequence collection of CT scan, pre-service is carried out to described projection image sequence collection and obtains described data for projection.
Wherein in an embodiment, the step that described non-negative image decomposes is comprised:
According to pre-conditioned, linear decomposition is carried out to described non-negative image, obtains described major component image and described component-part diagram picture, be describedly pre-conditionedly:
wherein, L represents described major component image, and E represents described component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of described L, α is regularization parameter,
be expressed as described component-part diagram value as determinant of a matrix.
Wherein in an embodiment, the step that described major component image and/or described component-part diagram picture carry out LS-SVM sparseness is comprised:
Extracting from described major component image and/or described component-part diagram picture can partly overlapping multiple image block;
Obtain the sparse coefficient that described multiple image block is corresponding;
Carry out optimization to described major component image and/or described component-part diagram picture, be met the optimization sparse solution of described objective function, described objective function is:
wherein, R
i∈ R
m × N, Δ represents described major component image or described component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
A kind of CT image re-construction system, described system comprises:
Acquisition module, for obtaining the data for projection that CT scan gathers;
Target image acquisition module, for carrying out iterative processing according to described data for projection, to obtain target image;
Non-negative image collection module, for carrying out non-negative process to described target image, obtains the non-negative image of described target image;
Decomposing module, for carrying out linear decomposition to described non-negative image, obtains major component image and time component-part diagram picture;
LS-SVM sparseness module, for carrying out LS-SVM sparseness to described major component image and/or described component-part diagram picture, obtains the optimization sparse solution meeting predetermined condition;
Rebuilding module, rebuilding image for obtaining CT according to described optimization sparse solution.
Wherein in an embodiment, described target image acquisition module is also for the imaging model based on CT image, and obtain the iterative model calculating target image according to described data for projection, the formula of described iterative model is expressed as:
wherein, k is iterations, and λ is relaxation factor, and its span is that 0 < λ < 2, i, j is respectively positive integer, and 1≤i≤N, 1≤j≤M, w
ijfor the element in above-mentioned matrix of coefficients, I
θit is the projection index set under a projection angle; The initial value of described target image is set, and utilize described iterative model to carry out iteration renewal to each pixel in described target image according to the iterations pre-set, obtain final target image, the current grayvalue of the pixel in described iterative model and the gray-scale value Uniform approximat of previous iteration.
Wherein in an embodiment, the pixel zero setting of described non-negative image collection module also for gray-scale value in described target image being less than 0.
Wherein in an embodiment, described decomposing module also for carrying out linear decomposition according to pre-conditioned to described non-negative image, obtains described major component image and described component-part diagram picture, is describedly pre-conditionedly:
wherein, L represents described major component image, and E represents described component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of described L, α is regularization parameter
be expressed as described component-part diagram value as determinant of a matrix.
Wherein in an embodiment, LS-SVM sparseness module comprises:
Image block extraction module, can partly overlapping multiple image block for extracting from described major component image and/or described component-part diagram picture;
Sparse coefficient acquisition module, for obtaining sparse coefficient corresponding to described multiple image block;
Optimization module, for carrying out optimization to described major component image and/or described component-part diagram picture, be met the optimization sparse solution of described objective function, described objective function is:
wherein, R
i∈ R
m × N, Δ represents described major component image or described component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
Above-mentioned CT image rebuilding method and system, by carrying out non-negative process to target image, obtain the non-negative image of target image, then non-negative image is decomposed, obtain major component image and time component-part diagram picture, finally LS-SVM sparseness is carried out to major component image and/or secondary component-part diagram picture, obtain optimization sparse solution, realize CT image reconstruction according to this optimization sparse solution, reduce the dimension of the image array in calculating process, improve the efficiency of image reconstruction.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of CT image rebuilding method in an embodiment;
Fig. 2 is the process flow diagram carrying out iterative processing in an embodiment according to data for projection;
Fig. 3 is the flow process of in an embodiment, major component image and/or secondary component-part diagram picture being carried out to LS-SVM sparseness;
Fig. 4 is the structured flowchart of CT image re-construction system in an embodiment;
Fig. 5 is the structured flowchart of LS-SVM sparseness module in an embodiment.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
At present, CT (Computed Tomography) scanning imaging system mainly divides following three parts: scanning system (X-ray tube, detector and scanning support), computer system, image display and store, photographic system.Wherein, the computer system of CT scan imaging system comprises principal computer and array computer two parts.Principal computer controls the work of whole CT scan imaging system, and its major function has: scanning monitoring, and data CT scan obtained store; The correction of CT value; The reconstruction control of image and the aftertreatment etc. of image.
As shown in Figure 1, in one embodiment, a kind of CT image rebuilding method is provided.Specifically describe the implementation procedure of the CT image reconstruction of the present embodiment below:
In a step 102, the data for projection of CT scan is obtained.
In some of them embodiment of the present invention, before beginning CT scan, character according to scanned object sets sweep parameter, the character of scanned object can be the physical propertys such as size, density, component, such as, scanned object can the object of different nature such as metal works, human body, insect, animal, plant, circuit board.Therefore for different scanned objects, need to set different sweep parameters, sweep parameter comprises projective amplification ratio, the data acquisition modes of detector, radiogenic voltage and power etc., and all sweep parameters remain unchanged in follow-up data acquisition.Such as, if scanned object is mouse, then set projective amplification than being 1:1, the data acquisition modes of detector is that radiogenic voltage is 80kv, and power is 15w continuously; If scanned object is ant, then set projective amplification than being 1:10, the data acquisition modes of detector is that radiogenic voltage is 20kv, and power is 10w continuously.
Gather darkfield image and bright-field image respectively, and obtain average dark field image and average bright-field image by sum-average arithmetic.Do not place scanned object in imaging viewing field, do not open light source and obtain some width darkfield images, such as, can gather 5 ~ 10 width darkfield images, darkfield image is sued for peace and is averaged obtain average dark field image according to the superposition of respective pixel gray-scale value.Open light source and gather some width bright-field image, and bright-field image sue for peace and is averaged obtain average bright-field image according to pixel grey scale superposition, significantly reduced the impact of reconstruction noise in image by darkfield image and bright-field image.
The scanned object rotation center that measurement is in imaging viewing field is to radiogenic distance and radiographic source to the distance of detector.Scanned object is placed in imaging viewing field, measures by the placement center of scanned object to radiogenic distance and radiographic source to the distance of detector, so that carry out image reconstruction.
Angularly spaced apart circumferential scanning is carried out to scanned object, obtains projection image sequence collection.The step of scanned object being carried out to angularly interval scan is: rotated one week continuous for turntable angularly compartment of terrain, and scan scanned object after rotating each time.Such as, the process of angularly interval scan can be: be placed in by scanned object on turntable, rotates 360 times continuously, and each rotation 1 degree, often rotates and once just once take, until turntable rotates a circle, obtain projection image sequence collection.
In addition, in some of them embodiment of the present invention, in order to avoid error, also need when obtaining projection image sequence collection the inspection carrying out turntable closed.Turntable closed refers to the degree of getting back to reference position after turntable rotates a week.Such as, initial position is A, place scanned object and rotate 360 times continuously by turntable, each rotation 1 degree, reach home position B after rotating a circle, the final position B obtained after rotating a circle in theory should overlap with initial position A, but due to error existing in actual mechanical system, result in initial position A and do not overlap with final position B.
Check that turntable closed is by angularly rotating preset times by scanned object, and at each angle shot image, to be captured complete after carry out subtracting each other of image, observe the image after subtracting each other, as long as the image after subtracting each other can carry out follow-up image scanning in desired extent, such as, if turntable closes completely, so " 0 degree of image " should be the same with " 360 degree of images ", and the image obtained after " 180 degree of images " being overturn also should be the same with " 0 degree of image ".Particularly, scanned object is placed in turntable, gather first 0 degree of image a, angularly (90 degree) rotate four times, according to the 90 degree of image b gathering scanned object, 180 degree of image c, 270 degree of image d and 360 degree of image e, use the flipped image of subtracted image c again after image a subtracted image e respectively, observable subtract each other after image, rule of thumb to judge whether the closed degree of turntable meets the demands, if do not met the demands, then to check that whether scanned object is firmly connected with turntable and whether turntable is stablized, to ensure that turntable closed can carry out follow-up scanning in allowed band.
Further, in some of them embodiment of the present invention, in order to remove the noise of projection image sequence collection, also anti-log operation is carried out to the projection image sequence collection that said process obtains, namely by following formula:
Wherein, x represent anti-log operate after data for projection, O represents thing field picture, i.e. the image of each scanning concentrated of projection image sequence, and R represents the average bright-field image of acquisition, and S represents average dark field image, obtains pretreated data for projection.
At step 104, carry out iterative processing according to data for projection, to obtain target image.
In some of them embodiment of the present invention, target image refers to initial image to be reconstructed.Utilizing the iterative model preset to carry out iterative processing to the pretreated CT scan data for projection that above-mentioned steps 102 obtains, obtaining the target image for rebuilding.
In one embodiment, as shown in Figure 2, carry out iterative processing according to data for projection to comprise with the step obtaining target image:
Step 124, based on the imaging model of CT image, obtains the iterative model calculating target image according to data for projection.
In some of them embodiment of the present invention, the imaging model of CT image can adopt following formula to represent:
p＝wx，
Wherein, p is data for projection, and w is system matrix, and x is target image.
The iterative model obtained can show with following formula table:
Wherein, iterations is k, λ is relaxation factor, and its span is that 0 < λ < 2, i, j is respectively positive integer, and 1≤i≤N, 1≤j≤M, w
ijfor the element in above-mentioned matrix of coefficients, I
θit is the projection index set under a projection angle.
The current grayvalue of the pixel in above-mentioned iterative model and the gray-scale value Uniform approximat of previous iteration.
Step 144, the initial value of Offered target image, and utilize iterative model to carry out iteration renewal to each pixel in target image according to the iterations pre-set, obtain final target image.
In some of them embodiment of the present invention, being composed by the initial value of target image x is 1, namely
j=1,2 ..., M, k are iterations, and generally getting k is positive integer between 10 to 200.
Utilize above-mentioned iterative model to carry out iteration to each pixel in target image according to the iterations k pre-set to upgrade each pixel in such target image x and carry out renewal correction according to above-mentioned iterative model, after iteration completes, just can obtain the iteration result X' of final target image.
Step 106, carries out non-negative process to target image, obtains the non-negative image of target image.
In some of them embodiment of the present invention, non-negative process refers to removes the minus pixel of gray-scale value in target image, can reduce the dimension of target image matrix like this, improve the efficiency of rebuilding.
Concrete, in some of them embodiment of the present invention, all pixels in the final target image X' obtain above-mentioned steps 144 carry out non-negative operation successively, that is:
After the operation of above-mentioned non-negative, just the gray-scale value can removed in the iteration result X' point that is less than 0, obtains non-negative image X
+.
Step 108, carries out linear decomposition to non-negative image, obtains major component image and time component-part diagram picture.
In some of them embodiment of the present invention, major component image refers to the element and structure that comprise " mainly " in image, and removes the image of noise and redundancy.Secondary component-part diagram similarly is the image referring to minutiae point or the unique point comprised in image.According to the decomposition model preset, decomposition is carried out to non-negative image and by original complex data dimensionality reduction, thus the speed of image reconstruction can be improved.
Concrete, in some of them embodiment of the present invention, in order to reduce operand, improve the speed of image reconstruction, by the non-negative image X that above-mentioned steps obtains
+be set to the linear unmixed model of major component image and time component-part diagram picture: X
+=a*L+b*E, wherein, L represents major component image, and E represents time component-part diagram picture, and a, b are respectively weight coefficient, and a, b are real number.In the present invention's some embodiments wherein, the value of a, b is respectively 1, and namely linear unmixed model is set to: X
+=L+E.
According to pre-conditioned, linear decomposition is carried out to non-negative image, obtains major component image and time component-part diagram picture, be pre-conditionedly:
Wherein, i, j are respectively positive integer, and L represents major component image, and E represents time component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of L, α is regularization parameter,
be expressed as the value of time component-part diagram as determinant of a matrix.
Solve decomposition model: X
+=L+E, the major component image L obtained and time component-part diagram make as E
value minimum.
Step 110, carries out LS-SVM sparseness to major component image and/or secondary component-part diagram picture, obtains the optimization sparse solution meeting objective function.
In one embodiment, as shown in Figure 3, the step that major component image and/or secondary component-part diagram picture carry out LS-SVM sparseness is comprised:
Step 11a, extracting from major component image and/or secondary component-part diagram picture can partly overlapping multiple image block.
Concrete, in some of them embodiment of the present invention, typical dictionary learning algorithm can be adopted, as K-SVD algorithm extracts from major component image and/or secondary component-part diagram picture can partly overlapping multiple image block.
Step 11b, obtains the sparse coefficient that multiple image block is corresponding.
Concrete, in some embodiments wherein, OMP (Orthogonal MatchingPursuit, orthogonal matching pursuit) algorithm or K-SVD algorithm can be adopted to obtain sparse coefficient corresponding to multiple image block.
Step 11c, carry out optimization to major component image and/or secondary component-part diagram picture, be met the optimization sparse solution of objective function, objective function is:
wherein, R
l∈ R
m × N, Δ represents major component image or secondary component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
Therefore, in this enforcement, optimization can be carried out to major component image L, be met the optimization sparse solution L of objective function
dL, also can carry out optimization to secondary component-part diagram as E, be met the optimization sparse solution E of objective function
dL, optimization can also be carried out to major component image L and time component-part diagram as E simultaneously, be met the optimization sparse solution L of objective function
dLand E
dL.
Step 112, obtains CT according to optimization sparse solution and rebuilds image.
According to the optimization sparse solution that above-mentioned steps 11c obtains, the decomposition model of integrating step 108 just can obtain rapidly rebuilds image X.
Based on the above embodiments, after obtaining optimization sparse solution sparse solution according to above-mentioned steps, just can according to the decomposition model of step 108, obtaining final CT reconstruction image can be following three kinds of situations:
(1) the optimization sparse solution L of major component image L
dLwith the superposition of former component-part diagram picture, i.e. X=L
dL+ E;
(2) the optimization sparse solution E of former major component image L and time component-part diagram picture
dLsuperposition, i.e. X=L+E
dL;
(3) the optimization sparse solution L of major component image L
dLwith the optimization sparse solution E of secondary component-part diagram picture
dLsuperposition, i.e. X=L
dL+ E
dL.
Above-mentioned CT image rebuilding method, by carrying out non-negative process to target image, obtain the non-negative image of target image, then non-negative image is decomposed, obtain major component image and time component-part diagram picture, finally LS-SVM sparseness is carried out to major component image and/or secondary component-part diagram picture, obtain optimization sparse solution, realize CT image reconstruction according to this optimization sparse solution, reduce the dimension of the image array in calculating process, improve the efficiency of image reconstruction.
In another embodiment, as shown in Figure 4, there is provided a kind of CT image re-construction system, this system comprises: acquisition module 402, target image acquisition module 404, non-negative image collection module 406, decomposing module 408, LS-SVM sparseness module 410 and reconstruction module 412.
The data for projection that acquisition module 402 gathers for obtaining CT scan.Target image acquisition module 404 for carrying out iterative processing according to data for projection, to obtain target image.Non-negative image collection module 506, for carrying out non-negative process to target image, obtains the non-negative image of target image.Decomposing module 408, for carrying out linear decomposition to non-negative image, obtains major component image and time component-part diagram picture.LS-SVM sparseness module 410, for carrying out LS-SVM sparseness to major component image and/or secondary component-part diagram picture, obtains the optimization sparse solution meeting predetermined condition.Rebuild module 412 and rebuild image for obtaining CT according to optimization sparse solution.
In one embodiment, target image acquisition module 404, also for the imaging model based on CT image, obtains the iterative model calculating target image according to data for projection; The initial value of Offered target image, and utilize iterative model to carry out iteration renewal to each pixel in target image according to the iterations pre-set, obtain final target image, the current grayvalue of the pixel in iterative model and the gray-scale value Uniform approximat of previous iteration.
In one embodiment, the pixel zero setting of non-negative image collection module 406 also for gray-scale value in target image being less than 0.
In one embodiment, decomposing module 408, also for decomposing non-negative image according to pre-conditioned, obtains major component image and time component-part diagram picture, is pre-conditionedly:
Wherein, L represents major component image, and E represents time component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of L, α is regularization parameter,
be expressed as the value of time component-part diagram as determinant of a matrix.
In one embodiment, as shown in Figure 5, LS-SVM sparseness module 410 comprises: image block extraction module 41a, sparse coefficient acquisition module 41b and optimization module 41c.
Image block extraction module 41a, can partly overlapping multiple image block for extracting in major component image and/or secondary component-part diagram picture.Sparse coefficient acquisition module 41b is for obtaining sparse coefficient corresponding to multiple image block.Optimization module 41c is used for carrying out optimization to major component image and/or secondary component-part diagram picture, and be met the optimization sparse solution of objective function, objective function is:
wherein, R
i∈ R
m × N, Δ represents major component image or secondary component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
Needs illustrate, the specific implementation process of the CT image re-construction system of some of them embodiment of the present invention is identical with CT image rebuilding method part, specifically see method section Example, can repeat no more here.
Fig. 1 is the schematic flow sheet of the CT image rebuilding method of one embodiment of the invention.Although it should be understood that each step in the process flow diagram of Fig. 1 shows successively according to the instruction of arrow, these steps are not that the inevitable order according to arrow instruction performs successively.Unless had explicitly bright herein, the order that the execution of these steps is strict limits, and it can perform with other order.And, step at least partially in Fig. 1 can comprise multiple sub-step or multiple stage, these sub-steps or stage are necessarily not complete at synchronization, but can perform in the different moment, its execution sequence does not also necessarily carry out successively, but with other steps or the sub-step of other steps or the executed in parallel at least partially in stage or alternately can perform.
The implementation of each embodiment only for corresponding steps in illustrating is set forth above, then in the not conflicting situation of logic, each embodiment above-mentioned be can mutually combine and form new technical scheme, and this new technical scheme is still in the open scope of this embodiment.
Through the above description of the embodiments, those skilled in the art can be well understood to the mode that above-described embodiment method can add required general hardware platform by software and realize, hardware can certainly be passed through, but in a lot of situation, the former is better embodiment.Based on such understanding, technical scheme of the present invention can embody with the form of software product the part that prior art contributes in essence in other words, this computer software product is carried in a non-volatile computer readable storage medium (as ROM, magnetic disc, CD, server cloud space), comprising some instructions in order to make a station terminal equipment (can be mobile phone, computing machine, server, or the network equipment etc.) perform method described in each embodiment of the present invention.
Each technical characteristic of the above embodiment can combine arbitrarily, for making description succinct, the all possible combination of each technical characteristic in above-described embodiment is not all described, but, as long as the combination of these technical characteristics does not exist contradiction, be all considered to be the scope that this instructions is recorded.
The above embodiment only have expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be construed as limiting the scope of the patent.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (10)
1. a CT image rebuilding method, described method comprises:
Obtain the data for projection of CT scan;
Iterative processing is carried out, to obtain target image according to described data for projection;
Non-negative process is carried out to described target image, obtains the non-negative image of described target image;
Linear decomposition is carried out to described non-negative image, obtains major component image and time component-part diagram picture;
LS-SVM sparseness is carried out to described major component image and/or described component-part diagram picture, obtains the optimization sparse solution meeting objective function;
Obtain CT according to described optimization sparse solution and rebuild image.
2. method according to claim 1, is characterized in that, describedly carries out iterative processing according to described data for projection and comprises with the step obtaining target image:
Based on the imaging model of CT image, obtain the iterative model calculating target image according to described data for projection, the formula of described iterative model is expressed as:
Wherein, k is iterations, and λ is relaxation factor, and its span is that 0 < λ < 2, i, j is respectively positive integer, and 1≤i≤N, 1≤j≤M, w
ijfor the element in above-mentioned matrix of coefficients, I
θit is the projection index set under a projection angle;
The initial value of described target image is set, and utilize described iterative model to carry out iteration renewal to each pixel in described target image according to the iterations pre-set, obtain described target image, the current grayvalue of the pixel in described iterative model and the gray-scale value Uniform approximat of previous iteration.
3. method according to claim 1 and 2, is characterized in that, describedly comprises the step that described target image carries out non-negative process: gray-scale value in described target image is less than the pixel zero setting of 0.
4. method according to claim 1, is characterized in that, before the step of the data for projection of described acquisition CT scan, described method also comprises:
Obtain the projection image sequence collection of CT scan, pre-service is carried out to described projection image sequence collection and obtains described data for projection.
5. method according to claim 1, is characterized in that, the step of described non-negative image being carried out to linear decomposition acquisition major component image and time component-part diagram picture comprises:
According to pre-conditioned, linear decomposition is carried out to described non-negative image, obtains described major component image and described component-part diagram picture, be describedly pre-conditionedly:
Wherein, L represents described major component image, and E represents described component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of described L, α is regularization parameter,
be expressed as described component-part diagram value as determinant of a matrix.
6. method according to claim 1, is characterized in that, comprises the step that described major component image and/or described component-part diagram picture carry out LS-SVM sparseness:
Extracting from described major component image and/or described component-part diagram picture can partly overlapping multiple image block;
Obtain the sparse coefficient that described multiple image block is corresponding;
Carry out optimization to described major component image and/or described component-part diagram picture, be met the optimization sparse solution of described objective function, described objective function is:
wherein, R
i∈ R
m × N, Δ represents described major component image or described component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
7. a CT image re-construction system, described system comprises:
Acquisition module, for obtaining the data for projection that CT scan gathers;
Target image acquisition module, for carrying out iterative processing according to described data for projection, to obtain target image;
Non-negative image collection module, for carrying out non-negative process to described target image, obtains the non-negative image of described target image;
Decomposing module, for carrying out linear decomposition to described non-negative image, obtains major component image and time component-part diagram picture;
LS-SVM sparseness module, for carrying out LS-SVM sparseness to described major component image and/or described component-part diagram picture, obtains the optimization sparse solution meeting predetermined condition;
Rebuilding module, rebuilding image for obtaining CT according to described optimization sparse solution.
8. system according to claim 7, is characterized in that, described target image acquisition module is also for the imaging model based on CT image, and obtain the iterative model calculating target image according to described data for projection, the formula of described iterative model is expressed as:
wherein, k is iterations, and λ is relaxation factor, and its span is that 0 < λ < 2, i, j is respectively positive integer, and 1≤i≤N, 1≤j≤M, w
ijfor the element in above-mentioned matrix of coefficients, I
θit is the projection index set under a projection angle; The initial value of described target image is set, and utilize described iterative model to carry out iteration renewal to each pixel in described target image according to the iterations pre-set, obtain final target image, the current grayvalue of the pixel in described iterative model and the gray-scale value Uniform approximat of previous iteration.
9. system according to claim 7, is characterized in that, described decomposing module also for carrying out linear decomposition according to pre-conditioned to described non-negative image, obtains described major component image and described component-part diagram picture, is describedly pre-conditionedly:
wherein, L represents described major component image, and E represents described component-part diagram picture, Σ
rσ
r(L) represent the nuclear norm of described L, α is regularization parameter,
be expressed as described component-part diagram value as determinant of a matrix.
10. system according to claim 7, is characterized in that, LS-SVM sparseness module comprises:
Image block extraction module, can partly overlapping multiple image block for extracting from described major component image and/or described component-part diagram picture;
Sparse coefficient acquisition module, for obtaining sparse coefficient corresponding to described multiple image block;
Optimization module, for carrying out optimization to described major component image and/or described component-part diagram picture, be met the optimization sparse solution of described objective function, described objective function is:
wherein, R
i∈ R
m × N, Δ represents described major component image or described component-part diagram picture, R
iΔ represents the image block extracted from Δ, || ||
2represent 2-norm, || ||
1represent 1-norm, γ is regularization parameter, and D represented complete dictionary, α
ibe i-th image block R
ithe sparse coefficient that Δ is corresponding, Γ is the sparse coefficient set of all image blocks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510346492.4A CN104992457B (en) | 2015-06-19 | 2015-06-19 | CT image rebuilding methods and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510346492.4A CN104992457B (en) | 2015-06-19 | 2015-06-19 | CT image rebuilding methods and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104992457A true CN104992457A (en) | 2015-10-21 |
| CN104992457B CN104992457B (en) | 2018-07-27 |
Family
ID=54304264
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510346492.4A Active CN104992457B (en) | 2015-06-19 | 2015-06-19 | CT image rebuilding methods and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104992457B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608717A (en) * | 2015-12-22 | 2016-05-25 | 肖古华 | CT system and CT image reconstruction method |
| CN106056644A (en) * | 2016-05-24 | 2016-10-26 | 深圳先进技术研究院 | Data processing method and data processing device for CT scanning |
| CN106650302A (en) * | 2017-02-24 | 2017-05-10 | 民勤县人民医院 | 3D root canal therapy plan control system based on internet of things |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060116567A1 (en) * | 2004-11-30 | 2006-06-01 | General Electric Company | Method and apparatus for image reconstruction using data decomposition for all or portions of the processing flow |
| CN102217934A (en) * | 2011-04-08 | 2011-10-19 | 中国科学院深圳先进技术研究院 | Magnetic resonance imaging method and system |
| CN103065342A (en) * | 2012-12-21 | 2013-04-24 | 深圳先进技术研究院 | Rebuilding method of computed tomography (CT) image |
| CN103150739A (en) * | 2013-03-04 | 2013-06-12 | 上海大学 | Video moving object partitioning algorithm based on multi-feature steady main component analysis |
| CN103268607A (en) * | 2013-05-15 | 2013-08-28 | 电子科技大学 | Common object detection method on weak supervision condition |
| CN103745445A (en) * | 2014-01-21 | 2014-04-23 | 中国科学院地理科学与资源研究所 | Gaussian and pulse mixed noise removing method and device |
| CN104274201A (en) * | 2014-10-10 | 2015-01-14 | 深圳先进技术研究院 | Method, system and equipment for tomography of mammary gland and image acquisition and processing method |
-
2015
- 2015-06-19 CN CN201510346492.4A patent/CN104992457B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060116567A1 (en) * | 2004-11-30 | 2006-06-01 | General Electric Company | Method and apparatus for image reconstruction using data decomposition for all or portions of the processing flow |
| CN102217934A (en) * | 2011-04-08 | 2011-10-19 | 中国科学院深圳先进技术研究院 | Magnetic resonance imaging method and system |
| CN103065342A (en) * | 2012-12-21 | 2013-04-24 | 深圳先进技术研究院 | Rebuilding method of computed tomography (CT) image |
| CN103150739A (en) * | 2013-03-04 | 2013-06-12 | 上海大学 | Video moving object partitioning algorithm based on multi-feature steady main component analysis |
| CN103268607A (en) * | 2013-05-15 | 2013-08-28 | 电子科技大学 | Common object detection method on weak supervision condition |
| CN103745445A (en) * | 2014-01-21 | 2014-04-23 | 中国科学院地理科学与资源研究所 | Gaussian and pulse mixed noise removing method and device |
| CN104274201A (en) * | 2014-10-10 | 2015-01-14 | 深圳先进技术研究院 | Method, system and equipment for tomography of mammary gland and image acquisition and processing method |
Non-Patent Citations (3)
| Title |
|---|
| QIONG XU等: "Low-Dose X-ray CT Reconstruction via Dictionary Learning", 《IEEE TRANS MED IMAGING》 * |
| 刘文磊: "低剂量CT成像与稀疏角度重建研究", 《中国优秀硕士学位论文全文数据库 信息科技辑(月刊）》 * |
| 郭威: "CT不完全投影数据重建算法研究", 《中国博士学位论文全文数据库 信息科技辑（月刊）》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608717A (en) * | 2015-12-22 | 2016-05-25 | 肖古华 | CT system and CT image reconstruction method |
| CN106056644A (en) * | 2016-05-24 | 2016-10-26 | 深圳先进技术研究院 | Data processing method and data processing device for CT scanning |
| CN106056644B (en) * | 2016-05-24 | 2020-03-17 | 深圳先进技术研究院 | Data processing method and device for CT scanning |
| CN106650302A (en) * | 2017-02-24 | 2017-05-10 | 民勤县人民医院 | 3D root canal therapy plan control system based on internet of things |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104992457B (en) | 2018-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104933744A (en) | Method and system for reestablishing CT image | |
| KR102174600B1 (en) | Method for processing multi-directional x-ray computed tomography image using artificial neural network and apparatus therefor | |
| US11354780B2 (en) | System and method for determining a trained neural network model for scattering correction | |
| US10984565B2 (en) | Image processing method using convolutional neural network, image processing device and storage medium | |
| Xu et al. | A practical cone-beam CT scatter correction method with optimized Monte Carlo simulations for image-guided radiation therapy | |
| Dong et al. | X-ray CT image reconstruction via wavelet frame based regularization and Radon domain inpainting | |
| US10825149B2 (en) | Defective pixel correction using adversarial networks | |
| CN105608717A (en) | CT system and CT image reconstruction method | |
| Zheng et al. | A dual-domain deep learning-based reconstruction method for fully 3D sparse data helical CT | |
| US11557071B2 (en) | Systems and methods for determining at least one artifact calibration coefficient | |
| CN106056644A (en) | Data processing method and data processing device for CT scanning | |
| CN104992457A (en) | CT image reconstruction method and system | |
| Wang et al. | An effective sinogram inpainting for complementary limited-angle dual-energy computed tomography imaging using generative adversarial networks | |
| CN113749618A (en) | Sports injury evaluation method, system, medium and terminal | |
| Wang et al. | One half-scan dual-energy CT imaging using the Dual-domain Dual-way Estimated Network (DoDa-Net) model | |
| CN117197349A (en) | CT image reconstruction method and device | |
| CN112001978B (en) | Method and device for reconstructing image based on dual-energy dual-90-degree CT scanning of generating countermeasure network | |
| Ma et al. | A neural network with encoded visible edge prior for limited‐angle computed tomography reconstruction | |
| Chen et al. | Dual-domain modulation for high-performance multi-geometry low-dose CT image reconstruction | |
| Patil et al. | Low performing pixel correction in computed tomography using deep learning | |
| KR102194009B1 (en) | POCS-based external information restoration method applicable to field modulation CT acquisition system | |
| Choi et al. | Chest tomosynthesis deblurring using CNN with deconvolution layer for vertebrae segmentation | |
| CN104382612A (en) | CT (Computed Tomography) data recovery method and device | |
| CN117672463B (en) | Data processing system and method for radiation therapy | |
| CN112509089B (en) | CT local reconstruction method based on truncated data extrapolation network |
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 |