Summary of the invention
The invention provides the bearing calibration of a kind of CT image metal artifacts, CT image metal artifacts means for correcting and a kind of CT equipment, it has overcome in existing CT image metal artifacts alignment technique, metal shadowing's data are difficult to location, the defects such as revised image malleable original image information, effectively improve CT image metal artifacts and remove efficiency and accuracy.
For addressing the above problem, the bearing calibration of a kind of CT image of the present invention metal artifacts, comprising:
By the original projection data acquisition original image of the CT diagnosis object that comprises objects interfered, and from described original image, extract original metal image;
Obtain the metal shadowing data corresponding with described original metal image;
Compare described original projection data and metal shadowing's data, obtain the diagnosis object data for projection of the CT diagnosis object of removing after objects interfered, based on described diagnosis object data for projection, proofread and correct described original projection data, obtain preliminary corrected projection data;
Obtain original correcting image corresponding to described preliminary corrected projection data;
Compare described original image and original correcting image, figure correction is carried out in image difference region between the two, set up model image, and obtain the model image data that described model image is corresponding;
With described model image data, again proofread and correct described original projection data, obtain target projection data;
By described target projection data, set up corresponding target image.
Alternatively, by described original projection data acquisition original image, obtained original correcting image and set up in the process of target image by described target projection data by described preliminary corrected projection data, at least one process adopts filtered back-projection method.
While alternatively, carrying out filtered back projection for the original projection data of the CT diagnosis object that comprises objects interfered, adopt the CT machine maximum scan visual field as rebuilding the visual field.
Alternatively, obtain the metal shadowing data corresponding with original metal image and comprise: the orthogonal projection that described original metal image is carried out based on frequency domain is processed, and obtains described metal shadowing data.
Alternatively, at original metal image, carry out in the orthogonal projection processing procedure based on frequency domain, after improving the resolution of described original metal image, then carry out successively two-dimensional Fourier transform and one dimension inversefouriertransform.
Alternatively, obtaining described preliminary corrected projection data comprises: described diagnosis object data for projection is carried out to linear interpolation correction, obtain interpolated projections data; Compare described original projection data and described interpolated projections data, obtain described preliminary corrected projection data.
Alternatively, described figure correction comprises: smoothing processing is carried out in the differential image region to described original image and original correcting image.
Alternatively, in described figure makeover process, repeatedly smoothing processing is carried out in the differential image region of described original image and original correcting image.
Alternatively, obtain diagnosis object data for projection and comprise: after obtaining described metal shadowing data, described metal shadowing data are done to orthogonal projection, thereby extract metal shadowing region; And in described metal shadowing regional extent, described in each, on original projection data basis, eliminate the described metal shadowing data corresponding with described original projection data, obtain the diagnosis object data for projection of removing after objects interfered.
Alternatively, obtain target projection data and comprise: in described metal shadowing region, adopt original projection data described in described model image Data correction.
Alternatively, described image difference region obtains by following manner: while comparing described original image and original correcting image, respectively described original image and described original correcting image are divided into some corresponding regions; Corresponding region between the two of comparison, and extract the corresponding region that cannot overlap between described original image and original correcting image and form described image difference region.
Alternatively, set up corresponding target image comprise by described target projection data: described in employing filtered back projection, target projection data are set up described target image, wherein, rebuilding the visual field is the reconstruction visual field that clinical interface is selected; Convolution kernel is the convolution kernel that clinical interface is selected; Building inconocenter is the center that clinical interface arranges.
The present invention also provides a kind of CT image metal artifacts means for correcting, comprising:
Original image acquiring unit by the original projection data acquisition original image of the CT diagnosis object that comprises objects interfered, and extracts original metal image from original image;
Artifact data capture unit, obtains the metal shadowing data corresponding with original metal image;
The first Data correction unit, compare described original projection data and metal shadowing's data, obtain the diagnosis object data for projection of the CT diagnosis object of removing after objects interfered, based on described diagnosis object data for projection, proofread and correct described original projection data, obtain preliminary corrected projection data;
Original correcting image acquiring unit, obtains original correcting image corresponding to described preliminary corrected projection data;
Model image processing unit, compares described original image and original correcting image, and figure correction is carried out in image difference region between the two, sets up model image, and obtains the model image data that described model image is corresponding;
The second Data correction unit, proofreaies and correct described original projection data again with described model image data, obtains target projection data;
Target image acquiring unit, sets up corresponding target image by described target projection data.
In addition, the present invention also provides a kind of CT equipment that comprises above-mentioned CT image metal artifacts means for correcting.
Compared with prior art, CT image metal artifacts of the present invention bearing calibration, device and CT equipment have the following advantages:
CT image metal artifacts of the present invention bearing calibration is determined the metal shadowing's scope as objects interfered according to original image corresponding to original projection data; Metal shadowing's data acquisition based within the scope of metal shadowing is removed the diagnosis object data for projection after objects interfered successively more afterwards, and based on described diagnosis object data for projection, model image is proofreaied and correct and set up to original projection data; And in the end according to the data for projection of model image, described original projection data are carried out to secondary correction, afterwards the target projection data based on after proofreading and correct according to clinical scanning used, build as condition and rebuild, obtain not having the target image of metal artifacts, thereby reach the object of calibration metal artifact.In its whole process, thereby the present invention has abandoned and has obtained the processing mode of image more clearly to obtained the direct modification processing of reconstruction image by raw data, in the present invention, all using original projection data as calibration object, thus spatial resolution and the low contrast ability of the image that assurance is obtained by correction data after processing.And original projection data have comprised all information of objects interfered completely and avoided introducing new artifact situation, when artifact is got rid of in its assurance, guarantee that the resolution of image is not suffered a loss.
In possibility, metal shadowing's data are obtained in the orthogonal projection of metal image based on carrying out based on frequency domain at original metal image, the accuracy while that it has guaranteed to obtain data, also do not affect processing speed, increased in CT image correction process the practicality of clinical implementation.
Embodiment
For above-mentioned purpose of the present invention, feature and advantage can be become apparent more, below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in detail.
Set forth in the following description a lot of details so that fully understand the present invention, but the present invention can also adopt other to be different from alternate manner described here, implement, so the present invention has not been subject to the restriction of following public specific embodiment.
Just as described in the background section, existing in CT image artifacts alignment technique, image resolution ratio after post-processing approach based on image area is processed is low, cause image blurring after correction, and in trimming process, tended to lose the relevant physical message of metal, the image after rebuilding tends to enter new artifact; And pre-treating method based on data for projection is in image correction process, the data of objects interfered are difficult to accurate location, cause the image after reconstruction to introduce new artifact or change original image information.CT image after defect based on above-mentioned CT image artifacts alignment technique makes to proofread and correct causes mistaken diagnosis, rate of missed diagnosis high.
The bearing calibration of a kind of CT image of the present invention metal artifacts has overcome above-mentioned existing CT image artifacts alignment technique defect, and original image corresponding to the original projection data of usining determined the metal shadowing's scope as objects interfered; Metal shadowing's data acquisition based within the scope of metal shadowing is removed the diagnosis object data for projection after objects interfered successively more afterwards, and based on described diagnosis object data for projection, model image is proofreaied and correct and set up to original projection data; And in the end according to the data for projection of model image, described original projection data are carried out to secondary correction; Afterwards the target projection data based on after proofreading and correct according to clinical scanning used, build as condition and rebuild, obtain not having the target image of metal artifacts, thereby reach the object of calibration metal artifact.
In the present invention, metal artifacts is proofreaied and correct in whole process, all usingd original projection data as calibration object, guaranteed spatial resolution and the low contrast ability of the image after processing.And the original projection data that original projection data of the present invention have comprised all information of objects interfered are completely as calibration object, avoided introducing new artifact situation.The described target image that adopts the present invention to obtain, has effectively reduced and has removed the CT image of the diagnosis object after objects interfered, effectively reduces and causes mistaken diagnosis, rate of missed diagnosis.
Shown in accompanying drawing 1, CT image metal artifacts bearing calibration detailed process of the present invention comprises:
Step S1: by the original projection data acquisition original image of the CT diagnosis object that comprises objects interfered; And from original image, extract original metal image, and obtain the corresponding metal shadowing of described original metal image data.
Step S2: compare described original projection data and metal shadowing's data, obtain the diagnosis object data for projection of the CT diagnosis object of removing after objects interfered; Based on described diagnosis object data for projection, proofread and correct described original projection data, obtain preliminary corrected projection data, and obtain original correcting image corresponding to described preliminary corrected projection data.
Step S3: compare described original image and original correcting image, figure correction is carried out in image difference region between the two, set up model image; And obtain the model image data that described model image is corresponding.
Step S4: again proofread and correct described original projection data with described model image data, obtain target projection data; Thereby obtain the target image corresponding with described target projection data.
Wherein, described objects interfered can comprise that high density material such as metal etc. can cause all materials of metal artifacts.
Below by conjunction with reference to figure 2, by specific embodiment, thereby set forth in more detail the present invention and advantage of the present invention.
Step S1: by the original projection data acquisition original image of the CT diagnosis object that comprises objects interfered; And from original image, extract original metal image, and obtain the corresponding metal shadowing of described original metal image data.
Specifically, first, by the original projection data filtering back projection of the CT diagnosis object that comprises objects interfered, obtain original image, wherein:
Original projection data P to the CT diagnosis object that comprises interfering picture
o p,n(P
o p,nfor the collimated beam equidistant data for projection of isogonism after resetting) carry out filtered back projection and obtain original image.
In specific implementation process, the parameter of setting up described original image comprises:
The reconstruction visual field F adopting
ovmaximum scan visual field F for CT machine
max, i.e. F
ov=F
max,
The convolution kernel K that filtered back projection's convolution kernel used is standard resolution
std,
And build the rotation center that inconocenter is CT machine, that is: ReconCenter=(0,0).
The described like this reconstruction visual field is the maximum scan visual field of CT machine, in CT original image trimming process, can not omit the CT original image information that comprises metal artifacts image afterwards.Thereby at CT image, compared with timing, cause new artifact, and the diagnostic message of disappearance, less mistaken diagnosis, the probability of failing to pinpoint a disease in diagnosis.
The reconstruction image that definition obtains by above-mentioned reconstruction parameter is I
o x, y, and the big or small N*N of image array.
Then, from original image, extract original metal image: wherein, described original metal image is defined as to I
m xy.
I
M x,y=I
O x,y*f(I
O x,y;t,w);
Wherein, threshold value t and threshold width w are pre-set metal threshold value and metal threshold widths.These two values depend on the filtration system of bulb, for difference, filter (head filtration, body filtration etc.) and have different parameter values.
Wherein, z is I
m x, y, a is t, b is w.
Then, original metal image is carried out to the orthogonal projection based on frequency domain, obtain metal shadowing's data.
Its orthogonal projection process based on frequency domain comprises:
The first step, by original metal image I
m x,ythe CT value of each pixel be converted to corresponding linear attenuation coefficient, the metal image I after conversion
μ x,yrepresent;
I
μ x,y=(I
M x,y+1000)/HU
scale;
Wherein, HU
scaledepend on output voltage, the collimating apparatus width of system, pass filter, the convolution kernel of the X ray bulb of system.
Second step, afterwards to the metal image I obtaining
μ x,ycarry out numerical values recited adjustment, the result after adjusting is used
represent;
Wherein, d is the physical separation size of equidistant collimated beam projection.
The 3rd step, by the image after adjusting
expand;
As, after expansion, the size of image is original four times, and the image of expansion represents with E, and extends equation is:
X=0,1 wherein ..., N-1, Y=0,1 ..., N-1, Δ=(3/2) N, wherein, N is pixel number.
By value corresponding to subscript index of the defined E of above-mentioned formula, be not set to 0.
It should be noted that in the present embodiment, by image spreading, to be original size 4 times, in practical operation, can specifically determine according to actual needs extend sizes.
The 4th step, to E
x,ycarry out two-dimensional Fourier transform, its result is used
represent;
After the expansion of above-mentioned the 3rd step image, can be to E
x,ywhile carrying out two-dimensional Fourier transform, to image spatial domain signal, implant enough interpolation, thereby guarantee the precision of two-dimensional Fourier transform, improve the frequency resolution of two-dimensional Fourier transform.
The 5th step, frequency interpolation, obtains
polar coordinates frequency-domain expression
Wherein, the frequency domain interval of rectangular coordinate is:
Polar frequency domain interval is:
Wherein, N
fFTlength for convolution in the reconstruction of step S1Zhong filtered back projection.And for the value of polar each point (p, n)
cubic(cubic polynomial around for value) interpolation.
Wherein for each projection angle:
Wherein, p=0,1 ..., N
p π-1, υ
ostart angle for first projection.
For each frequency indices:
n=0,1,...,N
FFT-1;
By interpolation process above, for each projection angle p, obtain following sequential value:
The 6th step, for what obtain
carry out deconvolution operation, obtain the Fourier transform of the data for projection of each projection angle,
The 7th step, for obtained above
carry out one dimension inversefouriertransform, obtain data for projection P corresponding in spatial domain
p, n,
By metal shadowing's data that method obtains above, be P
m p, n.
In above-mentioned steps, metal shadowing's data are obtained in the orthogonal projection of metal image based on carrying out based on frequency domain at original metal image, and it has guaranteed the data accuracy while obtaining, and has guaranteed processing speed, has increased in CT image correction process the practicality of clinical implementation.
Finally, determine metal shadowing region, it specifically comprises:
For the data P of metal shadowing
m p, n;
Carry out a plurality of angle orthogonal projection, and find out those non-vanishing successive ranges, and these non-vanishing successive ranges are exactly metal shadowing region, metal shadowing is [ S for region
i, E
irepresent.
Wherein, S
ifor the starting point in metal shadowing region, E
ifor the terminal in metal shadowing region, i=1,2 ..., k, wherein, k is the number of non-vanishing successive range.
In above-mentioned orthogonal projection process, the geometric parameter SID(bulb focus adopting is to the distance of frame rotation center), IDD(frame rotation center is to the distance of detecting device), the scan vision of SFOV(CT machine) numerical value be the physical geometry size that CT machine is corresponding, thereby make orthogonal projection geometric parameter used all consistent with the geometric parameter of actual machine scanning, guarantee the consistance of data, thereby reached the object of proofreading and correct original projection data.
Step S2: compare described original projection data and metal shadowing's data, obtain the diagnosis object data for projection of the CT diagnosis object of removing after objects interfered; And proofread and correct described original projection data based on described diagnosis object data for projection, and obtain preliminary corrected projection data, obtain original correcting image corresponding to described preliminary corrected projection data.
First, establish original projection data (the corresponding data for projection of original image) for P
o pn,
Original projection data P
o pnwith the data P of metal shadowing
m p,ndifference be P
d p,n,
P
D p,n=P
O pn-P
M p,n;
For each projection angle, in metallic region, difference data for projection is done to linear interpolation, obtain interpolated projections data
that is:
K ∈ (S, E) wherein.
Then, tentatively proofread and correct original projection data P
o p, nand obtain preliminary corrected projection data, its process is as follows:
Each subpoint for each metal shadowing region;
If interpolated data
be greater than difference data P
d p,n, revise original projection data P
o p,n, otherwise do not revise, i.e. preliminary corrected projection data,
Finally, according to setting up the condition of original image with original projection data in step S1, to described preliminary corrected projection data P
r o p,kcarry out filtered back projection's reconstruction, obtaining preliminary correcting image is I
c x, y.
Step S3: comparison original image I
o x,ywith preliminary correcting image I
c x,ymodel image is obtained in foundation:
Comprise: (1) comparison original image I
o x,ywith preliminary correcting image I
c x,yestablish target correcting area, its process comprises:
Set in advance two threshold value T
1and T
2, according to these two threshold values, image is divided into A district, three regions, B district, C district.A position is less than T
1the region that forms of pixel, B district be between the region of pixel composition, C district is for being greater than T
2the region that forms of pixel.
By original image I
o x,ydetermine B district, be defined as goal-selling correcting area, compare described original image I simultaneously
o x,ywith preliminary correcting image I
c x,ya district and C district, and within the scope of described A district and C district, described original image I
o x,ywith preliminary correcting image I
c x,ynonoverlapping part, and by these not lap be included into goal-selling correcting area, form target correcting area.Now, described target correcting area may be a discontinuous region.
In the present invention, the field of definition T in described B district
1and T
2described in step S1, the to establish region [ S of metal shadowing
i, E
i, described like this target correcting area is described original image I
o x,yimage I with preliminary correction
c x,ymetal shadowing region.
(2) according to target correcting area, set up model image, its process comprises:
According to determined target correcting area, based on original image I
o x,yand preliminary correcting image I
c x, yin the point that belongs in described target correction zone carry out repeatedly smoothing processing, but not point beyond target correcting area district do not process, the original image I after overcorrect
o x,ybe model image I
e x,y.
(3) model image data are obtained in the orthogonal projection of model image frequency domain:
By model image I
e x,ybased on frequency domain orthogonal projection, obtain model projection data P
e x,y.The method of its projection is carried out the orthogonal projection based on frequency domain as (3) in step S1 to metal image, thereby it is identical to obtain metal shadowing's data method, does not repeat them here.
Step S4: again proofread and correct described original projection data with described model image data, obtain target projection data; Thereby obtain the target image corresponding with described target projection data.
(1) with described model image data, again proofread and correct described original projection data, obtain target projection data:
At the region [ S of described metal shadowing
i, E
iin, with model projection data P
e x,yrevise original projection data P
o p, n, obtaining target projection data, the method for its correction is as follows:
P
R O p,k=P
R E p,k+(P
O p,Si-P
E p,Si)*(Ei-k)/(Ei-Si)+(P
O p,Si-P
E p,Si)*(k-Ei)/(Ei-Si)
In said process, first, when data for projection subscript value is view field's border numerical value, i.e. [ S
i, E
iinterior minimum value and maximal value, the revised projection data values of its correspondence equals the projection data values before correction, thereby can effectively guarantee to be corrected the continuity of data; Secondly, utilize the model data (P that there is no artifact
e x,y) revise the original projection data (P with artifact
o p, n), can have in the described original projection data of removal and produce the artifact data of metal artifacts, thereby obtain more accurate target projection data.
(2) obtain target image corresponding to target projection data:
By the result of final model correction, set up target image, and described target image is the result through metal artifacts correction, its condition is:
Rebuilding the visual field is the reconstruction visual field that clinical interface is selected;
Convolution kernel is the convolution kernel that clinical interface is selected;
Building inconocenter is the center that clinical interface arranges.
Thereby obtain the complete image-region that maximum diagnosis object is corresponding, guarantee the integrality of target image, so that inspection area scope and picture quality that Clinical CT operation technician can see according to hope require to select these parameters.
Shown in figure 3, the present invention also provides a kind of CT image metal artifacts means for correcting of realizing the bearing calibration of above-mentioned CT image metal artifacts, and it comprises:
Original image acquiring unit 10 by the original projection data acquisition original image of the CT diagnosis object that comprises objects interfered, and extracts original metal image from original image;
Artifact data capture unit 20, connects described original image acquiring unit 10, obtains the metal shadowing data corresponding with original metal image;
The first Data correction unit 30, connect described artifact data capture unit 20, compare described original projection data and metal shadowing's data, obtain the diagnosis object data for projection of the CT diagnosis object of removing after objects interfered, based on described diagnosis object data for projection, proofread and correct described original projection data, obtain preliminary corrected projection data;
Original correcting image acquiring unit 40, connects described the first Data correction unit 30, obtains original correcting image corresponding to described preliminary corrected projection data;
Model image processing unit 50, connect described original correcting image acquiring unit 40, compare described original image and original correcting image, figure correction is carried out in image difference region between the two, set up model image, and obtain the model image data that described model image is corresponding;
The second Data correction unit 60, connects described model image processing unit 50, with described model image data, again proofreaies and correct described original projection data, obtains target projection data;
Target image acquiring unit 70, connects described the second Data correction unit 60, by described target projection data, sets up corresponding target image.
CT image metal artifacts means for correcting of the present invention is after obtaining the CT original projection data of diagnosis object, and original image corresponding to the original projection data of usining determined the metal shadowing's scope as objects interfered; Metal shadowing's data acquisition based within the scope of metal shadowing is removed the diagnosis object data for projection after objects interfered successively more afterwards, and based on described diagnosis object data for projection, model image is proofreaied and correct and set up to original projection data; And in the end according to the data for projection of model image, described original projection data are carried out to secondary correction; Afterwards the target projection data based on after proofreading and correct according to clinical scanning used, build as condition and rebuild, obtain not having the target image of metal artifacts.Thus, described CT image metal artifacts means for correcting is usingd original projection data as calibration object, thus spatial resolution and the low contrast ability of the image that assurance is obtained by correction data after processing.And original projection data have comprised all information of objects interfered completely and avoided introducing new artifact situation, when artifact is got rid of in its assurance, guarantee that the resolution of image is not suffered a loss.The CT image quality that improves diagnosis object, obtains more accurate diagnostic message.
The present invention also provides a kind of CT equipment that comprises above-mentioned CT image metal artifacts means for correcting.
Although the present invention with preferred embodiment openly as above; but it is not for limiting the present invention; any those skilled in the art without departing from the spirit and scope of the present invention; can make possible change and modification, so protection scope of the present invention should be as the criterion with the scope that the claims in the present invention were defined.