CN103027682A - Dynamic contrast-enhanced magnetic resonance imaging method and system - Google Patents

Abstract

The invention relates to a dynamic contrast-enhanced magnetic resonance imaging method and a dynamic contrast-enhanced magnetic resonance imaging system. The dynamic contrast-enhanced magnetic resonance imaging method comprises the following steps of: scanning and obtaining K space data; conducting nonlinear image reconstruction on the K space data to obtain reconstructed images; conducting support detection on the reconstructed images or the sparsity coefficient of the reconstructed images; repetitively iterating image reconstruction and the support detection to convergence; and generating magnetic resonance images. By adopting the dynamic contrast-enhanced magnetic resonance imaging method and the dynamic contrast-enhanced magnetic resonance imaging system, since reconstructed images and supports are obtained by repetitively iterating the image reconstruction and the support detection to convergence and the supports are updated to image reconstruction conducted at the next iteration, the support information obtained through detection is more, the measuring data contained by signals used for reconstructing precise images is less, the scanning time is shortened and the goal of improving the image quality is achieved.

Description

Dynamic contrast strengthens MR imaging method and system

Technical field

The present invention relates to mr imaging technique, particularly relate to a kind of dynamic contrast and strengthen MR imaging method and system.

Background technology

Nuclear magnetic resonance MRI(Magnetic Resonance Imaging) be the another major progress of medical imaging diagnosis technology after computed tomography (CT), be a kind of spin motion characteristics of certain specific atoms nuclei in the tissue and the biomagnetism nuclear spin imaging technique of nmr phenomena of utilizing that develops rapidly along with the development of computer technology, electronic circuit technology, superconductor technology, have the advantages that to reach higher contrast in tissue without wound, without ionizing radiation.MRI can obtain shape information and the function information of check point simultaneously as a kind of diagnostic techniques that is widely adopted at present, has the unrivaled advantage of other technology, becomes the important means of current medical imaging.

In recent years, the dynamic contrast that technical development is got up based on fast imaging strengthens nuclear magnetic resonance (DCE-MRI Dynamic Contrast Enhanced MRI) method and is based on the functional MR I technology that paramagnetic contrast medium injection blood vessel causes organizing longitudinal relaxation time T1 shortening, the variation that its use to repeat record by imaging tissue signal intensity is diffused into situation in the surrounding tissue in time to follow the tracks of contrast agent, approximately dynamic imaging is finished at 5-10 minute interval behind the contrast agent first passage, is the functional MR I method of quantitative study microvascular endothelial permeability.

DCE-MRI needs higher spatial resolution and temporal resolution as a kind of quantitative evaluating method, to guarantee the accuracy of quantitative measurement.Yet in traditional Fourier MRI, temporal resolution and spatial resolution are these those long relations that disappear under certain condition, are difficult to realize simultaneously the high time-space resolution imaging.

Summary of the invention

Based on this, be necessary for magnetic resonance image (MRI) problem of low quality, provide the higher dynamic contrast of a kind of magnetic resonance image (MRI) quality to strengthen MR imaging method.

In addition, also be necessary to provide the higher dynamic contrast of a kind of magnetic resonance image (MRI) quality to strengthen magnetic resonance imaging system.

A kind of dynamic contrast strengthens MR imaging method, comprising: scan and obtain the K spatial data; Described K spatial data is carried out nonlinear images rebuild, obtain reconstructed image; Sparse coefficient to described reconstructed image or described reconstructed image carries out the support detection; Described reconstructed image and described support detect iteration to convergence; Generate magnetic resonance image (MRI).

Therein among embodiment, the sparse coefficient of described reconstructed image or described reconstructed image is carried out the step process that support detects be: predetermined threshold value; Obtain the sparse coefficient value of image reconstruction value or described reconstructed image according to described reconstructed image, described image reconstruction value or described sparse coefficient value then obtain support information greater than described predetermined threshold value.

Therein among embodiment, described image reconstruction and described support detect iteration to the step process of convergence and are: in described image reconstruction process, block l by described support information solution ₁Minimization problem, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

Δ ^(i-1) be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain direction, and ε is noise level;

Block l by described support information solution ₁The computing formula of minimization problem is converted to weight l ₁Minimization problem, and by FOCUSS Algorithm for Solving reconstructed image, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

W is the diagonal angle weighting matrix;

In described support testing process, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for described spatial domain and time-frequency domain.

Among embodiment, the computing formula of described threshold value is therein:

τ ⁽ⁱ⁾＝||ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

τ ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾0 increasing sequence is the exponential function of outside iterations i.

In addition, also be necessary to provide a kind of dynamic contrast to strengthen magnetic resonance imaging system, comprise: acquisition module is used for scanning and obtains the K spatial data; Image reconstruction module is used for that described K spatial data is carried out nonlinear images and rebuilds, and obtains reconstructed image; The support detection module is used for the sparse coefficient of described reconstructed image or described reconstructed image is carried out the support detection; The iteration module is used for described reconstructed image and described support are detected iteration to convergence; The image generation module is used for generating magnetic resonance image (MRI).

Among embodiment, described support detection module comprises therein: the predetermined threshold value unit is used for predetermined threshold value; Obtain the support information unit, be used for obtaining according to described reconstructed image the sparse coefficient value of image reconstruction value or described reconstructed image, described image reconstruction value or described sparse coefficient value then obtain support information greater than described predetermined threshold value.

Among embodiment, described iteration module comprises therein: the image reconstruction computing unit, block l by described support information solution ₁Minimization problem, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

Δ ^(i-1)Be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain direction, and ε is noise level;

Block l by described support information solution ₁The computing formula of minimization problem is converted to weight l ₁Minimization problem, and by FOCUSS Algorithm for Solving reconstructed image, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

W is the diagonal angle weighting matrix;

Support detection computations unit, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for described spatial domain and time-frequency domain.

The computing formula of threshold value described in embodiment is therein:

τ ⁽ⁱ⁾＝||ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

τ ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾0 increasing sequence is the exponential function of outside iterations i.

Above-mentioned dynamic contrast strengthens MR imaging method and system, by image reconstruction and support are detected iteration to convergence, namely obtain image and the support of a plurality of reconstructions, and support is updated in the image reconstruction that next iteration carries out, the support information that obtains of surveying more, it is also just fewer to reconstruct the measurement data that the applied signal of exact image comprises, thereby has shortened sweep time, and then has reached the purpose that has improved picture quality.

Description of drawings

Fig. 1 is the flow chart that dynamic contrast strengthens MR imaging method among the embodiment;

Fig. 2 is that the sparse coefficient to reconstructed image or reconstructed image carries out the method flow diagram that support detects among Fig. 1;

Fig. 3 is for rebuilding the magnetic resonance image (MRI) that obtains based on image reconstruction and support detection iteration to restraining the over-fitting of going forward side by side;

The magnetic resonance image (MRI) of Fig. 4 for only obtaining with traditional FOCUSS algorithm;

Fig. 5 is for adopting the resulting magnetic resonance image (MRI) of OMP method;

Fig. 6 is the structural representation that dynamic contrast strengthens magnetic resonance imaging system among the embodiment;

Fig. 7 is the structural representation of support detection module among Fig. 7;

Fig. 8 is the structural representation of iteration module among Fig. 7.

The specific embodiment

As shown in Figure 1, in one embodiment, dynamic contrast strengthens MR imaging method, comprises the steps:

Step S10: scan and obtain the K spatial data.Sweep signal by the radio-frequency coil of magnetic resonance equipment is launched tissue to be checked (for example mammary gland) receives and obtains the K spatial data.

Step S20: the K spatial data is carried out nonlinear images rebuild, obtain reconstructed image.

In the present embodiment, in the dynamic scan process of image, receive the signal that tissue to be checked discharges, obtain the K spatial data, and carry out nonlinear images according to the K spatial data and rebuild, obtain reconstructed image.

Step S30: the sparse coefficient to reconstructed image or reconstructed image carries out the support detection.

In the present embodiment, the Dynamic MRI image is sparse in spatial domain and time-frequency domain (x-f) territory, is the prior information of dynamic imaging.In the x-f territory, at first adopt compressed sensing (Compress Sensing is called for short CS) technology that primary data is carried out image reconstruction.

After first the data of spatial domain and time-frequency domain being carried out image reconstruction, the data of spatial domain and time-frequency domain are carried out support detect, perhaps to reconstructed image sparse coefficient carry out support and detect.Support is the position of nonzero element in the x-f territory, and the known support information of the part of Dynamic MRI image in the x-f territory (Partial Known Support is called for short PKS) is prior information, so detect the support information that obtains by support.

Further, by reference to the accompanying drawings 2, the process of step S30 is:

Step S310: predetermined threshold value.

Choosing of predetermined threshold value, if the position that less threshold value can cause a lot of mistakes to detecting support, after the image reconstruction iteration in can not get correction; If larger threshold value can cause seldom positional value to detecting support, need iteration many times.

So the setting of threshold value can be according to following computing formula:

τ ⁽ⁱ⁾＝||ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

τ ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾0 increasing sequence is the exponential function of outside iterations i, this exponential function is determined by exponential law:

B 〉=0, a 〉=0.

Step S320: acquire the sparse coefficient value of image reconstruction value or reconstructed image according to reconstructed image, image reconstruction value or sparse coefficient value then obtain support information greater than predetermined threshold value.

In the present embodiment, choose image reconstruction value or sparse coefficient value greater than the position study of predetermined threshold value and obtain the support information in x-f territory, be the process that support detects.

Step S40: image reconstruction and support detect iteration to convergence.

In the present embodiment, step (being step S20 and the step S30) iteration of image reconstruction and support detection is extremely restrained.

In image reconstruction process, block l by support information solution ₁Minimization problem, and then obtain intermediate reconstructed ρ ⁽ⁱ⁾, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}---\left(1\right)$

Δ ^(i-1)Be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain (k-t) direction, and ε is noise level, and i is iterations.

Block l by support information solution ₁The computing formula of minimization problem (being above-mentioned formula 1) is converted to weight l ₁Minimization problem, and by FOCUSS(gathering under determined system solution annual reporting law Focal Underdetermined System Solver, be called for short FOCUSS) Algorithm for Solving reconstruction ρ ⁽ⁱ⁾, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}---\left(2\right)$

W is the diagonal angle weighting matrix, and ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain (k-t) direction, and ε is noise level, and i is iterations.

In the support testing process, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}---\left(3\right)$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for spatial domain and time-frequency domain (x-f).

Step S50: generate magnetic resonance image (MRI).

In the present embodiment, image reconstruction and support detect and have passed through repeatedly iteration to restraining and obtaining reconstructed image, and this reconstructed image is magnetic resonance image (MRI).

Above-mentioned dynamic contrast strengthens MR imaging method, to utilize the Dynamic MRI image be sparse in the x-f territory and the known support information of part as prior information, by image reconstruction and support are detected iteration to convergence, spatial resolution and temporal resolution have been improved, and then reached and improved dynamic contrast, strengthen the purpose of the quality of nuclear magnetic resonance.

By reference to the accompanying drawings 3 ~ 5, design sketch is further set forth the effect of this programme by experiment.Fig. 3 namely adopts the resulting magnetic resonance image (MRI) of this programme for rebuilding the magnetic resonance image (MRI) that obtains based on image reconstruction and support detection iteration to restraining the over-fitting of going forward side by side; The magnetic resonance image (MRI) of Fig. 4 for only obtaining with traditional FOCUSS algorithm; Fig. 5 is for adopting OMP(two steps orthogonal matching pursuit) the resulting magnetic resonance image (MRI) of method.

By contrast as can be known (pay close attention in the image arrow indication zone), the magnetic resonance image (MRI) that the FOCUSS method that Fig. 4 adopts obtains tends to cause at phase-encoding direction more owes to sample pseudo-shadow; The magnetic resonance image (MRI) that the OMP method that Fig. 5 adopts obtains can cause more noise, causes picture quality relatively poor.And the image that adopts this programme to obtain, namely Fig. 3 has just overcome the defective of traditional scheme, and preferably magnetic resonance image (MRI) is provided.

Based on the above-mentioned method that provides, by reference to the accompanying drawings 6, also be necessary to provide a kind of dynamic contrast to strengthen magnetic resonance imaging system, comprising:

Acquisition module 10 is used for scanning and obtains the K spatial data.Sweep signal by the radio-frequency coil of magnetic resonance equipment is launched tissue to be checked (for example mammary gland) receives and obtains the K spatial data.

Image reconstruction module 20 is used for that the K spatial data is carried out nonlinear images and rebuilds, and obtains reconstructed image.

In the present embodiment, in the dynamic scan process of image, receive the signal that tissue to be checked discharges, obtain the K spatial data, and it is carried out the data that sparse conversion obtains spatial domain and time-frequency domain (x-f).

Support detection module 30 is used for the sparse coefficient of described reconstructed image or described reconstructed image is carried out the support detection.

In the present embodiment, the Dynamic MRI image is sparse in spatial domain and time-frequency domain (x-f) territory, is the prior information of dynamic imaging.In the x-f territory, at first adopt compressed sensing (Compress Sensing is called for short CS) technology that primary data is carried out image reconstruction.

After first the data of spatial domain and time-frequency domain being carried out image reconstruction, the data of spatial domain and time-frequency domain are carried out support to be detected, support is the position of nonzero element in the x-f territory, and the known support information of the part of Dynamic MRI image in the x-f territory (Partial Known Support, be called for short PKS) be prior information, so detect the support information that obtains by support.

Further, by reference to the accompanying drawings 7, support detection module 30 also comprises:

Predetermined threshold value unit 310 is used for predetermined threshold value.

The positional value of predetermined threshold value is chosen, if the position that less threshold value can cause a lot of mistakes to detecting support, after the image reconstruction iteration in can not get correction; If larger threshold value can cause seldom positional value to detecting support, need iteration many times.

So the setting of threshold value can be according to following computing formula:

τ ⁽ⁱ⁾＝||ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

τ ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾0 increasing sequence is the exponential function of outside iterations i, this exponential function is determined by exponential law:

B 〉=0, a 〉=0.

Obtain support information unit 320, be used for acquiring according to reconstructed image the sparse coefficient value of image reconstruction value or reconstructed image, image reconstruction value or sparse coefficient value then obtain support information greater than predetermined threshold value.

In the present embodiment, choose image reconstruction value or sparse coefficient value greater than the position study of predetermined threshold value and obtain the support information in x-f territory, be the process that support detects.

Iteration module 40 is used for image reconstruction and support are detected iteration to convergence.

In the present embodiment, image reconstruction and support are detected iteration to convergence, be the process that support detects.

Further, by reference to the accompanying drawings 8, iteration module 40 comprises:

Image reconstruction computing unit 410 in image reconstruction process, blocks l by support information solution ₁Minimization problem, and then obtain intermediate reconstructed ρ ⁽ⁱ⁾, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}---\left(1\right)$

Δ ^(i-1)Be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain (k-t) direction, and ε is noise level, and i is iterations.

Block l by support information solution ₁The computing formula of minimization problem (being above-mentioned formula 1) is converted to weight l ₁Minimization problem, and by FOCUSS(gathering under determined system solution annual reporting law Focal Underdetermined System Solver, be called for short FOCUSS) Algorithm for Solving reconstruction ρ ⁽ⁱ⁾, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}---\left(2\right)$

W is the diagonal angle weighting matrix, and ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain (k-t) direction, and ε is noise level, and i is iterations.

Support detection computations unit 420, in the support testing process, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}---\left(3\right)$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for spatial domain and time-frequency domain (x-f).

Image generation module 50 is used for generating magnetic resonance image (MRI).

In the present embodiment, image reconstruction and support detect and have passed through repeatedly iteration to restraining and obtaining reconstructed image, and this reconstructed image is magnetic resonance image (MRI).

Above-mentioned dynamic contrast strengthens magnetic resonance imaging system, to utilize the Dynamic MRI image be sparse in the x-f territory and the known support information of part as prior information, by image reconstruction and support are detected iteration to convergence, improve spatial resolution and temporal resolution, and then reached the purpose of the quality that has improved dynamic contrast enhancing nuclear magnetic resonance.

The above embodiment has only expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to claim of the present invention.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 (7)

1. a dynamic contrast strengthens MR imaging method, comprising:

Scanning also obtains the K spatial data;

Described K spatial data is carried out nonlinear images rebuild, obtain reconstructed image;

Sparse coefficient to described reconstructed image or described reconstructed image carries out the support detection;

Described reconstructed image and described support detect iteration to convergence;

Generate magnetic resonance image (MRI).

2. dynamic contrast according to claim 1 strengthens MR imaging method, it is characterized in that, the sparse coefficient of described reconstructed image or described reconstructed image is carried out the step process that support detects be:

Predetermined threshold value;

Obtain the sparse coefficient value of image reconstruction value or described reconstructed image according to described reconstructed image, described image reconstruction value or described sparse coefficient value then obtain support information greater than described predetermined threshold value.

3. dynamic contrast according to claim 2 strengthens MR imaging method, it is characterized in that, described image reconstruction and described support detect iteration to the step process that restrains and is:

In described image reconstruction process, block l by described support information solution ₁Minimization problem, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

Δ ^(i-1)Be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain direction, and ε is noise level;

Block l by described support information solution ₁The computing formula of minimization problem is converted to weight l ₁Minimization problem, and by FOCUSS Algorithm for Solving reconstructed image, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

W is the diagonal angle weighting matrix;

In described support testing process, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for described spatial domain and time-frequency domain.

4. dynamic contrast according to claim 3 strengthens MR imaging method, it is characterized in that the computing formula of described threshold value is:

τ ⁽ⁱ⁾＝ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

τ ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾0 increasing sequence is the exponential function of outside iterations i.

5. a dynamic contrast strengthens magnetic resonance imaging system, it is characterized in that, comprising:

Acquisition module is used for scanning and obtains the K spatial data;

Image reconstruction module is used for that described K spatial data is carried out nonlinear images and rebuilds, and obtains reconstructed image;

The support detection module is used for the sparse coefficient of described reconstructed image or described reconstructed image is carried out the support detection;

The iteration module is used for described reconstructed image and described support are detected iteration to convergence;

The image generation module is used for generating magnetic resonance image (MRI).

6. dynamic contrast according to claim 5 strengthens magnetic resonance imaging system, it is characterized in that described support detection module comprises:

The predetermined threshold value unit is used for predetermined threshold value;

Obtain the support information unit, be used for obtaining according to described reconstructed image the sparse coefficient value of image reconstruction value or described reconstructed image, described image reconstruction value or described sparse coefficient value then obtain support information greater than described predetermined threshold value.

7. dynamic contrast according to claim 6 strengthens magnetic resonance imaging system, it is characterized in that described iteration module comprises:

The image reconstruction computing unit blocks l by described support information solution ₁Minimization problem, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|{\mathrm{\ρ}}_{{\mathrm{\Δ}}^{(i-1)}}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

Δ ^(i-1)Be the outer nonzero element position of known support, ρ is the image sequence in spatial domain and the time-frequency domain, and d is transform-domain signals, and F is the two-dimensional Fourier transform of spatial frequency domain and time domain direction, and ε is noise level;

Block l by described support information solution ₁The computing formula of minimization problem is converted to weight l ₁Minimization problem, and by FOCUSS Algorithm for Solving reconstructed image, computing formula is:

$\underset{\mathrm{\ρ}}{\min }{\left|\right|W^{(i-1)}\mathrm{\ρ}\left|\right|}_{1}s.t.{\left|\right|d-\mathrm{F\ρ}\left|\right|}_2\≤\mathrm{\ϵ}$

W is the diagonal angle weighting matrix;

Support detection computations unit, the computing formula that obtains support information is:

$T^{\left(i\right)}:=\{z:|{\mathrm{\ρ}}_z^{\left(i\right)}|>{\mathrm{\τ}}^{\left(i\right)}\}$

Be ρ ⁽ⁱ⁾Z element, ρ ⁽ⁱ⁾Reconstructed image data value for described spatial domain and time-frequency domain.

8, dynamic contrast according to claim 7 strengthens magnetic resonance imaging system, it is characterized in that the computing formula of described threshold value is:

τ ⁽ⁱ⁾=||ρ ⁽ⁱ⁾|| _∞/δ ⁽ⁱ⁾

f ⁽ⁱ⁾Be threshold value, δ ⁽ⁱ⁾The O increasing sequence is the exponential function of outside iterations i.

Images