Dynamic enhanced magnetic resonance imaging method based on magnetic resonance fingerprint
Technical Field
The invention belongs to the technical field of magnetic resonance imaging, and particularly relates to a Dynamic Contrast Enhanced (DCE) magnetic resonance imaging method based on Magnetic Resonance Fingerprinting (MRF).
Background
Dynamic Enhanced magnetic resonance imaging (DCE-MRI), by intravenous bolus injection of a paramagnetic Contrast agent, using rapid T1Weighted imaging sequence (T)1weighted imaging) to obtain a change curve of the signal intensity of the region of interest along with time, and then obtaining quantitative parameters reflecting tissue perfusion and vascular permeability based on pharmacokinetic model analysis: volume transfer constant Ktrans(volume transfer constant), interstitial-plasma rate constant Kep(Intersatiumto plasma rate), extracellular space volume fraction Ve(fractional extracellular space volume), plasma volume fraction Vp(fractional plasma volume)。
DCE-MRI is used as a non-invasive examination method, and can quantitatively evaluate the vascular characteristics such as tissue blood flow perfusion, permeability and the like, so that more abundant and accurate information can be provided for the fields of evaluating tumor angiogenesis, brain glioma grading, breast cancer diagnosis and treatment, liver cirrhosis, hepatic fibrosis evaluation and the like.
The existing DCE quantitative analysis method needs to establish a theoretical model under various assumptions, and then extracts K from experimental data by a model parameter fitting methodtrans、Kep、Vp、VeAnd the like. The perfusion signal synthesis reflects the dynamic process of blood flow and the infiltration of capillary vessels to surrounding tissues, and the simplified model is difficult to describe the complex perfusion process, so that the quantitative analysis error of DCE is larger.
Numerical simulation can overcome many limitations of theoretical models, make a more realistic description of complex perfusion processes, and can simulate contrast agent flow in blood, exchange between blood vessels and surrounding tissue, and diffusion processes outside blood vessels. And the numerical simulation has stronger expandability and can be used for meticulously depicting different research objects.
Magnetic Resonance Fingerprinting (MRF) is a new type of parametric quantitative imaging technique. Before the experiment, magnetic resonance signals under different parameters are simulated to establish a dictionary. Then, the magnetic resonance signals measured by the experiment are used as fingerprints to be matched with all channels in the dictionary, the most similar channel is selected, and the corresponding parameter information is extracted to be used as the final experiment result.
Disclosure of Invention
In order to overcome the defects of the DCE quantitative analysis, the invention provides a dynamic enhanced (DCE) magnetic resonance imaging method based on Magnetic Resonance Fingerprint (MRF).
Dynamic enhancement based on magnetic resonance fingerprint provided by the inventionThe magnetic resonance imaging method, abbreviated as DCE-MRF, is shown in FIG. 4. The invention obtains K from DCE signal reconstruction by using MRF dictionary matching methodtransAnd (3) the equal-parameter image replaces the original parameter acquisition mode of performing model fitting on the DCE signal.
The invention provides a dynamic enhanced magnetic resonance imaging method based on magnetic resonance fingerprints, which comprises the following specific steps:
step 1: before injecting contrast agent, it is first necessary to acquire the longitudinal relaxation time T of the measured region1Parametric quantitative images. Magnetic resonance acquisition sequences may be used, such as: a Look-Locker sequence, a MOLLI (modified Look-Locker) sequence, and a shmoli (shortenered MOLLI) sequence based on the inversion recovery pulse, or a sasha (preservation recovery acquisition) sequence, an MLLSR (modified Look-Locker) sequence, and the like based on the saturation recovery pulse;
step 2: the DCE raw image sequence is acquired. Firstly, injecting paramagnetic contrast agent into vein group, continuously collecting T in tested area in a period of time1Weighting the image;
and step 3: and (5) image registration. For a series of T acquired1Weighted image and T1Carrying out spatial position registration on the parameter quantitative images to ensure that the same voxels of all the images correspond to the same spatial position;
and 4, step 4: extracting a curve of the signal intensity of each voxel along the time as a DCE fingerprint of the voxel;
and 5: acquiring an Arterial Input Function (AIF) from the DCE raw image;
step 6: t obtained according to step 11Parametric quantitative images for different T1The parameter values establish a DCE dictionary, respectively. The dictionary is built as follows: at a given set of perfusion parameters (K)trans、Kep、Ve、Vp) And an arterial input function AIF, simulating the physiological processes of the paramagnetic contrast agent in the tissue of interest (including distribution of the contrast agent in the blood, exchange in the blood vessels and the surrounding tissue, in the surrounding tissue)Diffusion process of water molecules, etc.), and the diffusion process of water molecules, on the basis of the Bloch equation, a time-dependent curve of the magnetic resonance signal in the perfusion situation is derived as the set of perfusion parameters (K)trans、Kep、Ve、Vp) The corresponding DCE fingerprint. Different perfusion parameters are combined to form a parameter space, and DCE fingerprints corresponding to each group of parameters in the parameter space are calculated to obtain a DCE dictionary. For each T obtained in step 11Values, all using the above procedure to establish the T1A DCE dictionary corresponding to the value;
and 7: single voxel based dictionary matching. DCE fingerprint F of each voxel obtained in step 4acqWith the voxel T1Matching DCE dictionary corresponding to value, and screening out and F from dictionaryacqOne entry F with the largest correlation coefficientdicExtracting FdicCorresponding perfusion parameter (K)trans、Kep、Ve、Vp) A parameter value for the voxel;
and 8: and (5) reconstructing parameters of each voxel according to the step 6 to finally obtain a perfusion parameter quantitative image of the region of interest.
Compared with the prior art, the invention has the following advantages:
1. provides a brand-new DCE parameter quantitative analysis mode. The method of dictionary matching replaces the mode of model parameter fitting, and the irrationality caused by model simplification is made up;
2. the establishment of the dictionary is based on mathematical simulation, and compared with the prior DCE simplified model, the dictionary can describe richer and more real physiological processes: such as uneven distribution of contrast agent in blood, non-uniform effect of blood flow, diffusion process of contrast agent and water molecules, etc.;
3. the dictionary has extensibility. A targeted dictionary can be established according to different research requirements, and the content and size of the dictionary can be flexibly adjusted according to the interested parameters, the magnetic resonance sequence and the interested time period;
4. the invention mainly improves the parameter analysis process of DCE experiment, is compatible with the clinically existing DCE sequence, and is easy to popularize and realize.
Drawings
Fig. 1 shows the DCE sequence acquisition resulting in the original image (circles correspond to brain tumor regions).
FIG. 2 shows a voxel T in a brain tumor region1The values correspond to a schematic of the DCE dictionary.
FIG. 3 shows a voxel F in a brain tumor regionacq(corresponding to discrete points) matching result Fdic(corresponding to the solid line).
FIG. 4 is a flow chart of the method of the present invention (DCE-MRF).
Detailed Description
In the following, the embodiments of the present invention will be described in detail with reference to the accompanying drawings by taking DCE-MRF for analyzing brain tumor as an example, and fig. 4 is a flowchart of the DCE-MRF method provided by the present invention. It should be noted that several variations and modifications of the following steps are within the scope of the present invention without departing from the spirit of the present invention.
Step 1: scanning a patient's head for a regular T1、T2The magnetic resonance image is weighted to locate the tumor position and determine the scanning range of the DCE.
Step 2: before injection of contrast agent, T is acquired within the scan range determined in step 11Parametric quantitative images, using for example a Look-Locker sequence based on inversion recovery pulses.
And step 3: the DCE raw image sequence is acquired. Firstly, injecting a paramagnetic contrast agent Gd-DTPA into a vein, and continuously acquiring 36-phase T within 3 minutes for the scanning range determined in the step 1 in a period of 5 seconds1The image is weighted. In particular by successive application of T1And (4) weighting sequence, reconstructing the k-space data collected at each time point to obtain the DCE original image sequence. As shown in fig. 1.
And 4, step 4: for the DCE original image obtained in the step 3 and the T obtained in the step 21And (4) carrying out rigid registration on the parameter quantitative images to ensure that the same voxels in all the images correspond to the same spatial position, so as to prepare for a voxel-based dictionary matching process in the step 7.
And 5: the DCE fingerprint of each voxel inside the brain tumor was extracted. In the DCE image sequence, a curve of the signal intensity of each voxel along the time is extracted as the DCE fingerprint of the voxel. And, all fingerprints are standardized.
Step 6: an Arterial Input Function (AIF) is obtained from the DCE raw image. The specific method is to draw a small area on the artery near the brain tumor and extract the DCE curve of the area as the artery input function.
And 7: t obtained according to step 21Parametric quantitative images for different T1The parameter values establish a DCE dictionary, respectively. Dictionary passage perfusion parameter set (K) with different valuestrans、Ve、Vp) Stretch forming, wherein: ktransThe range is 0-0.3/min, and the value interval is 0.01/min; veAnd VpThe range of (1) is 0-1, and the value interval is 0.05. And calculating the time-varying curve of the magnetic resonance signals under each dictionary channel as a DCE fingerprint, and opening 13671 dictionary channels to form a DCE dictionary. Each fingerprint in the dictionary is also normalized. FIG. 2 shows a certain T1Schematic representation of the DCE dictionary under value.
And 8: DCE fingerprint F for each voxel in brain tumoracqDictionary matching is performed. The voxel is selected from the T1-DCE dictionary corresponding to the FacqOne entry F with the largest correlation coefficientdicExtracting FdicCorresponding perfusion parameter (K)trans、Ve、Vp) For the parameter value of the voxel, use Kep=Ktrans/ VeTo obtain KepAnd (4) parameters. FIG. 3 shows a schematic diagram of the matching process, wherein the scatter represents the experimentally acquired DCE fingerprint FacqThe solid line represents the sum F selected from the dictionaryacqItem F with highest relevancedic。
And step 9: calculating the corresponding parameter set for each voxel in the brain tumor according to step 8, (K)trans、Kep、Ve、Vp) Finally obtaining the K of the brain tumortransImage, KepImage, VeImage sum VpAnd (4) an image. While step 8 matching process can be utilizedCorrelation coefficient and match error (e.g. F)acqAnd FdicMean square error) of the DCE-MRF method, and an error map of the DCE-MRF method is obtained through reconstruction, so that the accuracy of the perfusion parameters obtained by the method can be conveniently evaluated.