TWI638335B - Method for correcting a diffusion image having an artifact - Google Patents

Abstract

The present invention relates to a method of correcting a diffused image having artifacts, the method comprising: (a) providing a set of diffused images comprising the diffuse image having artifacts; (b) calculating each of the set of diffused images a first signal intensity of the image; (c) a plot of the slice number of the set of diffused images relative to the first signal intensity; (d) calculating the diffused image with artifacts by interpolation on the map a second signal strength; and (e) correcting the diffused image with artifacts based on the second signal strength.

Description

 Method for correcting diffused image with artifact  

The present invention relates to a method of correcting a diffuse image with artifacts.

Magnetic resonance imaging (MRI) tractography is a mapping of nerve fiber structures based on tissue anisotropy diffusion-sensitive MRI. In order to be a survey tool of general value, neurosurgery must provide a schematic and accurate representation of neural connectivity in most real-world situations. Complex fiber structures are a ubiquitous feature of neuroanatomy, so neurospatial methods that reliably examine neural tissue structures can be very valuable. MRI neurosurgery, which began within the framework of diffusion tensor angiography (DTI), did not achieve this goal (Basser, PJ, Pajevic, S., Pierpaoli, C., Duda, J., Aldroubi, A., 2000. In Vivo fiber tractography using DT-MRI data. Magn. Reson. Med. 44, 625-632). The tensor model it relies on cannot distinguish the multiple fiber directions in the MRI voxel, so it is impossible to distinguish the fiber intersection, whether it is the tract intersection in the white matter or the fiber cross in the intricate structure of the gray matter. (Mori, S., van Zijl, PC, 2002. Fiber tracking: principles and strategies-a technical review. NMR Biomed. 15, 468-480). Methods to address this limitation in the DTI architecture have been proposed (eg, Behrens, TE, Woolrich, MW, Jenkinson, M., Johansen-Berg, H., Nunes, RG, Clare, S., Matthews, PM, Brady, JM, Smith, SM, 2003. Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn. Reson. Med. 50, 1077-1088) but does not imply the imaging of direct and detailed complex fiber structures according to principles.

Inspired by these limitations, the MRI method has been described as having the ability to resolve the heterogeneity of fiber orientation in each resolved tissue volume (stereopixel) (Tuch, DS, Reese, TG, Wiegell, MR, Makris, N., Belliveau). , JW, Wedeen, VJ, 2002. High angular resolution diffusion imaging reveals intravoxel white matter fiber heterogeneity. Magn. Reson. Med. 48, 577-582), and provides new insights into the structure of white matter fiber bundles (Schmahmann, JD, Pandya) , DN, Wang, R., Dai, G., D'Arceuil, HE, de Crespigny, AJ, Wedeen, VJ, 2007. Association fibre pathways of the brain: parallel observations from diffusion spectrum imaging and auto-radiography. Brain 130,630 -653). These methods go beyond DTI. They describe the diffusion in each voxel in a general function (Probability Density Function (PDF)) that specifies the 3D distribution of the microscopic displacement of the MR visible spins it contains for each voxel. These methods include diffusion spectrum MRI (DSI) (Lin, CP, Wedeen, VJ, Chen, JH, Yao, C., Tseng, WY, 2003. Validation of diffusion spectrum magnetic resonance imaging with manganese-enhanced rat optic tracts and ex vivo phantoms.NeuroImage 19, 482-495) (where PDF is all imaged with displaced 3D Fourier coding), and Q-ball method that is primarily sensitive to PDF angles (Tuch, DS, Reese, TG, Wiegell, MR, Wedeen, VJ, 2003) .Diffusion MRI of complex neural architecture. Neuron 40, 885-895) and related methods (Jansons, K., Alexander, D., 2003. Persistent angular structure: new insights from diffusion magnetic resonance imaging data. Inverse Problems 19, 1031-1046) .

Since Stejskal and Tanner described the basic diffusion sequence, diffusion weight magnetic resonance imaging (DWI) has become an established tool in clinical and research settings (Stejskal EO, Tanner JE. Spin Diffusion Measurements: Spin Echoes in the Presence of a Time-Dependent) Field Gradient. J. Chem. Phys. 1965; 42(1): 288-92). Today, single echo planar imaging (EPI) is the method of choice for most applications, unless very high spatial resolution or very small spatial distortion is required, in this case multi-shot (multi-shot) Method) (Pipe JG, Farthing VG, Forbes KP. Multishot diffusion-weighted FSE using PROPELLER MRI. Magn Reson Med. Jan; 2002 47(1): 42-52) is preferred. Diffusion imaging is susceptible to several artifacts (Le Bihan D, Poupon C, Amadon A, Lethimonnier F. Artifacts and pitfalls in diffusion MRI. J Magn Reson Imaging. Sep; 2006 24(3): 478-88). Some of these artifacts are related to the scanner hardware, such as distortion caused by eddy currents, and some artifacts are associated with the capture, such as the magnetization of the entire brain tissue when read using echo floor imaging (EPI). Geometric distortion caused by the difference in rate. Other artifacts are associated with subjects such as physiological movements (heart beats, breathing) and bulk motion. The uncorrected, slow overall motion left during the capture results in inconsistent image data. This causes edge artifacts and blurring in the image. In addition, rapid global motion can result in uneven signal attenuation artifacts (signal dropouts) in the diffuse weight image, which is caused by the additional phase term (Anderson AW, Gore JC. Analysis and Correction). Of motion artifacts in diffusion weighted imaging. Magn Reson Med. Sep; 1994 32(3): 379-87). Artifacts caused by physiological motion are usually small and can be mitigated by gating, however, when uncorrected, overall motion artifacts often result in images that are not available. While slow overall motion can be corrected retrospectively, immediate motion correction reduces motion-related effects and reduces the variance between volumes compared to retrospective motion correction. If the spatial resolution is anisotropic, as is often the case with clinical protocols, interpolation will also result in sub-optimal image quality. External motion tracking can also be used to correct motion online, for example using optical methods (Zaitsev M, Dold C, Sakas G, Hennig J, Speck O. Magnetic resonance imaging of freely moving objects: prospective real- Time motion correction using an external optical motion tracking system. Neuroimage.Jul 1;2006 31(3):1038-50;Dold C,Zaitsev M,Speck O,Firle EA,Hennig J,Sakas G.Advantages and limitations of prospective head Motion compensation for MRI using an optical motion tracking device. Acad Radiol.Sep;2006 13(9):1093-103;Zaremba AA,MacFarlane DL,Tseng WC,Stark AJ,Briggs RW,Gopinath KS,Cheshkov S,White KD. Optical head tracking for functional magnetic resonance imaging using structured light. J Opt Soc Am A Opt Image Sci Vis. Jul; 2008 25(7): 1551-7). However, these methods require additional external hardware and software and time-consuming calibration steps to be used.

Diffused images with signal attenuation artifacts cannot be easily retrospectively corrected. In order to avoid errors caused by these artifacts, these image data need to be excluded from the analysis, which will lead to a reduction in the signal-to-noise ratio (SNR) of the derived map and deviations in the tensor calculation. In order to solve these problems, a reacquisition method may be used in which a segment or image affected by motion artifacts is retrieved during the same capture period (Porter DA, Heidemamn RM. High resolution diffusion-weighted imaging using readout) -segmented echo-planar imaging, parallel imaging and a two-dimensional navigator-based reacquisition.Magn Reson Med.Aug;2009 62(2):468-75;Benner,T.;van der Kouwe,AJ.;Sorensen,AG .Diffusion imaging with prospective motion correction and reacquisition. Magn Reson Med. 2011 Jul; 66(1): 154-167).

Diffusion MRI is becoming more and more important for clinical and neuroscience research because it is capable of depicting the microstructural properties of white matter. In order to more accurately estimate the diffusion index to reflect the microstructure properties, recent advances in diffusion MRI techniques (such as diffusion spectral imaging (DSI) and high-angle resolution diffusion imaging) have obtained diffusion weights (DW) with multiple diffusion sensitivities and directions. image. Due to the use of strong diffusion gradients, these techniques are sensitive to head motion, which can cause image attenuation artifacts (signal loss) and errors in diffusion index calculations (Yendiki, Anastasia; Koldewyn, Kami; Kakunoori, Sita; Kanwisher, Nancy; Fischl, Bruce.: Spurious group differences due to head motion in a diffusion MRI study. NeuroImage Volume 88, March 2014, Pages 79-90). Accordingly, the present invention is directed to developing a post-processing algorithm for correcting artifacts of a diffuse image. Correction performance is tested by simulating signal loss in in vivo diffusion data and interpolation is used to correct for signal loss. This method successfully corrects missing images and restores accurate generalized diffusion anisotropy (GFA) calculations.

The invention provides a method for correcting a diffused image with artifacts, the method comprising: (a) providing a set of diffused images comprising the diffused images with artifacts; (b) calculating each image of the set of diffused images a first signal strength; (c) a plot of a slice number of the set of diffused images relative to the first signal intensity; (d) calculating the diffused image with artifacts by interpolation on the map a second signal strength; and (e) correcting the diffused image with artifacts based on the second signal strength.

In a preferred embodiment of the invention, the set of diffused images is a diffuse weight image, a diffuse spectrum image, a diffused tensor image, a high-angle resolution image, or a q-ball image.

In a preferred embodiment of the invention, the diffuse image with artifacts is caused by subject movement.

In a preferred embodiment of the invention, the interpolation method is a linear interpolation method, a polynomial interpolation method or a spline interpolation method. In a more preferred embodiment of the invention, the spline interpolation method is a B-spline interpolation method.

Figure 1 shows an example of an iterative interpolation. The steps of the iterative interpolation are shown in (B) to (D).

Figure 2 shows the following reference image (top row) and corrected image (salvaged image) (bottom row): (A) b = 310 s / mm ² , diffusion direction [0.0601, -0.9895, -0.1314], (B) b = 2760 s/mm ² , diffusion direction [0.3834, -0.5477, -0.7437], and (C) b = 4000 s/mm ² , diffusion direction [0.0419, -0.9022, 0.4292].

Figure 3 shows a representative generalized diffusion anisotropy (GFA) plot for a single slice from the reference (left), degraded (middle), and corrected (right) DSI data sets.

Figure 4 shows a plot of signal loss versus functional difference (FD) in a corrupted and corrected DSI data set.

Figure 5 shows two consecutive missing corrected original images (A), corrected images (B), and the difference between the original image and the corrected image (C). The top line shows the first one lost and the bottom line shows the second one missing.

The following examples are non-limiting and represent only a variety of aspects and features of the invention.

Imaging

DSI data was obtained on a 3T MRI system (TIM Trio, Siemens, Erlangen) with a 32 channel phased array coil. The DSI pulse sequence uses two refocusing balanced echo-diffusion echographic contrast sequences (Reese TG, Heid O, Weisskoff RM, Wedeen VJ. Reduction eddy-current-induced distortion in diffusion MRI using a twice- refocused spin echo. Magn Reson Med. 2003 Jan; 49(1): 177-82), TR/TE = 9600/130 ms, FOV = 200 x 200 mm ² , matrix size = 80 x 80, 56 slices, and the slice thickness is 2.5 mm. In 3D q space (|q| A total of 102 diffusion coding gradients (maximum diffusion sensitivity bmax=4000 s/mm ² ) are sampled at grid points in the hemisphere of 3.6 units (Li-Wei Kuo, Jyh-HorngChen, Van Jay Wedeen, and Wen-Yih Isaac Tseng. Optimization of diffusion spectrum imaging and q-ball imaging on clinical MRI system. NeuroImage 41 (2008) 7-18).

Signal loss simulation

Fifteen DSI data sets with no signal loss were selected for reference. These DSI data sets are used to simulate data sets that are corrupted by signal loss; the original DW images are randomly replaced with signal loss images. For each data set, 15 different losses were used, ie 10, 30, 50, 70, 90, 130, 170, 210, 250, 290, 340, 380, 420, 460 and 500 lost, creating 15 Analog data set. These losses are discretely lost. A similar approach is used to simulate continuous loss, such as two consecutive losses.

Loss correction

For each volumetric data corresponding to a particular diffusion code, we find that the signal strength along the z (slice) direction is a smooth curve. In this embodiment, the signal loss at any z position in the curve is recovered by using the least squares fit on the B-spline curve (David Eberly.: Least-Squares Fitting of Data with B-Spline Surfaces) .Geometric Tools 2005).

For discrete loss, both sides of the missing image are retained, so interpolation is performed directly to correct the missing image. For continuous loss, the information on both sides of the lost image is incomplete so that the lost image cannot be directly corrected. Therefore, iterative interpolation is used to correct for continuous loss. An example of an iterative interpolation method is as follows: When the consecutive lost images are in slice 27 and slice 28 (Fig. 1A), four steps are used to correct the consecutive lost images:

Step 1: Discard the missing information of slice 27 and slice 28, otherwise the missing information will produce an erroneous result.

Step 2: Use the information from slice 1 to slice 26 and slice 29 to slice 56 to interpolate the missing portion caused by the loss of slice 27 and slice 28 (the interpolated image is considered to be a complex of slice 27 and slice 28, ie slice 27.5) (Fig. 1B).

Step 3: The information of slice 1 to slice 26, slice 27.5, and slice 29 to slice 56 is used to interpolate the image of slice 27 (both sides of slice 27 are obtained, so the result of the interpolation is better) (Fig. 1C).

Step 4: Using the information of slice 1 to slice 27.5 and slice 29 to slice 56 to interpolate the image of slice 28, iterative interpolation is performed (Fig. 1D).

Loss correction verification

We use whole brain tract-based automatic analysis (Yu-Jen Chen, Yu-Chun Lo, Yung-Chin Hsu, Chun-Chieh Fan, Tzung-Jeng Hwang, Chih-Min Liu, Yi-Ling Chien, Ming H.Hsieh, Chen-Chung Liu, Hai-Gwo Hwu, Wen-Yih Isaac Tseng: Automatic Whole Brain Tract-Based Analysis Using Predefined Tracts in a Diffusion Spectrum Imaging Template and an Accurate Registration Strategy.Human Brain Mapping 36: 3441-3458 (2015)) A generalized diffusion anisotropy (GFA) profile of 76 white matter nerve bundles was obtained for each DSI data set. We tested the performance of this correction method by evaluating the functional difference (FD) between the GFA profile derived from the reference DSI dataset and the GFA profile derived from the impairment and correction dataset (Sylvain Gouttard, Casey B. Goodlett Marek Kubicki, and Guido Gerig: Measures for Validation of DTI Tractography. Proc SPIE Int Soc Opt Eng. 2012 February 23; 8314:83). Functional differences are defined as: Where f ( t _i , tb _j , s _k ) and f _ref ( t _i , tb _j , s _k ) are the GFA values measured in the t _i step of the nerve bundle tb _j of the subject s _k , respectively derived The self-corrupted/corrected DSI data set and the reference DSI data set, and l, m, n indicate the total number of steps of the nerve bundle, the total number of nerve bundles, and the total number of subjects, respectively.

Result

To prevent errors in GFA calculations caused by signal loss, the conventional approach is to discard data from the analysis. This method wastes a lot of data sets and may deviate from the research matrix. To solve this problem, we provide a way to correct DSI data sets that are corrupted by signal loss. In the simulation data set, we show that this method can successfully correct discrete missing images (Figure 2). As shown in Figure 2, we can see that the missing image is corrected to the image displayed in the bottom row, which looks very similar to the reference image in the top row. As shown in Figure 3, the corrupted GFA map is clearly different from the reference GFA map, and the corrected GFA map looks very similar to the reference GFA map. As shown in Fig. 4, the FD value of the damage increases sharply with the number of losses. In contrast, the corrected FD value increases slightly with the number of losses. The results of the GFA plot and FD confirmed that the correct GFA values were restored after correction (Figures 3 and 4). For continuous loss, this method also successfully corrected for two consecutive losses. As shown in Figure 5, we can see that the corrected image (B) looks very similar to the original image (A).

Claims (5)

 A method of correcting a diffused image having artifacts, the method comprising: (a) providing a set of diffused images comprising the diffuse image having artifacts; (b) calculating a first image of each of the set of diffused images Signal strength; (c) plotting the slice number of the set of diffused images relative to the first signal intensity; (d) calculating the second signal of the diffused image with artifacts by interpolation on the map And (e) correcting the diffused image with artifacts based on the second signal strength.    The method of claim 1, wherein the set of diffused images is a diffusion weight image, a diffused spectrum image, a diffusion tensor image, a high-angle resolution image, or a q-ball image.    The method of claim 1, wherein the diffuse image with artifacts is caused by movement of the subject.    The method of claim 1, wherein the interpolation method is a linear interpolation method, a polynomial interpolation method, or a spline interpolation method.    The method of claim 4, wherein the spline interpolation method is a B-spline interpolation method.