A new method to achieve superresolution using diffusion mri has been recently introduced 1. Application of sparsecoding superresolution to 16bit. Super resolution imaging sr is a class of techniques that enhance increase the resolution of an imaging system. Nedjatigilani et al, regularized superresolution for diffusion mri, in. Request pdf superresolution in magnetic resonance imaging. Brain mri superresolution using deep 3d convolutional. Diffusion weighted imaging dwi enables noninvasive investigation and characterization of the whitematter but suffers from a relatively poor resolution. Poot,1 ben jeurissen,2 yannick bastiaensen,2 jelle veraart,2 wim van hecke,2,3 paul m. A compressedsensing approach for super resolution reconstruction of diffusion mri lipeng ning, 1 kawin setsompop, 2 oleg michailovich, 3 nikos makris, 2 carlfredrik westin, 1 and yogesh rathi 1.
Our approach combines the twin concepts of compressed sensing cs and classical superresolution to reduce acquisition time while increasing spatial resolution. The obtained real data results demonstrate a first time proof of concept that gans can be useful in super resolution problem of diffusion mri for upscaling in the spatial dimension. It allows the mapping of the diffusion process of molecules, mainly water, in biological tissues, in vivo and noninvasively. Subpixel shifting in the inplane dimension was proposed to obtain multiple low resolution images to reconstruct the high resolution images 7. Increas36 ing image resolution through super resolution is a key to more accurate understanding of 37 the anatomy greenspan2008. Robust determination of the fibre orientation distribution in. Superresolution reconstruction for diffusionweighted images. We utilize generative adversarial networks gans to obtain a higher resolution diffusion mr image in the spatial dimension from lower resolution diffusion images. Examplebased single image superresolution sr has recently shown outcomes with high reconstruction performance. Advanced photonics journal of applied remote sensing. A joint compressedsensing and superresolution approach for. In this paper, we study deep 15 threedimensional convolutional neural networks for the superresolution of brain magnetic 16 resonance imaging data. Diffusion weighted images were independently and subjectively scored on le sion conspicuity, internal architecture definition, and overall image.
Diffusionweighted imaging dwi is a key technique for investigating white matter noninvasively. This technique, known as superresolution trackdensity imaging tdi, is able to increase the spatial resolu tion of the reconstructed images beyond the acquired mri resolutio n. For a complete list of gans in general computer vision, please visit reallyawesomegan. From quantitative measurement to invivo neuroanatomy. Superresolution reconstruction of mr image with a novel. Then, a generative adversarial network gan is trained with each dataset to generate super resolution mr images. The resolution of this reconstructed sr mri is improved in all directions and its information is increased comparing to the initial scans. Image quality transfer and applications in diffusion mri.
Robust determination of the fibre orientation distribution. Highresolution diffusionweighted imaging of the prostate. In this paper, we propose to combine the concepts of compressed sensing and superresolution to reconstruct very high resolution diffusion data. Recently, super resolution methods for diffusion mri capable of retrieving high resolution diffusion weighted images were proposed, yielding a resolution beyond the scanner hardware limitations. Superresolution trackdensity imaging studies of mouse brain. However, most rely on a specialised acquisition with partially over. Superresolution sr approaches were originally developed for the reconstruction of highresolution hr images from a set of lowresolution lr images in video sequences. Superresolution in diffusionweighted imaging ncbi nih. Magnetic resonance imaging mri is a noninvasive technique that is used in clinical applications such as diseases diagnosis and monitoring and treatment progress. Moreover, this approach can be used in conjunction with any diffusion model. Joint superresolution using only one anisotropic low. Diffusion mri is a magnetic resonance imaging mri method that produces in vivo images of biological tissues weighted with the local microstructural characteristics of water diffusion.
Multiscale brain mri superresolution using deep 3d. The multi image resolution diffusion mri data set is obtained from one long scanning session 19 days of a perfusion fixated vervet monkey brain using a 4. Modelbased superresolution of diffusion mri springerlink. Pdf revealing the hippocampal connectome through super. By using the postprocessing methods to gain spatial resolution based on diffusion mri fiber tracking, the superresolution tdi can reveal structures beyond the resolution of the acquired imaging voxel calamante et al. Standard dmri acquisitions facilitate a proper analysis e. Inference of multiple fiber orientations in high angular resolution diffusion imaging. High resolution t1w and t2w datasets with both normal and pathology multiple sclerosis from the publicly available brainweb database were used 15, 16. Diffusionweighted magnetic resonance imaging dwi or dwmri is the use of specific mri sequences as well as software that generates images from the resulting data that uses the diffusion of water molecules to generate contrast in mr images. Our approach combines the twin concepts of compressed sensing cs and classical super resolution to reduce acquisition time while increasing spatial resolution. A waveletbased superresolution method for multislice mri. Resolution reconstruction of diffusion parameters from. Despite the recent development of several superresolution fluorescence microscopic techniques, there are still few techniques that can be readily employed in. Image quality transfer and applications in diffusion mri sciencedirect.
In magnetic resonance imaging typical clinical settings, both low and high resolution images of different types are routinarily acquired. Superresolution reconstruction to increase the spatial resolution of diffusion weighted images from orthogonal anisotropic acquisitions. Using diffusion mri to discriminate areas of cortical grey matter. Superresolution reconstruction for diffusionweighted. Superresolved lowresolution images yielded left ventricular volumes comparable to those from fullresolution images p. Introduction 30 magnetic resonance imaging mri is a powerful imaging modality for in vivo brain 31 visualization with a typical image resolution of 1mm. This technique is able to increase the spatial resolution of the reconstructed images beyond the acquired mri resolution by incorporating information from wholebrain fibretracking results. Diffusionweighted images were independently and subjectively scored on le sion conspicuity, internal architecture definition, and overall image.
In optical sr the diffraction limit of systems is transcended, while in geometrical sr the resolution of digital imaging sensors is enhanced in some radar and sonar imaging applications e. However, classical interpolation techniques are not. Diffusion mri dmri has been widely adopted to analyse white matter microstructure, but scarcely used to distinguish grey matter regions because of the reduced anisotropy there. Super resolution sr aims to enhance the resolution 35 of an imaging system using single or multiple data acquisitions milanfar2010. Using diffusion mri to discriminate areas of cortical grey. Deep learning singleframe and multiframe superresolution. Parizel,3 and jan sijbers2 diffusion weighted magnetic resonance images are often. Revealing the hippocampal connectome through superresolution. This technique, known as superresolution trackdensity imaging tdi, is able to increase the spatial resolu tion of the reconstructed images beyond the acquired mri resolutio n by incorporating information contained.
Nov 25, 20 this work introduces a modelbased super resolution reconstruction srr technique for achieving high resolution diffusion weighted mri. A joint compressedsensing and superresolution approach for very. The superresolution tdi maps 19, 20 were generated from the wholebrain csd tractography maps 21, 22 by calculating the total number of tracks present in each element of a 0. A joint compressedsensing and superresolution approach. In many clinical applications such as neurosurgical planning and deep brain stimulation, it is.
To validate the proposed method, a synthetic dataset was used. Fiberbased methods reconstruct superresolution information on nerve. Poot,1 ben jeurissen,2 yannick bastiaensen,2 jelle veraart,2 wim van hecke,2,3. Purpose diffusion mri is hampered by long acquisition times, low spatial resolution, and a low signal.
The aim of this study was to superresolve in vivo human brain 3 t mri diffusion tractography data from an ultrahigh angular resolution dataset using track density imaging to. Several methods based on neural networks have successfully introduced techniques into sr problem. They acquired three 3d orthogonal volumes at a resolution of 1. Diffusion mri dmri can provide invaluable information about the structure of different tissue types in the brain. Superresolution for whole body diffusionweighted mri x. Bayesian superresolution in brain diffusion weighted. We tracked the wholebrain connections of white matter fibres from the hippocampus using ultrahigh angular resolution diffusion mri in both a single 1150direction dataset and a large normal. Or if more than two diffusion weighted images are used, linear regression can be used to. Recently, methods have been proposed to improve the trade. Super resolution reconstruction, diffusion mri, spherical. A joint compressedsensing and superresolution approach for very highresolution diffusion imaging.
Unlike the majority of existing super resolution algorithms, our proposed solution does not entail acquiring multiple scans from the same subject which renders it practical in clinical settings and applicable to legacy data. A curated list of awesome gan resources in medical imaging, inspired by the other awesome initiatives. Superresolution trackdensity imaging studies of mouse. Diffusion mri available for download and read online in other formats. Robust determination of the fibre orientation distribution in diffusion mri. Using diffusion mri to discriminate areas of cortical grey matter tharindu ganepolaa,b, zoltan nagyc, aurobrata ghoshb, theodore papadopouloe, daniel c. Sr techniques have also been applied to anatomical magnetic resonance imaging mri 4. High angular resolution diffusionweighted imaging hardi. Frontiers papez circuit observed by in vivo human brain. Super resolution for whole body diffusion weighted mri x. This technique is able to increase the spatial resolution of the reconstructed images beyond the acquired mri resolution by incorporating information from wholebrain. First, our work delves into the relevance of several factors in the.
If two diffusion weighted images is used different bvalues, then diffusivity can be calculated by. In our study, we first enlarged low resolution images using 5 commonly used super resolution algorithms and obtained differentially enlarged image datasets with complementary priors. Jan 01, 2016 purpose diffusion mri is hampered by long acquisition times, low spatial resolution, and a low signal. Diffusionweighted imaging dwi enables noninvasive investigation and characterization of the whitematter but suffers from a relatively poor resolution.
Nonnegativity constrained superresolved spherical deconvolution. Iqt can reconstruct high resolution images from low resolution input by learning a mapping. This may also preclude the use of advanced diffusion mri techniques, such as spherical deconvolution 373839, super resolution 40, 41, and the apparent fiber density 42. Of particular significance is the major eigenvector of the diffusion tensor, corresponding to the direction of fastest diffusion basser, 1995. The accuracy of further analysis such as brain morphometry can be. Resolution reconstruction srr was applied on diffusion weighted imaging in ref. To avoid the long scanning time, superresolution sr acquisition emerged as an effective technology been initially proposed for mri and then adapted into dwi soon. Superresolution imaging sr is a class of techniques that enhance increase the resolution of an imaging system. Superresolution methods 12 and noise reduction techniques, 14 may also be considered as acceleration schemes since they circumvent the need to collect multiple averages andor extra. Superresolution tech niques could have strong impacts on structural magnetic resonance imaging when focusing 14 on cortical surface or nescale structure analysis for instance.
Mri superresolution using selfsimilarity and image priors. However, as the image resolution is proportional to the scan. To avoid the long scanning time, super resolution sr acquisition emerged as an effective technology been initially proposed for mri and then adapted into dwi soon. The image superresolution sr technique offers an alternative approach to improve the spatial resolution. In multislice magnetic resonance imaging mri, the resolution in the slice direction is usually reduced to allow faster acquisition times and to reduce the amount of noise in each 2d slice. Most of the methods involve the use of prior models of the mri data as. Application of sparsecoding superresolution to 16bit dicom. The proposed method combines the twin concepts of compressed sensing and super resolution to model the diffusion signal at a given bvalue in a basis of spherical ridgelets with totalvariation tv regularization to account for signal correlation in neighboring voxels. Highresolution magnetic resonance imaging mri allows the visu. Dictionary based superresolution for diffusion mri core. We present an innovative framework for reconstructing highspatialresolution diffusion magnetic resonance imaging dmri from multiple lowresolution lr images. The obtained real data results demonstrate a first time proof of concept that gans can be useful in superresolution problem of diffusion mri for upscaling in the spatial dimension. This work introduces a modelbased superresolution reconstruction srr technique for achieving highresolution diffusionweighted mri.
The image super resolution sr technique offers an alternative approach to improve the spatial resolution. Superresolution techniques have been used, for example, for reconstructing an image from a few frames of a movie or from pictures taken by a moving satellite. Diffusionweighted mri is a key technique in studying the neural architecture and connectivity of the brain. Super resolution methods 12 and noise reduction techniques, 14 may also be considered as acceleration schemes since they circumvent the need to collect multiple averages andor extra diffusion directions. Diffusion mri spatial superresolution using generative. Serenod a department of cognitive, perceptual and brain sciences, ucl, london, uk b centre for medical image computing, department of computer science, ucl, london, uk c laboratory for social and neural systems. It can be utilized as imagingbased biomarkers for investigating several brain disorders such as alzheimers disease, schizophrenia, mild traumatic brain injury, etc. Acquisition time of mri data and 32 signaltonoise ratio are two parameters that drive the choice of an appropriate image reso 33 lution for a given study. Conference proceedings papers presentations journals. In this work we propose a superresolution reconstruction srr technique based on the acquisition of multiple anisotropic orthogonal dwi scans. In particular, we focus on a specific qspace sampling scheme known as high angular resolution diffusion imaging hardi which uses several diffusion measurements at a single bvalue shell 23, 24. Super resolution, sparsecoding super resolution scsr, mri, dicom, 16bit 1. Superresolution reconstruction of diffusion parameters from diffusionweighted images with different slice orientations.
Super resolution reconstruction of diffusion parameters from diffusion weighted images with different slice orientations. Larson 1departments of radiology and electrical engineering and computer science, northwestern university, chicago, il, united states, 2departments of electrical engineering and computer science, northwestern university, evanston, il, united states, 3department of. Revealing the hippocampal connectome through super. Diffusion weighted imaging dwi is a key technique for investigating white matter noninvasively. Larson 1departments of radiology and electrical engineering and computer science, northwestern university, chicago, il, united states, 2departments of electrical. To address this issue, a number of super resolution sr methods have been proposed to improve the resolution of 3d mri volumes.
Superresolution reconstruction to increase the spatial. In this research, we propose a novel super resolution sr technique for reconstructing high resolution hr mri using a sequence of orthogonal low resolution lr mri scans. Unlike the majority of existing superresolution algorithms, our proposed solution does not entail acquiring multiple scans from the same subject which renders it practical in clinical settings and applicable to legacy data. Diffusion weighted magnetic resonance images are often acquired with single shot multislice imaging sequences, because of their short scanning times and robustness to motion. The aim of this study was to super resolve in vivo human brain 3 t mri diffusion tractography data from an ultrahigh angular resolution dataset using track density imaging to compare it to ex. We present an innovative framework for reconstructing highspatial resolution diffusion magnetic resonance imaging dmri from multiple low resolution lr images. The recently proposed trackdensity imaging tdi technique was introduced as a means to achieve superresolution using diffusion mri. Introduction the spatial resolution of magnetic resonance imaging mri is a crucial factor related to image quality, and affects the identification of anatomical features in medical imaging. In this paper, we propose to combine the concepts of compressed sensing and super resolution to reconstruct very high resolution diffusion data. Techniques such as parallel mri 2, 3, propeller, and compressed sensing affect these relations at the acquisition level. Recently, superresolution methods for diffusion mri capable of retrieving highresolution diffusionweighted images were proposed, yielding a resolution beyond the scanner hardware limitations. Superresolution, sparsecoding superresolution scsr, mri, dicom, 16bit 1. Superresolution for multislice diffusion tensor imaging.
1247 144 387 1343 1572 1679 1248 683 1154 33 1585 608 1028 741 959 581 175 840 44 133 1632 1390 999 365 1220 147 1389 966 774 180 1068 917 1415 986 717 842 178 808 636