Xue Li¹, Kun Sun¹, Wei Liu², Robert Grimm³, Caixia Fu², and Fuhua Yan^1
1Department of Radiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China, ²Application Development, Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China, ³Application Predevelopment, Siemens Healthcare, Erlangen, Germany

This study compared the diagnostic performance of the ADC derived from the diffusion-weighted readout-segmented EPI DWI accelerated with SMS technique (SMS rs-EPI DWI) with that derived from the conventional rs-EPI DWI on breast lesions using whole-tumor histogram analysis. Our results showed that the SMS technique can increase the spatial resolution of the rs-EPI DWI sequence without prolonging the scan time. It can also improve the diagnostic performance of the derived ADC map based on whole-tumor histogram analysis in distinguishing benign and malignant breast lesions.

Óscar Peña-Nogales¹, Rodrigo de Luis-Garcia¹, and Santiago Aja-Fernández^1
1Laboratorio de Procesado de Imagen, Universidad de Valladolid, Valladolid, Spain

Estimation of Intravoxel Incoherent Motion (IVIM) parameter maps from a set of diffusion-weighted (DW) images acquired at multiple b-values usually suffers from low SNR, which may increase the variance of the estimated maps. Unfortunately, there is no consensus on the optimal b-values to maximize the noise performance of IVIM parameters. In this work, we determine the optimal b-values to maximize the performance of IVIM parameter mapping by using a Cramér-Rao Lower Bound approach under realistic noise assumptions. The reduction of the estimation variance on the IVIM parameters compared to state-of-the-art b-values suggests the utility of this approach to optimize DW-MRI.

Naoki Ohno¹, Tosiaki Miyati¹, Tetsuo Ogino², Yu Ueda², Yuki Koshino^1,3, Yudai Shogan^1,3, Toshifumi Gabata⁴, and Satoshi Kobayashi^1
1Faculty of Health Sciences, Institute of Medical, Pharmaceutical and Health Sciences, Kanazawa University, Kanazawa, Japan, ²Philips Japan, Tokyo, Japan, ³Radiology Division, Kanazawa University Hospital, Kanazawa, Japan, ⁴Department of Radiology, Kanazawa University Graduate School of Medical Sciences, Kanazawa, Japan

In this study, we compared diffusion parameters with intravoxel incoherent motion (IVIM) analysis of the brain between second-order motion-compensated (2^nd-MC) and conventional (non-MC) diffusion encoding schemes. Perfusion-related diffusion coefficient with non-MC was strongly affected by bulk motion in the pons which has the largest motion in the brain. By contrast, the 2^nd-MC diffusion gradients compensated the bulk motion-induced signal loss and improved the fitting accuracy of biexponential model. The 2^nd-MC diffusion encoding reduces the bulk motion effect on IVIM analysis of the brain, thereby improving the measurement accuracy.

Dan Wu¹, Dapeng Liu², Yi-Cheng Hsu³, Haotian Li¹, Yi Sun³, Qin Qin², and Yi Zhang^1
1Key Laboratory for Biomedical Engineering of Ministry of Education, Department of Biomedical Engineering, College of Biomedical Engineering & Instrument Science, Zhejiang University, Hangzhou, China, ²Johns Hopkins University School of Medicine, BALTIMORE, MD, United States, ³MR Collaboration, Siemens Healthcare Ltd., Shanghai, China

Oscillating gradient enables access of short diffusion times for time-dependent diffusion MRI (dMRI), but poses challenges for clinical use, including limited oscillating frequencies and b-values, low SNR, and relatively long scan times. This study proposes a 3D oscillating gradient prepared gradient spin-echo sequence (OGprep-GRASE) to improve the SNR and shorten the acquisition time for OG-dMRI. The proposed sequence reduced the scan time by a factor of 1.38 and increased the SNR by 1.74 times, compared with the existing 2D echo-planar imaging (EPI) approach, leading to improved diffusion tensor reconstruction. Diffusivity measurements showed similar time-dependency using the GRASE and EPI sequences.

Ting Yuan¹, Yan Sha², Zhongshuai Zhang³, Xilan Liu⁴, Xinpei Ye⁵, Yaru Sheng⁵, Kun Zhou⁶, and Caixia Fu^6
1Shanghai Insititute of Medical Imaging, Shanghai, China, ²Eye & ENT Hospital of Shanghai Medical School, Fudan University, Shanghai, China, ³Siemens Healthcare Ltd, Shanghai, China, ⁴Department of Radiology, Shanghai Ninth People’s Hospital, Shanghai JiaoTong University, Shanghai, China, ⁵Department of Radiology, Eye & ENT Hospital of Shanghai Medical School, Fudan University, Shanghai, China, ⁶Department of Digitalization, Siemens Shenzhen Magnetic Resonance Ltd., Shenzhen, China

This study investigated the role of RESOLVE and TGSE DWI sequences in the evaluation of optic neuritis and compared their image qualities qualitatively and quantitatively. We found that TGSE significantly improved the image quality for the evaluation of optic neuritis by reducing the susceptibility induced image distortion compared with RESOLVE. However, it appeared lower SNR and CNR than that of RESOLVE images.

Qi Zhang¹, Xiaoduo Yu¹, Jieying Zhang¹, Xinming Zhao¹, Han Ouyang¹, Hongmei Zhang¹, Qinglei Shi², Xiang Feng², and Xiaoye Wang^2
1Department of Imaging Diagnosis, National Cancer Center/Cancer Hospital, Chinese Academy of Medical Science and Peking Union Medical College, Beijing, China, ²MR Scientific Marketing, Diagnostic Imaging, Siemens Healthcare Ltd, Beijing, China

This study proposed an automatic image quality evaluation method using no-reference image quality metrics of structural similarity index (SSIM), blind/referenceless image spatial quality evaluator (BRISQUE), perception based Image quality evaluator (PIQE), SNR Wavelet and Contrast. The SNR Wavelet was calculated by the quotient of the image before wavelet filtering and the difference between the image before wavelet filtering and the image after wavelet filtering. The contrast was calculated using the average signal difference of three to five gray bars in the middle position. This study showed the automatic no-reference image quality metrics have potentials in future application of evaluating the image quality of uterine malignancy DWI at 3T with a higher efficiency.

Huiting Zhang¹, Ankang Gao², Shaoyu Wang¹, Yang Song³, Jingliang Cheng², Guang Yang³, and Xu Yan^1
1MR Scientific Marketing, Siemens Healthcare, Shanghai, China, ²The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China, ³Shanghai Key Laboratory of Magnetic Resonance, East China Normal Univeristy, Shanghai, China

This study aimed to evaluate the performance of three b-value sampling schemes in calculating multiple diffusion models, including the DTI, DKI, NODDI and the newly proposed mean apparent propagator (MAP)-MRI models. The three schemes includes the conventional diffusion spectrum imaging (DSI) acquisition scheme based on Cartesian grid sampling in q-space, multi-shell sampling with the same (MDDW) or different (FREE) gradient directions in each shell. Each scheme supports the estimation of all the four models. The results showed that generally three schemes generated very similar parameters, and could be all used in future studies.  

Carl-Fredrik Westin^1,2 and Filip Szczepankiewicz^1,2
1Harvard Medical School, Boston, MA, United States, ²Brigham and Women's Hospital, Boston, MA, United States

In this work we propose novel sampling schemes for diffusion MRI required for encoding with objects with more than one directional axis. The presented solution is general and suitable for both axisymmetric and non-axisymmetric encoding schemes. An important feature of the presented sampling schemes is that they can be interleaved and be complimentary. This means that any combination of them can be used to define an isotropic sampling scheme with number of samples: (15, 30, 45, 60, 75, 90). 

Jan Morez¹, Jan Sijbers¹, and Ben Jeurissen^1
1imec-Vision Lab, Dept. Physics, University of Antwerp, Antwerp, Belgium

Multi-tissue constrained spherical deconvolution of multi-shell diffusion weighted MRI data estimates the white matter fiber orientation distribution function, together with the densities of gray matter  and cerebrospinal fluid. In this work, we propose a 5-minute scanning protocol that allows a more precise estimation of WM and GM densities, while maintaining a high angular resolution. 

Kayoko Abe¹, Kazufumi Suzuki¹, Masami Yoneyama², and Shuji Sakai^1
1Diagnostic Imaging and Nuclear Medicine, Tokyo Women's Medical University, Tokyo, Japan, ²Philips Japan, Tokyo, Japan

High b-value single-shot echo planar diffusion-weighted imaging (EPI-DWI) has been expected to provide more detail information about brain structure and diseases. However, higher b-value causes lower image quality due to an increase in noise-like artifacts. Compressed SENSE (C-SENSE), which is a combination of compressed sensing and parallel imaging technique: SENSE is an accelerating scan technique, which includes noise reduction methods. In this study, we revealed that EPI-DWI images with C-SENSE using high b-values (b:1000, 2000, 3000, 4000, 5000 s/mm2) showed higher SNR and ADC values than EPI-DWI images with SENSE. 

Grant Kaijuin Yang^1,2, Ek Tsoon Tan³, Eric Fiveland⁴, Thomas Foo⁴, and Jennifer McNab^2
1Electrical Engineering, Stanford University, Stanford, CA, United States, ²Radiology, Stanford University, Stanford, CA, United States, ³Hospital for Special Surgery in Manhattan, New York, NY, United States, ⁴GE Global Research, Niskayuna, NY, United States

In this work, q-space trajectory imaging was implemented on a whole-body system equipped with a high-performance head-only gradient system capable of 200 mT/m maximum gradient amplitude and 500 T/m/s slew rate. The improved gradient performance enabled the acquisition of q-space trajectory imaging with sub-millimeter in-plane resolution and reduced voxel volume compared to previously published work, while simultaneously improving head coverage, and reducing susceptibility induced image distortions.

Rita G. Nunes¹, Keerthi Sravan Ravi², Sairam Geethanath², and J. Thomas Vaughan Jr^2
1Instituto Superior Técnico, Lisbon, Portugal, ²Columbia University Magnetic Resonance Research Center, New York, NY, New York, NY, United States

Diffusion-weighted imaging is an essential sequence for many clinical applications. While the post-processing tools for diffusion are widely available, vendor-neutral, open-source acquisition implementations have not been shared for research purposes.   We develop a cross-vendor, open source package of a multi-slice single-shot spin echo-planar imaging based diffusion pulse sequence, capable of multiple b values and directions. We demonstrate this on an in vitro phantom, measuring plausible Apparent Diffusion Coefficient values and in vivo human brain data, obtaining good quality Fractional Anisotropy and Mean Diffusivity maps. We process our data with freely available post-processing tools to generate quantitative diffusion maps.

Neville D Gai¹ and John A Butman^1
1National Institutes of Health, Bethesda, MD, United States

While most brain imaging sequences now favor their 3D counterparts, diffusion imaging is an exception. This is due to large diffusion gradients resulting in increased sensitivity to motion exhibited by 3D acquisition. Prior schemes have used limited brain coverage and/or triggering or acquired multiple 3D slabs along with modified reconstruction schemes. The modified sequence used here employs first-order motion compensated diffusion gradients in addition to real-time alignment to acquire whole brain 3D-DWI images as a single slab. Relatively shorter TE (using enhanced gradients) and TR along with other modifications result in faster, reduced artifact diffusion images while providing higher SNR.

Qiuyun Fan¹, Thomas Witzel¹, Slimane Tounekti¹, Qiyuan Tian¹, Chanon Ngamsombat¹, Maya Polackal¹, Aapo Nummenmaa¹, and Susie Huang^1
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, MA, United States

We report the acquisition of whole brain, 2-mm isotropic resolution DDE data in a healthy volunteer using an orientationally invariant sampling scheme and quantify the mean DDE signal intensity across cortical regions as a measure of diffusion restriction within different cortices. Higher mean signal intensities were observed in the cerebellum and limbic cortices, which are thought to reflect a higher degree of restriction in the tissue microstructural environment and may correspond to densely packed, small granule and pyramidal cells known to be present in these regions.

Jianbo Cao¹, Stephen Pickup¹, Hanwen Yang¹, Victor Castillo¹, Cynthia Clendenin^2,3, Peter O’Dwyer^2,3, Mark Rosen^1,3, and Rong Zhou^1,3
1Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, ²Pancreatic Cancer Research Center, University of Pennsylvania, Philadelphia, PA, United States, ³Abramson Cancer Center, University of Pennsylvania, Philadelphia, PA, United States

Diffusion weighted (DW)-MRI is sensitive to tumor microenvironment (TME) thus useful for assessing pancreatic cancer responses to stroma-directed drugs, as they change TME by degradation or reduction of extracellular matrix. Motion-sensitive location of pancreatic tumor and fast respiration rate of mice impose a big challenge for quantitative DW-MRI. We compared radial k-space and echo-planar imaging based DW protocol for their accuracy and test-retest reproducibility. EPI-DW consistently underestimates water ADC value at 37C (reference to literature) where radial-DW does not. Better test-retest producibility measured by within-subject CV is obtained with radial-DW compared to EPI-DW.

Benjamin C Tendler¹, Sean Foxley², Moises Hernandez-Fernandez³, Michiel Cottaar¹, Olaf Ansorge⁴, Saad Jbabdi¹, and Karla Miller^1
1Wellcome Centre for Integrative Neuroimaging, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, ²Department of Radiology, University of Chicago, Chicago, IL, United States, ³Centre for Biomedical Image Computing and Analytics, University of Pennsylvania, Philadelphia, PA, United States, ⁴Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom

Diffusion-weighted steady-state free precession (DW-SSFP) generates high SNR diffusivity estimates in whole, post-mortem human brains. Improved estimates at 7T has motivated its use at ultra-high field. However, the DW-SSFP signal has a strong dependence on flip angle. This translates into both variable signal amplitude and diffusion contrast. At 7T, transmit-($$$B_{1}^{+}$$$) inhomogeneity leads to $$$B_{1}^{+}$$$-dependent SNR and ADC estimates. Previous work corrected for $$$B_{1}^{+}$$$-inhomogeneity by acquiring DW-SSFP datasets at two flip angles. Here, this approach is extended, utilising the full Buxton model of DW-SSFP to model non-Gaussian diffusion. A noise-floor correction and signal weighting are also incorporated to improve diffusivity estimates.

Yuan Zheng¹, Tao Feng¹, Sirui Li², Wenbo Sun², Qing Wei³, Samo Lasic⁴, Danielle van Westen⁵, Karin Bryskhe⁴, Daniel Topgaard^4,5, and Haibo Xu^2
1UIH America, Houston, TX, United States, ²Zhongnan Hospital of Wuhan University, Wuhan, China, ³United Imaging Healthcare, Shanghai, China, ⁴Random Walk Imaging, Lund, Sweden, ⁵Lund University, Lund, Sweden

Multidimensional diffusion MRI (dMRI) is a powerful tool that even in its simplest form provides more detailed microstructural information than conventional dMRI, such as microscopic anisotropy (µFA) unconfounded by orientation dispersion. However, it requires multiple diffusion encoding modes (usually directional and isotropic encodings) and, for the more advanced versions, prolonged scan and post-processing times. We proposed using convolutional neural networks (CNN) to accelerate multidimensional dMRI data acquisition and analysis, and have demonstrated that satisfactory µFA maps can be generated in real-time with only 50% of the encodings, which might help to better adapt multidimensional dMRI to clinical practices.

Gerhard S Drenthen^1,2, Walter H Backes¹, and Jacobus FA Jansen^1,2
1Department of Radiology and Nuclear Medicine, Maastricht University Medical Center, Maastricht, Netherlands, ²Department of Electrical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands

In this study we aim to accelerate the acquisition time of myelin-water imaging by acquiring fewer slices and applying machine learning to extract myelin-specific information from anatomical (T1w and T2w) and diffusion-weighted imaging (DWI), which are commonly available in many clinical research studies. It is shown that with a 6-fold acceleration (from 7:30min to 1:15min) the myelin content can be reconstructed using neural networks with an agreement to the ground-truth that is comparable to the reproducibility of the scan itself.

Lisha Yuan¹, Yi-Cheng Hsu², Dan Wu¹, Hongjian He¹, and Jianhui Zhong^1,3
1Center for Brain Imaging Science and Technology, Department of Biomedical Engineering, Key Laboratory for Biomedical Engineering of Ministry of Education, Zhejiang University, Hangzhou, China, ²MR Collaboration, Siemens Healthcare Ltd., Shanghai, China, ³Department of Imaging Sciences, University of Rochester, Rochester, NY, United States

Compared to traditional echo-planar imaging (EPI)-based schemes, spatiotemporal encoding (SPEN) is largely insensitive to magnetic field and chemical shift heterogeneities. However, excitation gradient has different effects for each position, thus the interaction between imaging and diffusion gradients introduces spatial-dependent diffusion weightings along the SPEN axis. A new method named polarities average mode (PAM) was proposed to obtain accurate apparent diffusion coefficient (ADC) map, with two acquisitions of different polarities between excitation and diffusion gradients. Simulation, phantom, and human experiments were designed to assess method performance. The proposed method enables SPEN to obtain ADC maps easily and accurately.

Fei Wang¹, Mengxiao Liu², and Juan Zhu^1
1Department of MRI,AnQing Municipal Hospital, Anqing, China, ²MR scientific Marketing, Diagnostic Imaging, Siemens Healthcare Ltd, Shanghai, China

To evaluate the feasibility of applying simultaneous multi-slice (SMS) single-shot echo planar imaging (EPI) to accelerate MR diffusion imaging for breast carcinoma, fibroadenoma of breast and normal breast. SMS sequences with double and triple acceleration were compared with conventional DWI sequences, respectively. These SMS DWI sequences were compared to conventional DWI in terms of image quality parameters (5-point Likert scale) and SNR, ADC measurements. Comparing with conventional EPI-DWI, the SMS markedly reduces the diffusion scan time and the image SNR still shows a good quality. Thus, SMS technique is recommended for DWI of the MR breast examinations.

R. Allen Waggoner¹, Thorsten Feiweier², and Keiji Tanaka^1
1Laboratory for Cognitive Brain Mapping, RIKEN - Center for Brain Science, Wako-shi, Japan, ²Siemens Healthcare GmbH, Erlangen, Germany

On clinical scanners, high b-value diffusion studies using SE-EPI suffer from the need for long TEs, which leads to signal loss due to T2 decay.  Stimulated-Echo EPI permits high b-values together with short TEs on clinical scanners.  We demonstrate that tractograms obtained from high b-value STE-EPI images are clean even in regions where tractograms from SE-EPI images with the same b-values break down.

Loxlan W Kasa¹, Terry Peters², Roy AM Haast³, and Ali R Khan^4
1School of Biomedical Engineering, Imaging Research Laboratories, Robarts Research Institute, Western University, LONDON, ON, Canada, ²Imaging Research Laboratories, Robarts Research Institute, School of Biomedical Engineering,,Department of Medical Biophysics,Departments of Medical Imaging, Western University, London, ON, Canada, ³Imaging Research Laboratories, Robarts Research Institute, Western University, London, ON, Canada, ⁴Imaging Research Laboratories, Robarts Research Institute, School of Biomedical Engineering, Department of Medical Biophysics, Western University, London, ON, Canada

Diffusion kurtosis imaging (DKI), an extension to diffusion tensor imaging (DTI), aims to improve quantification of the hindered/restricted diffusion pattern due to microstructural complexity in the brain. But in order to capture the non-Gaussian diffusion behaviour of water molecules in biological tissues, stronger gradients larger than those employed in standard diffusion weighted imaging (DWI) are required. Here, we explored the test-retest reliability of DKI derived metrics with respect to different gradient strength in a high spatial resolution dataset. It was observed that DKI precision was comparable between b-value=1000, 2000, 3000 s/mm² and b-value=1000 & 3000 s/mm² dataset.

Qi Zhang¹, Jieying Zhang¹, Xiaoduo Yu¹, Han Ouyang¹, Xinming Zhao¹, Hongmei Zhang¹, Qinglei Shi², Xiang Feng², and Xiaoye Wang^2
1Department of Imaging Diagnosis, National Cancer Center/Cancer Hospital, Chinese Academy of Medical Science and Peking Union Medical College, Beijing, China, ²MR Scientific Marketing, Diagnostic Imaging, Siemens Healthcare Ltd, Beijing, China

In the evaluation of uterine malignancy, conventional DWI based on single-shot echo-planar imaging (ss-EPI) is prone to imaging artifacts, including susceptibility artifacts from gas, imaging blurring, which limit its diagnostic value, especially in detecting and staging uterine malignancy. The purpose of this study is to compare the detection of uterine malignancy and image quality among DWI based on integrated slice-specific dynamic shimming (iShim), readout segmentation of long variable echo trains (RESOLVE) and ss-EPI sequence. Our results indicated that iShim DWI showed better image quality than ss-EPI and RESOLVE DWI in the terms of subjective image scores and objective quantitative metrics.

Min-xiong Zhou¹, Huiting Zhang², Yang Song³, Guang Yang³, and Xu Yan^2
1Shanghai University of Medicine & Health Sciences, Shanghai, China, ²MR Scientific Marketing, Siemens Healthcare, Shanghai, China, ³Shanghai Key Laboratory of Magnetic Resonance, East China Normal Univeristy, Shanghai, China

Advanced diffusion models such as NODDI, MAP-MRI are of high interests in brain research, but suffer from long acquisition time. Advanced under-sampling scheme were reported in previous studies for acceleration but are not commercially available. This study evaluates a simple and commercial available under-sampling scheme using the symmetric property of q-space, which could accelerate the acquisition by 2 fold. Results showed that it did not significant sacrifice the accuracy of quantitative maps. In addition, a symmetrically data copy step is needed to improve the estimation accuracy for both MAP-MRI and NODDI models.

Melina Hosseiny¹, KyungHyun Sung¹, Teeravut Tubtawee¹, Voraparee Suvannarerg¹, Shabnam Mortazavi¹, Soheil Kooraki¹, Saurab Gupta¹, Afshin Azadikhah¹, Justin Ching¹, Ely R Felker¹, David Lu¹, and Steven S Raman^1
1Abdominal Radiology, David Geffen School of Medicine at UCLA, Los Angeles, CA, United States

Single-Shot Echo-Planar Imaging (ssEPI)is highly susceptible to T2* blurring and geometrical distortion. This study aimed to compare the image quality between ssEPI and Readout-Segment Echo-Planar Imaging (rsEPI) for acquiring prostate DWI on 162 patients. Geometrical distortion was ranked on ssEPI and rsEPI using a five-point scale. The geometrical distortion was significantly less observed in rsEPI compared to ssEPI (P<0.01).  Geometrical distortion scores of three and higher were observed in 30 individuals in ssEPI, with all having scores < 3 on rsEPI. In conclusion, using rsEPI for DWI acquisition may augment or replace ssEPI on 3T prostate mpMRI.

Abdulaziz Alshehri^1,2, Oun Al-iedani^1,2, Jameen Arm^1,2, Neda Gholizadeh¹, Rodney Lea³, Jeannette Lechner-Scott^3,4,5, and Saadallah Ramadan^1,2
1School of Health Sciences, University of Newcastle, Newcastle, Australia, ²Imaging center, Hunter Medical Research Institute, Newcastle, Australia, ³Hunter Medical Research Institute, Newcastle, Australia, ⁴School of Medicine and Public Health, University of Newcastle, Newcastle, Australia, ⁵Department of Neurology, John Hunter Hospital, Newcastle, Australia

This study aims to evaluate and compare DTI parameters in relapsing-remitting MS patients with age and sex-matched healthy controls, and to correlate these DTI metrics with clinical symptoms and brain volumetric measures. As a result, There was a statistically significant increase in most of DTI parameters for RRMS patients compared with healthy controls. FA correlated positively with clinical parameters like EDSS and cognitive assessment. Both MD and RD correlated negatively with cognition parameters and positively with EDSS. Quantitative DTI parameters not only differentiate between RRMS patients and HCs, but are also associated with disability and mental health of RRMS.

Malte Steinhoff¹, Alfred Mertins¹, and Peter Börnert^2,3
1Institute for Signal Processing, University of Luebeck, Luebeck, Germany, ²Philips Research Europe, Hamburg, Germany, ³Department of Radiology, LUMC, Leiden, Netherlands

We propose an extra-navigated SENSE-based multi-shot DWI reconstruction algorithm that comprises navigator-based phase and rigid in-plane motion corrections at fast reconstruction times. Furthermore, this approach exploits the low-resolution navigator signal to perform diffusion contrast corrections explicitly within the model. The extra-navigated method is compared in-vivo to a self-navigated reference algorithm. The extra-navigated motion estimation from low-resolution navigator data yields decent reconstructions which perfectly coincide with self-navigated results. Moreover, extra-navigation allows for fast reconstruction at the cost of lower scan efficiency and appears to be more robust for strong motion corruption and high segmentations.

Xinzeng Wang¹, Daniel Litwiller², Ali Ersoz³, Marc Lebel⁴, Sagar Mandava⁵, Lloyd Estkowski³, Arnaud Guidon⁶, Ann Shimakawa⁷, and Ersin Bayram^1
1Global MR Applications & Workflow, GE Healthcare, Houston, TX, United States, ²Global MR Applications & Workflow, GE Healthcare, New York, NY, United States, ³Global MR Applications & Workflow, GE Healthcare, Waukesha, WI, United States, ⁴Global MR Applications & Workflow, GE Healthcare, Calgary, AB, Canada, ⁵Global MR Applications & Workflow, GE Healthcare, Tucson, AZ, United States, ⁶Global MR Applications & Workflow, GE Healthcare, Boston, MA, United States, ⁷Global MR Applications & Workflow, GE Healthcare, Menlo Park, CA, United States

PROPELLER DWI, a FSE based DWI method, is increasingly used to reduce susceptibility artifacts and motion artifacts. Multi-shot Echo-Planar diffusion method also can reduce susceptibility artifacts, but PROPELLER DWI shows better image quality where susceptibility artifacts are most problematic, such as in skull base, head-neck and pelvis. However, the acquisition time is often longer compared to ms-DW-EPI, therefore SNR is usually compromised to reduce acquisition time. In this work, we evaluated a deep-learning based reconstruction method (DL Recon PROP) intended to improve image quality and ADC measurements by reducing the noise and artifacts without increasing acquisition time.

Ye Wu¹, Yoonmi Hong¹, Weili Lin¹, Pew-Thian Yap¹, and the UNC/UMN Baby Connectome Project Consortium^1
1Department of Radiology and BRIC, University of North Carolina, Chapel Hill, Chapel Hill, NC, United States

We propose a rapid and automated method to rectify incorrect gradient orientations resulting from inconsistencies in coordinate frame conventions across scanners, file formats, and processing tools. Using these incorrect gradient orientations will invalidate subsequent derived quantities that are dependent on local orientation information, particularly tractography. Our approach to correcting the gradient orientations is based on maximizing an orientation continuity index that is computed directly from the diffusion-weighted images without the need for model fitting.

Zhangxuan Hu¹, Zhe Zhang², Yishi Wang³, Yajing Zhang⁴, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, ²China National Research Center for Neurological Diseases, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, ³Philips Healthcare, Beijing, China, ⁴MR Clinical Science, Philips Healthcare (Suzhou), Suzhou, China

Single-shot EPI (SS-EPI) is widely used for diffusion-weighted imaging (DWI), but suffers from susceptibility-induced distortion and T2* blurring, which limit its resolution and ability to detect detailed structures. Parallel imaging and multi-shot techniques can be used to improve the resolution and reduce image distortion. However, these techniques have their own drawbacks, such as limited achievable acceleration factors or prolonged acquisition time. In this study, a deep-learning based method is proposed to achieve high-resolution distortion-free DWI using SS-EPI thus to improve the acquisition efficiency and clinical applicability.

Malte Steinhoff¹, Itamar Ronen², Andrew Webb², Alfred Mertins¹, and Peter Börnert^2,3
1Institute for Signal Processing, University of Luebeck, Luebeck, Germany, ²Department of Radiology, LUMC, Leiden, Netherlands, ³Philips Research Europe, Hamburg, Germany

Segmented diffusion imaging with iterative motion corrected reconstruction (SEDIMENT) is studied at 7T for self-navigated multi-shot DWI reconstruction including rigid in-plane motion correction. Motion-corrupted datasets contain intra-shot motion corrupted data with imperfect diffusion-sensitizing gradient reversal, which have to be identified and removed. The iterative SEDIMENT framework is evaluated in-vivo in conjunction with tailored data rejection strategies to detect corrupted shot datasets and generally improve convergence. The proposed algorithm provides high-quality multi-shot DWI and DTI reconstructions in the presence of gross motion allowing for efficient navigator-free DWI acquisition schemes.

Shreyas Fadnavis¹, Hu Cheng², and Eleftherios Garyfallidis^1
1Intelligent Systems Engineering, Indiana University Bloomington, Bloomington, IN, United States, ²Psychological and Brain Sciences, Indiana University Bloomington, Bloomington, IN, United States

NoiseFactors is a probabilistic graphical model to suppress and remove additive noise in a single DWI image. It mitigates the issues caused by noise by preserving correlations in the signal components and suppressing the uncorrelated noise within local neighbourhoods. We solve the low-rank approximation problem by learning a best m-component approximation of a factor model. To do so we also introduce a novel flipped bi-crossvalidation to estimate the factor model. It outperforms the state-of-the-art PCA based methods such as Marchenko-Pastur PCA and Local PCA. The proposed method for denoising will be made available with an open-source implementation in DIPY.

Yiman Huang¹, Xinlin Zhang¹, Hua Guo², Huijun Chen², Di Guo³, and Xiaobo Qu^1
1Department of Electronic Science, Xiamen University, Xiamen, China, ²School of Medicine, Tsinghua University, Beijing, China, ³School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, China

Multi-shot DWI improves the image resolution, while it induces phase variation at the same time. We introduce a smooth phase constraint of each shot image into multi-shot DWI reconstruction procedures by imposing the low-rankness of Hankel matrix constructed from the k-space data. The image is further improved with a partial sum of singular values in low-rank matrix reconstruction. Results on brain imaging data show that the proposed method outperforms the state-of-the-art methods in terms of artifacts removal and is compatible to partial Fourier sampling in accelerated DWI.

Banafshe Shafieizargar¹, Ben Jeurissen¹, Arnold Jan den Dekker¹, and Jan Sijbers^1
1imec-Vision Lab, Department of Physics, University of Antwerp, Antwerp, Belgium

To address the issue of phase induced artifacts in multi-shot diffusion weighted imaging, we propose a model-based framework which enables the joint estimation of diffusion and phase parameters directly from the multi-shot k-q-space. In a simulation study, we show that using this framework, diffusion parameters can be estimated more accurately and precisely than with the conventional method (image reconstruction followed by voxel-wise model fitting) that ignores phase differences.

Simin Liu¹, Yuhui Xiong^1,2, Erpeng Dai^1,3, Jieying Zhang¹, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, ²Neusoft Medical Systems Co., Ltd., Shanghai, China, ³Department of Radiology, Stanford University, Stanford, CA, United States

In EPI-based diffusion imaging, the geometric distortions become more severe with increased image resolution. Various post-processing methods have been proposed to correct for distortions, such as the top-up method and the field-mapping method. Nonetheless, for 3D isotropic high-resolution diffusion imaging, distortion correction becomes more challenging due to decreased SNR. In this study, we applied a modified distortion correction method, which was previously proposed for 2D imaging, to isotropic high-resolution diffusion imaging using 3D simultaneous multi-slab (SMSlab) acquisition. The modified distortion correction method performed well in both phantom and in vivo experiments. It also outperformed the conventional top-up and field-mapping methods.

Gregory R. Lee^1,2
1Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Radiology, University of Cincinnati College of Medicine, Cincinnati, OH, United States

In this work it is demonstrated that computationally efficient deniosing can be done using thresholding of wavelet coefficients without sacrificing image quality relative to state-of-the-art patch-based methods. This is achieved by combining use of a redundant, directional tight wavelet frame with the Karhunen-Loeve transform along the "directions" dimension. An efficient GPU implementation of the algorithm required less than 30 seconds to process even relatively large DTI datasets (e.g. 96x96x60x203). The proposed approach should find use in SNR-challenged acquisitions such high resolution DTI, DKI and DSI.

Ye Wu¹, Yoonmi Hong¹, Weili Lin¹, Pew-Thian Yap¹, and the UNC/UMN Baby Connectome Project Consortium^1
1Department of Radiology and BRIC, University of North Carolina, Chapel Hill, Chapel Hill, NC, United States

Automated brain extraction using diffusion MRI is challenging owing to the low spatial resolution. Unremoved residual non-brain voxels characteristically manifest as voxels with high fractional anisotropy (FA). In this abstract, we introduce a fast and robust method to identify non-brain voxels for clean extraction of the brain using diffusion MRI. We show that our method is effective for both adult and infant data.

Xinyu Ye¹, Guangqi Li¹, Yuan Lian¹, Yishi Wang², and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, ²Philips Healthcare, Beijing, China

Recently, a distortion- and blurring-free acquisition method called PSF-EPI has been used in DWI. However, when field inhomogeneity is severe, DW images may become noisy and more shots are needed to get the reliable results. In this work, we introduce a low-rank based reconstruction method using signal correlation along the k_y-encoding dimension in PSF-EPI to improve image quality and reduce needed shot number to shorten scan time. High-resolution in-vivo data were used to test the performance of the proposed method. The results show that the quality of the images is improved.

Fasil Gadjimuradov^1,2, Seung Su Yoon^1,2, Thomas Benkert², Marcel Dominik Nickel², Karl Engelhard³, and Andreas Maier^1
1Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ²Magnetic Resonance, Siemens Healthcare GmbH, Erlangen, Germany, ³Department of Radiology, Martha-Maria Hospital, Nürnberg, Germany

Partial Fourier (PF) acquisition schemes are a common way to increase the inherently low signal-to-noise ratio in diffusion-weighted (DW) images. The naïve solution of zero-filling k-space results in visible blurring and Gibbs ringing. Based on the circumstance that traditional methods such as homodyne reconstruction or POCS often fail to remove blurring and ringing without introducing new artifacts, this work aims to use a Convolutional Neural Network for robust PF reconstruction in prostate DWI. We show that our data-driven approach, which efficiently uses correlations across different b-values, outperforms traditional methods in terms of quantitative measures and visual impression of the images.

Xin Tang¹, Jun Xie², Guobin Li², Meng Jiang³, and Chenxi Hu^4
1Department of Medical Information Engineering, Wuyuzhang honors college, Sichuan University, Chengdu, China, ²United Imaging Healthcare Co., Ltd, Shanghai, China, ³Department of Cardiology, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China, ⁴Institute of Medical Imaging Technology, School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China

A novel method was proposed to accelerate apparent-diffusion-coefficient(ADC) mapping to shorten the EPI echo-train and/or improve resolution. The method was based on SUPER--a Cartesian k-space undersampling strategy for parametric mapping acceleration—and adapted to account for the nonlinear phase variation in diffusion-weighted imaging at different b-values. In healthy subjects, the phase-corrected SUPER(R=2) and SUPER-SENSE(combining SUPER and parallel imaging, R=4) demonstrated similar image quality, reasonable noise amplification, and similar reconstruction time compared with the non-acceleration gold standard, despite a 2-fold and 4-fold reduction of reconstruction data. This suggests that SUPER is a practical and accurate approach for accelerating ADC mapping.

Dariya Malyarenko¹ and Thomas L Chenevert^1
1University of Michigan, Ann Arbor, MI, United States

Quantitative tissue diffusion parameters derived from diffusion weighted imaging (DWI) models hold promise for diagnostic and prognostic clinical oncology applications. System-dependent spatial DW bias due to gradient nonlinearity (GNL) is known confounding factor for quantitative DWI metrics. Improved accuracy and multiplatform reproducibility was previously demonstrated for mono-exponential apparent diffusion coefficient with correction for platform-dependent GNL bias (GNC). Complex tumor microenvironment often exhibits multi-exponential diffusion described by isotropic kurtosis model. This study proposes analytical extension and demonstrates empirical confirmation for GNC of parametric maps derived from diffusion kurtosis model.

Liyuan Liang¹, Mei-Lan Chu², Nan-Kuei Chen^3,4, Shihui Chen¹, and Hing-Chiu Chang^1
1Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong, China, ²Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, ³Department of Biomedical Engineering, University of Arizona, Tucson, AZ, United States, ⁴Brain Imaging and Analysis Center, Duke University Medical Center, Durham, NC, United States

Recently, a self-navigated interleaved block-segmented EPI (iblocks-EPI) has been proposed to acquire DTI data with high spatial resolution and less geometric diction. In addition, the oversampling of central k-space of iblock-EPI can benefit the SNR performance. However, same as the other multi-shot EPI techniques, iblocks-EPI is highly susceptible to minuscule and macroscopic motions during data acquisition. In this study, we developed a self-calibrated and collaborative iblocks-DTI reconstruction framework that can correct image artifacts and diffusion-encoding contrast change caused by minuscule and macroscopic motions.

Thomas L. Chenevert¹ and Dariya I. Malyarenko^1
1University of Michigan Hospitals, Ann Arbor, MI, United States

Systematic errors confound wide-spread clinical use of apparent diffusion coefficient (ADC) for diagnostic and prognostic applications. Standard clinical diffusion sequences using single-spin-echo echo-planner-imaging are susceptible to gradient channel-specific eddy-currents for b>0 inducing distortions of voxel magnetization-density (MD), as well as geometric distortion.  Unlike geometric distortion that are largely correctable by image registration to b=0, persisting signal amplitude distortions lead to systematic spatially-dependent errors mimicking, but physically distinct from, non-uniform diffusion weighting induced by gradient nonlinearity (GNL). This study proposes the use of geometric distortion parameters derived from in-plane image registration for MD correction of ADC in presence of GNL.

 

Xiaodong Ma^1,2, Yilong Liu^1,2, Zheyuan Yi^1,2,3, Alex T. Leong^1,2, Hua Guo⁴, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, China, ²Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China, ³Electrical and Electronic Engineering, Southern University of Science and Technology, Shenzhen, China, ⁴Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China

We propose a novel joint calibrationless reconstruction for accelerating multi-shot navigator-free DTI, using a low-rank completion approach. The redundant information across different directions is utilized to facilitate the reconstruction, including sharable coil sensitivities and anatomical structures. A 3D Hankel tensor was constructed and its concatenated Hankel matrices were used for low-rank approximation. In vivo human brain DTI experiment shows that the proposed joint reconstruction can reduce artifacts in diffusion-weighted images, and yield more accurate DTI metrics, when compared with separate reconstruction for different directions. This method also presents a new potential reconstruction strategy for fast high-resolution DTI.

Xinzeng Wang¹, Ersin Bayram¹, Daniel Litwiller², Tetsuya Wakayama³, Alan B McMillan⁴, Lloyd Estowski⁵, Ty A Cashen⁵, and Ali Pirasteh^4
1Global MR Applications & Workflow, GE Healthcare, Houston, TX, United States, ²Global MR Applications & Workflow, GE Healthcare, New York, NY, United States, ³Global MR Applications & Workflow, GE Healthcare, Hino, Japan, ⁴Radiology, University of Wisconsin Madison, Madison, WI, United States, ⁵Global MR Applications & Workflow, GE Healthcare, Waukesha, WI, United States

While echo-planar diffusion-weighted imaging (EP-DWI) is the main sequence for cancer detection in the prostate peripheral zone, it is susceptible to signal loss and distortion due to B0-field inhomogeneities secondary to a variety of causes, including rectal gas or metal hardware in the pelvis. We were able to demonstrate that a spin-echo based DWI sequence with radial k-space sampling (PROPELLER) can overcome such artifacts and the addition of a deep-learning reconstruction algorithm can overcome the poor signal-to-noise (SNR) profile of the PROPELLER-DWI, overall generating images with minimal-to-no appreciable artifact and favorable SNR.

Peidong He¹, Zifei Liang¹, Marco Muccio¹, Florian Knoll¹, Jiangyang Zhang¹, and Yulin Ge^1
1Department of Radiology, New York University School of Medicine, New York, NY, United States

In diffusion MRI, a higher number of gradient directions benefit to SNR and robustness of fiber rotational invariant estimation for tensor computation, however, it makes the acquisition time to lengthy to be clinically viable. This study was to apply a deep learning approach to generate new individual direction diffusion weighted (DW) source images (e.g., 60 more directions) from original 30 direction DW images based on high-angular-resolution (90 directions) dataset. Such an approach not only significantly reduce the scan time using one-third original DW images, but also be able to compute dMRI-derived parametric maps using a standard tensor model.

Cornelius Eichner¹, Michael Paquette¹, Angela D Friederici¹, and Alfred Anwander^1
1Department of Neuropsychology, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany

Advanced diffusion MRI (dMRI) data with high resolution and strong diffusion contrast typically suffer from low SNR levels. Therefore, denoising algorithms such as MP-PCA became an essential part of current dMRI processing pipelines. To overcome challenges related to violations of MP-PCAs assumption of tissue homogeneity in typical dMRI data, we here introduce an informed-MP-PCA (iMP-PCA) algorithm taking local differences in tissue composition into account. Denoising-performance of iMP-PCA was compared to conventional MP-PCA and evaluated on both magnitude and real-valued dMRI data. iMP-PCA was shown to significantly improve denoising-performance, especially at tissue boundaries and in regions of low SNR.

Xiaoping Wu¹ and Kamil Ugurbil^1
1Center for Magnetic Resonance Research, Radiology, Medical School, University of Minnesota, Minneapolis, MN, United States

We introduce a new denoising framework for denoising magnitude diffusion MRI. The framework synergistically combines the variance stabilizing transform with optimal singular-value manipulation. The usefulness of the proposed framework is demonstrated using both simulation and real-data experiments. Our results show that the proposed denoising framework can significantly improve signal-to-noise ratios across the entire brain, leading to substantially enhanced performances for estimating diffusion-tensor-related indices and for resolving crossing fibers when compared to another competing method. As such, the proposed denoising method is expected to have great utility for high-quality, high-resolution whole-brain diffusion MRI, desirable for many neuroscience and clinical applications.

Elizabeth Powell^1,2, Torben Schneider³, Marco Battiston², Francesco Grussu^2,4, Ahmed Toosy², Jonathan D Clayden⁵, and Claudia A. M. Gandini Wheeler-Kingshott^2,6,7
1Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ²NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Queen Square Institute of Neurology, Faculty of Brain Sciences, University College London, London, United Kingdom, ³Philips Healthcare, Guildford, United Kingdom, ⁴Centre for Medical Image Computing, Department of Computer Science, University College London, London, United Kingdom, ⁵Developmental Imaging and Biophysics Section, Great Ormond Street Institute of Child Health, University College London, London, United Kingdom, ⁶Department of Brain and Behavioural Sciences, University of Pavia, Pavia, Italy, ⁷Brain MRI 3T Center, IRCCS Mondino Foundation, Pavia, Italy

Nyquist sampling errors in echo planar imaging (EPI) often require 2D phase correction during reconstruction to remove unwanted ghost artefacts; however, phase corrections can be challenging to translate to high b-value diffusion weighted imaging (DWI) owing to associated noise amplification. We introduce SPECTRE (SENSE with 2D PhasE CorrecTion and channel-wise noise REmoval), and demonstrate that the SNR gains achieved by denoising complex channel data enable robust ghost correction without biasing diffusion parameter estimates.

Merry P. Mani¹, Hemant Kumar Aggarwal², Sanjay Ghosh¹, and Mathews Jacob^2
1Department of Radiology, University of Iowa, Iowa City, IA, United States, ²Department of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United States

We propose a model-based deep learning architecture for the reconstruction of highly accelerated diffusion MRI. We introduce the use of a pre-trained denoiser as the regularizer in a model-based recovery for diffusion weighted data from k-q under-sampled acquisition in a parallel MRI setting. The denoiser is designed based on a general tissue microstructure diffusion signal model with multi-compartmental modeling. A neural network was trained in an unsupervised manner using a convolutional auto-encoder to learn the diffusion MRI signal subspace. To demonstrate the acceleration capabilities of the proposed method, we perform MRI reconstruction experiments on a simulated brain dataset.

Lebina Shrestha Kakkar¹, Muhammad Usman¹, Alex Kirkham², Simon Arridge¹, and David Atkinson^1
1Centre for Medical Imaging, University College London, London, United Kingdom, ²Department of Radiology, University College Hospital, London, United Kingdom

Prostate diffusion-weighted (DW) MRI based on single-shot echo planar imaging (EPI) can suffer from geometric distortions caused by off-resonance magnetic field at the prostate-rectal air interface. Although techniques exist for effective distortion correction in brain imaging, such approaches may struggle with the severe distortions observed in prostate imaging. Here we focus on applying a recently developed model-based reconstruction technique to correct distortions in typical clinically used prostate DW-MRI. The distortion correction is feasible and may improve prostatic zonal anatomy in DW-MRI. Additionally, corrected averaged high b-value datasets show higher SNR than uncorrected dataset and may further help to identify tumours.

Shihui Chen¹, Mei-Lan Chu², Chun-Jung Juan³, Liyuan Liang¹, and Hing-Chiu Chang^1
1The University of Hong Kong, Hong Kong, Hong Kong, ²Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan, Taipei, Taiwan, ³Department of Medical Imaging, China Medical University Hsinchu Hospital, Taiwan, Taipei, Taiwan

Low SNR and long acquisition time hinder the clinical feasibility of intra-voxel incoherent motion (IVIM) diffusion-weighted imaging. In this work, we used a joint reconstruction framework based on POCSMUSE algorithm to simultaneously reconstruct under-sampled multi-b multi-shot DW-EPI data without undesired noise amplification. A proposed parametric correction scheme (PCS) was incorporated into POCSMUSE algorithm for robust reconstruction based on either conventional bi-exponential or simplified IVIM model. The proposed method demonstrated that the high-quality brain IVIM images could be reconstructed from highly accelerated (R=4) data acquired with multi-b multi-shot DW-EPI.

Oliver Maier¹, Stefan M Spann¹, Lea Bogensperger^1,2, and Rudolf Stollberger^1,3
1Institute of Medical Engineering, Technical University Graz, Graz, Austria, ²Institute for Computer Graphics and Vision, Technical University Graz, Graz, Austria, ³Biotechmed, Graz, Austria

Multi-Shell DTI suffers from low SNR for high b-value data and prolonged scan time. The Gaussian noise assumption is typically violated due to multi-coil imaging and magnitude forming thus requiring special treatment to avoid biases in the DTI estimates. To this end, we propose a model-based reconstruction technique to exploit the Gaussian noise in the raw k-space data and enable acceleration of the DTI measurement. We show the acceleration potential and quantitative accuracy of the proposed method for mono- and bi-exponential fitting approaches on freely available DTI data and full brain DTI measurements of one healthy volunteer.

SeyyedKazem HashemizadehKolowri¹, Rong-Rong Chen¹, Ganesh Adluru^2,3, and Edward V. R. DiBella^1,2,3
1Electrical and Computer Engineering, University of Utah, SALT LAKE CITY, UT, United States, ²Radiology and Imaging Science, University of Utah, SALT LAKE CITY, UT, United States, ³Biomedical Engineering, University of Utah, SALT LAKE CITY, UT, United States

Simultaneous multi-slice (SMS) acquisition combined with blipped controlled aliasing in parallel imaging is commonly used to accelerate diffusion imaging with single-shot EPI sequences.  In this work, we propose a new method, termed regularized image domain split slice-GRAPPA (RI-SSG), which allows an efficient image domain implementation of SSG coupled with total variation regularization to improve the quality of SMS reconstruction. We process two single-shot EPI datasets acquired using diffusion protocol of Human Connectome Project in Aging to evaluate performance of SMS reconstructions. The RI-SSG yields less noisy results than SENSE and SSG in estimating diffusion-weighted images and parametric maps of diffusion. 

Malte Steinhoff¹, Alfred Mertins¹, and Peter Börnert^2,3
1Institute for Signal Processing, University of Luebeck, Luebeck, Germany, ²Philips Research Europe, Hamburg, Germany, ³Department of Radiology, LUMC, Leiden, Netherlands

We propose a self-navigated iterative reconstruction algorithm for multi-shot DWI which effectively performs the shot phase updates with a fixed joint image prior. This framework further nicely incorporates deep learning generated image priors into the shot phase estimation while keeping the joint image production isolated. A U-Net is trained on extra-navigated data to mitigate phase cancellation artifacts. The algorithm with and without U-Net support is compared to self- and extra-navigated reference algorithms. The U-Net approach effectively mitigates phase-related signal cancellation artifacts. The improved multi-shot image prior regularizes the shot phase estimation enabling highly segmented self-navigated diffusion echo-planar imaging.

Yunsong Liu¹, Congyu Liao², Kawin Setsompop², and Justin P. Haldar^1
1Signal and Image Processing Institute, University of Southern California, Los Angeles, CA, United States, ²Martinos Center for Biomedical Imaging, Charlestown, MA, United States

gSlider is a diffusion MRI method that achieves fast high-resolution data acquisition using a novel slab-selective RF-encoding strategy.  Recent work has proposed subsampling of the multidimensional gSlider encoding space (diffusion-encoding/RF-encoding) for further improved scan efficiency.  Two different q-space regularization approaches (i.e., Laplace-Beltrami smoothness and spherical ridgelet sparsity) have been proposed to compensate for missing data, but there have been no systematic comparisons between the two. We compare and evaluate the potential synergies of these regularization approaches.  Results suggest that there can be small advantages to combining both regularization strategies together, although Laplace-Beltrami regularization alone is simpler and not much worse.

Gabriel Ramos-Llordén¹, Lipeng Ning¹, Congyu Liao², Rinat Mukhometzianov³, Oleg Michailovich³, Kawin Setsompop², and Yogesh Rathi^1
1Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ²Massachusetts General Hospital, Harvard Medical School, Charlestown, MA, United States, ³University of Waterloo, Waterloo, ON, Canada

gSlider is an efficient, super-resolution, technique to achieve submillimeter diffusion MRI data circumventing the trade-off between image resolution and SNR. Yet, the long acquisition time is still an issue. In this work, we extend gSlider by allowing under-sampling both in q-space and Radio-Frequency (RF)-encoded data, achieving then shorter acquisition time that gSlider. Our method, gSlider-SR, uses a basis of Spherical-Ridgelets to exploit the redundancy of the dMRI data, while at the same time enhancing SNR. We demonstrate that only ten minutes are needed to reconstruct 64 diffusion directions (b=2000s/mm²) at 860 μm data with reliably signal preservation.

Yi-Hang Tung¹, Frank Godenschweger¹, Myung-Ho In², Alessandro Sciarra³, and Oliver Speck^1
1Biomedical Magnetic Resonance, Otto-von-Guericke University Magdeburg, Magdeburg, Germany, ²Department of Radiology, Mayo Clinic, Rochester, MN, United States, ³Medicine and Digitalization, Otto-von-Guericke University Magdeburg, Magdeburg, Germany

     VAT-DIADEM is able to accelerate the distortion-free T2 weighting and diffusion weighting imaging sequence DIADEM at ultra-high field. The VAT gradient is turned on during imaging read-out in the slice selection direction. To further increase the acquisition efficiency, we developed a blipped version of the VAT gradient and applied simultaneous multi-slice imaging, based on blipped-CAIPI. The result is distortion reduction in EPI or acceleration in DIADEM and the increased volume coverage without increasing TR.

Samuel St-Jean¹, Max A. Viergever¹, and Alexander Leemans^1
1Image sciences institute, University Medical Center Utrecht, Utrecht, Netherlands

Small variations in diffusion MRI metrics between subjects are ubiquitous due to differences in scanner hardware and are entangled in the genuine biological variability between subjects, including abnormality due to disease. In this work, we propose a new harmonization algorithm based on adaptive dictionary learning to mitigate the unwanted variability caused by different scanner hardware while preserving the biological variability of the data. Results show that unpaired datasets from multiple scanners can be mapped to a scanner agnostic space while preserving genuine anatomical variability, reducing scanner effects and preserving simulated edema added to test datasets only.

Matthew J. Middione¹, Michael Loecher¹, Kévin Moulin^1,2, and Daniel B. Ennis^1,2,3
1Department of Radiology, Stanford University, Palo Alto, CA, United States, ²Cardiovascular Institute, Stanford University, Palo Alto, CA, United States, ³Veterans Administration Health Care System, Division of Radiology, Palo Alto, CA, United States

In DWI, applied diffusion gradients are assumed to be linear in space, but in practice are accompanied by additional undesired concomitant gradient (CG) fields that arise as a consequence of Maxwell’s equations. These CG fields contribute a residual gradient moment for asymmetric diffusion encoding strategies, which may impact the quantitative accuracy of the measured ADC. In this work, a CG correction method that does not increase the minimum achievable TE was implemented for clinically relevant asymmetric diffusion encoding DWI protocols and the mean ADC error reduction was characterized.

Yoojin Lee¹, Klaas P Pruessmann², and Zoltan Nagy^1
1Laboratory for Social and Neural Systems Research (SNS Lab), University of Zurich, Zurich, Switzerland, ²Institute of Biomedical Engineering, ETH, Zurich, Switzerland

Diffusion MRI commonly proceeds with single-shot echo planar imaging (EPI) readout, which renders this modality sensitive to Eddy current and susceptibility-induced distortions. Both image-based and k-space-based correction methods have been suggested previously. The aim of this project was to use a standard dataset and compare output images of two image-based correction methods (topup/Eddy and Tortoise) and the extended signal model method. The latter relies on separate magnetic field monitoring data and a B_o field map. Qualitatively the output of three pipelines were similar. Future work will make the diffusion acquisition more challenging for a more complete systematic evaluation.

Xiaobing Fan¹, Aritrick Chatterjee¹, Milica Medved¹, Aytekin Oto¹, and Gregory S. Karczmar^1
1Radiology, The University of Chicago, Chicago, IL, United States

We present a new concept based on matrix analysis for analyzing hybrid multi-dimensional T2-weighted and diffusion-weighted MRI of the prostate. This study evaluates whether matrix analysis is useful in diagnosis of prostate cancer. The hybrid data were linearized first by taking natural logarithms. Then the hybrid symmetric matrix was formed by multiplying by its own transpose matrix for each pixel. The eigenvalues and eigenvectors are calculated for this symmetric matrix to generate color maps. The preliminary results suggest that the combined color eigenvalue map provides new information that could help to identify and stage prostate cancer.

MinJung Jang¹, Seokha Jin¹, MungSoo Kang¹, SoHyun Han², and HyungJoon Cho^1
1Biomedical Engineering, Ulsan national institute of science and technology, Ulsan, Republic of Korea, ²Center for Neuroscience Imaging Research, Institute of Basic Science, Suwon, Republic of Korea

The purpose of this study is to demonstrate automatic segmentation for reduced diffusion and perfusion areas in ischemic stroke brains with pattern-recognition (constrained non-negative matrix factorization (cNMF)) to directional intravoxel incoherent motion MRI (IVIM-MRI). The robustness of region segmentation with pattern-recognition was observed in both simulations and in vivo experiments. Using pattern-recognition analysis of IVIM signal, white matter in which flow may have been aligned, and lesion regions are automatically segmented in ischemic stroke brain. In this study, we successfully implemented pattern-recognition to directional IVIM signals.

Maurizio Bergamino¹, Ryan R Walsh², and Ashley M Stokes^1
1Division of Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States, ²The Muhammad Ali Parkinson Center, Barrow Neurological Institute, Phoenix, AZ, United States

The objective of this study is to investigate differences in white matter (WM) integrity between Alzheimer’s disease (AD) and healthy subjects (HC) using diffusion tensor imaging (DTI) metrics from standard DTI and free-water (FW)-DTI. Regional changes in DTI metrics were found with both standard and FW-DTI, while FW-DTI improves the reliability and inter-parameter consistency of DTI metrics in the presence of atrophy. We hypothesize that the implementation of FW correction algorithm for DTI may provide more sensitive and specific insight into AD-related pathological changes in WM.

Di Guo¹, Zhangren Tu^1,2, Zifei Zhang³, Tianyu Qiu³, Xiaofeng Du¹, Min Xiao², Zhong Chen³, and Xiaobo Qu^3
1School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, China, ²School of Opto-Electronic and Communication Engineering, Xiamen University of Technology, Xiamen, China, ³Department of Electronic Science, Xiamen University, Xiamen, China

 The magnetic resonance diffusion ordering spectrum has been widely used in the separation of mixture due to the mobility of molecular self-diffusion reaction molecules. We propose a hybrid time and exponential decay signal recovery method based on low rank Hankel matrix¹ to accelerate the acquisition. The experimental results show that this method enables to reduce the error between the recovery diffusion spectrum and the fully sampled signal, and enhance the peak intensity.

Suguru Yokosawa^1,2, Toru Shirai¹, Hisaaki Ochi¹, and Yoshitaka Bito^3
1Research & Development Group, Hitachi, Ltd., Tokyo, Japan, ²Graduate School of Engineering, Chiba University, Chiba, Japan, ³Healthcare Business Unit, Hitachi, Ltd., Tokyo, Japan

   In this study, we proposed a method for characterizing intravoxel spatial distribution of diffusion by using texture analysis with GLCM (gray level cooccurrence matrix) from single-shell diffusion MRI data, which can be acquired in practical scan time. Since the method does not assume a diffusion distribution model, unstable calculation such as fitting process is not required. The method could provide novel diagnostic indices of diffusion MRI without additional acquisition.

Suheyla Cetin Karayumak^1,2, Lily O’Sullivan², Monica Gabriella Lyons², Marek Kubicki^1,2, and Yogesh Rathi^1,2
1Harvard Medical School, Boston, MA, United States, ²Brigham and Women's Hospital, Boston, MA, United States

This study thoroughly explores the reliability of a statistical multi-site diffusion-MRI harmonization method, ComBat, using a large diffusion-MRI dataset involving 5 sites. Additionally, we demonstrate the effects of using different software packages to estimate the diffusion tensor on the harmonization performance of ComBat. Even when the processing steps (and software) were identical for all sites, ComBat altered the inter-group variability at each site and in some regions flipped the direction of effect-sizes. Finally, when different software packages were used to estimate the diffusion tensor at different sites, ComBat introduced much drastic effects, thereby reducing the reliability of the results.

Suheyla Cetin Karayumak^1,2, Lily O’Sullivan², Monica Gabriella Lyons², Tashrif Billah², Ofer Pasternak^1,2, Sylvain Bouix^1,2, Marek Kubicki^1,2, and Yogesh Rathi^1,2
1Harvard Medical School, Boston, MA, United States, ²Brigham and Women's Hospital, Boston, MA, United States

This study attempts to highlight the reproducibility problem in clinical diffusion MRI studies by comparing different software in a large diffusion MRI data, collected from multiple sites with 291 schizophrenia patients and 251 healthy controls.  We fit diffusion tensor using FSL, MRtrix and Slicer with ordinary-least-squares (OLS) and weighted-least-squares (WLS) methods. To assess biological differences (measured by FA) between controls and patients, we computed effect-sizes (Cohen’s d) at each site using each software. We observed significant software effects with each software package producing significantly different results, when either OLS or WLS, was used.

Kurt G Schilling¹, Bennett A Landman^1,2,3,4, Alexander Leemans⁵, Adam W Anderson^1,2, and Chantal M.W. Tax^6
1Vanderbilt University Institute of Imaging Science, Nashville, TN, United States, ²Biomedical Engineering, Vanderbilt University, Nashville, TN, United States, ³Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, TN, United States, ⁴Electrical Engineering, Vanderbilt University, Nashville, TN, United States, ⁵Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands, ⁶CUBRIC, School of Physics, Cardiff University, Cardiff, United Kingdom

We investigated the diffusion fiber response function by characterizing the modelled diffusivity, the shape, and the size of the diffusion signal in regions of single fiber populations. We show that the various descriptors of the diffusion signal vary significantly across white matter pathways within a subject, and across subjects for a given pathway. Analysis of variance suggests that variability between pathways is greater than across subjects. Understanding the response function and its variation are necessary for accurate fiber orientation characterization, as well as characterizing microstructural effects from geometrical effects within a voxel. 

Jiyoon Yoo¹, Leevi Kerkelä¹, Patrick W. Hales¹, Kiran K. Seunarine¹, Noemi G. Gyori^1,2, Enrico Kaden², and Christopher A. Clark^1
1UCL Great Ormond Street Institute of Child Health, London, United Kingdom, ²UCL Centre for Medical Image Computing, London, United Kingdom

Several neurological diseases are associated with microstructural changes in the hippocampus which can be observed using diffusion MRI. However, DTI derived microstructural metrics such as fractional anisotropy (FA) are not able to differentiate between orientation dispersion and microscopic anisotropy. In this study, we applied an optimised multidimensional diffusion encoding sequence to measure microscopic fractional anisotropy (µFA) and normalised size variance (C_MD) in the human hippocampus of healthy subjects. We also defined a clinically feasible acquisition protocol by subsampling the full data set.

Zhaoqing Li¹, Huan Gao², Kedi Xu², and Ruiliang Bai^1,3
1Interdisciplinary Institute of Neuroscience and Technology (ZIINT), College of Biomedical Engineering and Instrument Science, Zhejiang University, HangZhou, China, ²Qiushi Academy for Advanced Studies (QAAS), Key Laboratory of Biomedical Engineering of Ministry of Education, College of Biomedical Engineering and Instrument Science, Zhejiang University, HangZhou, China, ³School of Medicine, Zhejiang Univerisity, HangZhou, China

Focal cortical ischemia animal model has been widely used to study neuroanatomical reorganization of cortex with intracortical microstimulation. However, the whole-brain structural changes following focal ischemia on motor cortex is unknown. In this study, high-resolution DTI is performed on focal unilateral motor cortex ischemia rat model at 14T. Voxel-wise based analysis (VBA) and VBA-guided connectivity analysis are performed to explore potential whole-brain microstructure and global structural connectivity changes. Our results show that large-scale white matter microstructural changes mainly distribute in corpus callosum, external capsule, cerebral peduncle. Significant reorganization of sensorimotor network associated with these regions were also found after ischemia.

Guangyu Dan^1,2, Yuxin Zhang^3,4, Zheng Zhong^1,2, Kaibao Sun¹, M. Muge Karaman^1,2, Diego Hernando^3,4, and Xiaohong Joe Zhou^1,2,5
1Center for Magnetic Resonance Research, University of Illinois at Chicago, Chicago, IL, United States, ²Department of Bioengineering, University of Illinois at Chicago, Chicago, IL, United States, ³Department of Medical Physics, University of Wisconsin-Madison, Madison, WI, United States, ⁴Department of Radiology, School of Medicine and Public Health, University of Wisconsin-Madison, Madison, WI, United States, ⁵Department of Radiology and Neurosurgery, University of Illinois at Chicago, Chicago, IL, United States

Parameters in many diffusion models depend on diffusion time. However, time-dependent diffusion behaviors in the long diffusion time regime have not been well studied because a longer diffusion time would lead to a longer TE, substantially reducing signal-to-noise ratio in conventional spin-echo diffusion pulse sequences. In this study, we employed a STEAM diffusion sequence to investigate the diffusion-time dependence of parameters in a fractional order calculus diffusion model. Our results showed substantial dependence of all diffusion parameters on diffusion times in the range of 100-1000 ms.

Dmitri Shastin^1,2, Maxime Chamberland¹, Greg Parker¹, Chantal M. W. Tax¹, Kristin Koller¹, Khalid Hamandi^1,2, William Gray^2,3, and Derek Jones^1
1School of Psychology, Cardiff University Brain Research Imaging Centre, Cardiff, United Kingdom, ²School of Medicine, Cardiff University Brain Research Imaging Centre, Cardiff, United Kingdom, ³BRAIN Biomedical Research Unit, Cardiff, United Kingdom

Anatomically constrained tractography aims to reduce the number of false positive streamlines by applying anatomically realistic priors. Grey matter-white matter interface is currently extracted based on T1 information requiring the availability of a T1 volume and its accurate co-registration with diffusion MRI data. Here we describe an alternative method based on multi-shell multi-tissue constrained spherical deconvolution that produces the interface without the need for T1. We go ahead and compare our method with the existing one and show a potential application.

Tokunori Kimura¹, Kousuke Yamashita¹, and Kouta Fukatsu^1
1Department of Radiological Science, Shizuoka College of Medicare Science, Hamamatsu-Shi, Japan

We proposed a new Diffusion Weighted Imaging (DWI) with T2-based water suppression technique (Wsup-DWI) and additionally combined with synthetic-MRI technique. Our water suppression was achieved by subtracting heavy T2-weighted (long-TE) image from the standard DWI images after correcting signal intensities due to b-value. We demonstrated that hyperintense artifacts due to CSF partial volume effects (PVE) were dramatically suppressed even at lower b-values (b); and quantitative maps of ADC and FA became close to the tissue values due to CSF reduction. Furthermore, fiber tracts became longer for Wsup-DWI data than for standard-DWI data in a tractography with the same condition.

Hu Cheng¹, Jian Wang^1,2, and Sharlene Newman^1
1Psychological and Brain Sciences, Indiana University, Bloomington, IN, United States, ²School of Information Science and Engineering, Shandong Normal University, Jinan, China

We propose a fully automated brain tissue segmentation method based on sparse representation of diffusion weighted imaging (DWI) signal. Learning a dictionary from DWI signals, brain voxels are classified into gray matter, white matter, and CSF according to their sparse representation of clustered dictionary atoms. The proposed method was tested on three subjects of the HCP DWI datasets and achieved good agreement with the segmentation on T1-weighted images using SPM12. The method is very fast and robust for a wide range of sparse coding parameter selection and works well on DWI data with less number of shells or gradient directions.

Anna-Katinka Bracher¹, Meinrad Beer¹, Volker Rasche², and Neubauer Henning^1
1Department of Diagnostic and Interventional Radiology, Ulm University Medical Center, Ulm, Germany, ²Department of Internal Medicine II, Ulm University Medical Center, Ulm, Germany

In standard diffusion-weighted imaging, the parameter maps are calculated based on a set of magnitude images. The standard fitting model of the parameter maps does not take a noise-induced bias into account. A noise offset model for Intravoxel Incoherent Motion (IVIM) in Diffusion Weighted Imaging (DWI) is introduced and the impact on diffusion parameter maps is evaluated for muscle and synovia of the knee joint. Parameter fitting using the noise-offset model results in faster signal decay of the fitted curve and therefore in increased diffusion coefficients compared to standard IVIM mapping.

Anouk van Rijn¹, Alexander Leemans¹, Geert Jan Biessels², and Alberto de Luca ^1
1Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands, ²Department of Neurology, University Medical Center Utrecht Brain Center, Utrecht, Netherlands

DTI metrics are often used without assessing the goodness of fit of the estimation model. We investigated local changes in fit residuals induced by pathology. To this end, a template of expected normalized residuals was created with 10 healthy controls, then voxel-wise comparisons were performed against the residuals of subjects affected by dementia. Results show that the residuals of infarcted regions are significantly different as compared to healthy tissue, whereas no differences were observed in hyperintensities. The fit residuals of the DTI model can be used to complement the information of DTI metrics at detecting microstructural changes in brain lesions.

Yaxin Niu¹, Ailian Liu¹, Ye Li¹, and Lizhi Lizhi Xie^2
1The First Affiliated Hospital of Dalian Medical University, Dalian Medical University, Da Lian, China, ²GE Healthcare, MR Research China, BeiJing, China, Beijing, China

Diffusion-weighted imaging reflects the micro-movements of water molecules. 

Catherine A Spilling¹, Franklyn A Howe¹, and Thomas R Barrick^1
1Neurosciences Research Centre, Molecular and Clinical Sciences Research Institute, St George's University of London, London, United Kingdom

Quasi-diffusion image (QDI) is a new ultra-high b-value diffusion magnetic resonance imaging technique which provides standard and non-Gaussian diffusion images. We use permutation analysis to optimise a clinical QDI tensor acquisition from a gold standard multi b-value protocol (28 non-zero b-values in 6 diffusion directions) to identify a 2 minute acquisition protocol with excellent tissue contrast. We achieve this by comparing different b-value combinations (2 non-zero b-values) to the gold standard using χ² difference in parameterised signal decay curves. We obtain an optimal acquisition protocol of b=0, 1080, 5000 s mm^-2 that may be acquired in a clinically acceptable time.

Xuedong Wang¹, Ailian Liu¹, ShiFeng Tian¹, Lizhi Xie², and Qingwei Song^1
1The first affiliated hospital of dalian medical university, DaLian, China, ²GE Healthcare, MR Research China, Beijing, Beijing, China

Diffusion tensor imaging (DTI), a type of MR functional imaging, can reflect the direction of the diffusion motion of water moleculesas well as describe the free diffusion rate of water molecules. The characteristic suggested the DTI were helpful in distinguishing endometrial carcinoma (EC) from endometrial polyp (EP).

Erick O Buko¹, Afis Ajala¹, Jiming Zhang², Pei Herng Hor¹, and Raja Muthupillai^2
1Physics, Texas Center for Superconductivity at University of Houston, Houston, TX, United States, ²Diagnostic and Interventional Radiology, Baylor St. Luke's Medical Center, Houston, TX, United States

Determination of perfusion fraction (f) from IVIM-MRI data is challenging. Recent studies have shown that estimation of f is influenced by echo time (TE). Eliminating thisTE dependency on f requires acquiring data sets at multiple TEs.  Here, we propose an analytical segmented approach to correct for the TE dependency on f arising from T2 differences between the tissue and fluid compartments (AS-T₂).  Numerical simulations predict, and phantom experiments confirm that, -compared to commonly used segmented approaches to estimate f, AS-T₂ approach permits the determination of perfusion fraction without TE dependence, and with fewer measurements.

Erick O Buko¹, Afis Ajala¹, Jiming Zhang², and Pei Herng Hor^1
1Physics, Texas Center for Superconductivity at University of Houston, Houston, TX, United States, ²Diagnostic and Interventional Radiology, Baylor St. Luke's Medical Center, Houston, TX, United States

We present a simple phantom set-up to model diffusion weighted imaging signal arising from intra-voxel incoherent motion (IVIM).  The model provides means to independently control both the perfusion fraction (f) as well as the pseudodiffusion coefficient (D_f).  

Patrik Brynolfsson¹, Markus Nilsson², Christian Jamtheim Gustafsson^1,3, Tim Sprenger⁴, Pia C Sundgren^5,6, Lars E Olsson¹, and Filip Szczepankiewicz^7,8
1Dept. of Translational Medicine, Division of Medical Radiation Physics, Lund University, Lund, Sweden, ²Dept. of Clinical Sciences, Division of Radiology, Lund University, Lund, Sweden, ³Dept. of Hematology, Oncology and Radiation Sciences, Lund University, Lund, Sweden, ⁴MR Applied Science Laboratory Europe, GE Healthcare, Stockholm, Sweden, ⁵Diagnostic Radiology, Clinical Sciences, Lund University, Lund, Sweden, ⁶Lund University Bioimaging Center, Lund University, Lund, Sweden, ⁷Lund University, Lund, Sweden, ⁸Brigham and Women's Hospital, Boston, MA, United States

Brain and head and neck cancer patients need fixation while undergoing radiotherapy, which prevents the use of a dedicated head coil when imaging patients using MRI in radiotherapy setup. We investigate the impact of coil setup on the technical feasibility of tensor-valued diffusion MRI in a radiation therapy setting.  Three coil configurations were evaluated with respect to SNR and parameter bias in the resulting parametric maps using Bland-Altman plots. As expected, the coils configured for radiation therapy imposed a penalty on the SNR, but accuracy was not negatively impacted.

Li Guo^1,2,3, Xinyuan Zhang^2,3, Changqing Wang⁴, Jian Lyu^2,3, Yingjie Mei⁵, Ruiliang Lu¹, Mingyong Gao¹, and Yanqiu Feng^2,3
1Department of MRI, The First People’s Hospital of Foshan (Affiliated Foshan Hospital of Sun Yat-sen University), Foshan, China, ²School of Biomedical Engineering, Southern Medical University, Guangzhou, China, ³Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, China, ⁴School of Biomedical Engineering, Anhui Medical University, Hefei, China, ⁵Philips Healthcare, Guangzhou, China

The noncentral Chi noise in magnitude image may significantly affect the reliability of quantitative analysis in diffusion-weighted (DW) magnetic resonance imaging (MRI), especially at high b-value and/or higher order modeling of diffusion signal such as diffusion kurtosis imaging (DKI). We developed a novel first-moment noise-corrected curve fitting model with adaptive neighborhood regularization (MN¹CM-ANR) algorithm for DKI. By fitting the signal to its first-moment (i.e. the expectation of the signal), MN¹CM-ANR can effectively compensate the bias due to the noncentral Chi noise. In addition, by exploiting the neighboring pixels to regularize the curve fitting, MN¹CM-ANR can reduce the measurement variance.

Umberto Villani^1,2, Simona Schiavi³, and Alessandra Bertoldo^1,2
1Padova Neuroscience Center, University of Padova, Padova, Italy, ²Department of Information Engineering, University of Padova, Padova, Italy, ³Department of Computer Science, University of Verona, Verona, Italy

Generalized sensitivity functions provide insights about which temporal observations are most informative for the estimation of biological model parameters. We formulate the same concept in the dMRI field to investigate how biophysical models/data representations react to HARDI acquisitions of different b-values. This approach handily shows how different parameters feature enhanced estimation precision at different b-values and exposes potential correlations between them, shedding light on possible a posteriori identifiability issues. Requiring only byproducts of standard optimization routines, generalized sensitivity functions can easily be integrated in standard analyses when proposing either a new model or a modification of existing ones. 

Óscar Peña-Nogales¹, Carlos Castillo^2,3,4, Carlos Lizama⁵, Rodrigo de Luis-Garcia¹, Santiago Aja-Fernández¹, and Pablo Irarrazaval^2,3,4,6
1Laboratorio de Procesado de Imagen, Universidad de Valladolid, Valladolid, Spain, ²Instituto de Ingeniería Biológica y Médica, Pontificia Universidad Catolica de Chile, Santiago, Chile, ³Millennium Nucleus for Cardiovascular Magnetic Resonance, Santiago, Chile, ⁴Biomedical Imaging Center, Pontificia Universidad Católica de Chile, Santiago, Chile, ⁵Departamento de Matemáticas y Ciencia de la Computación, Universidad de Santiago de Chile, Santiago, Chile, ⁶Electrical Engineering Department, Pontificia Universidad Católica de Chile, Santiago, Chile

Diffusion-Weighted (DW) imaging has a monoexponential signal decay at low b-values related to the statistical mechanics of Brownian motion. However, at high b-values the DW signal is no longer monoexponential. Various models have been proposed to better fit the data; nevertheless, none of them assumes that the DW signal is governed by the fractional-time order dynamics of the generalized Brownian motion (a.k.a., anomalous diffusion).  In this work, by assuming anomalous diffusion we solve the fractional-time order Bloch-Torrey equation for smooth diffusion-weighting gradients and compare the obtained model to existing ones. The proposed model outperforms state-of-the-art on synthetic anomalous phantom experiments.

Junying Wang¹, Hansen Schie², Weiqiang Dou³, and Xiaoyi He^1
1Medical Imaging, Shandong First Medical University, Jinan, Shandong, China, China, ²Medical Imaging, Shandong Provincial Qianfoshan Hospital，the First Hospital Affiliated with Shandong First Medical University, Jinan City, Shandong Province, China, China, ³GE Healthcare, MR Research China,Bejing, Bejing,China, China

This study aimed to test the reproducibility of DKI technique in normal liver and to explore the microscopic and diffusion differences between the left and right hepatic lobes.

32 healthy volunteers were scanned in DKI twice and the interval time was two weeks. All DKI-derived parametrical maps were reconstructed and mean values in eight liver segments were calculated.

We demonstrated that DKI in the liver showed excellent reproducibility between two measurements and found interesting regional distributions of the parameters in liver. Additionally, significant difference was found in DKI-derived parameters between the left and right hepatic lobes

Thomas R Barrick¹, Catherine A Spilling¹, Carson Ingo², Jeremy Madigan³, Jeremy D Isaacs³, Philip Rich³, Timothy L Jones³, Richard L Magin⁴, Matt G Hall^5,6, and Franklyn A Howe^1
1Neurosciences, St George's, University of London, London, United Kingdom, ²Northwerstern University, Chicago, IL, United States, ³St George's University Hospitals NHS Foundation Trust, London, United Kingdom, ⁴University of Illinois at Chicago, Chicago, IL, United States, ⁵National Physical Laboratory, Teddington, United Kingdom, ⁶University College London, London, United Kingdom

We present a new ultra-high b-value diffusion magnetic resonance imaging (dMRI) methodology, Quasi-Diffusion Imaging (QDI). The QDI technique includes a tensor representation of the dMRI data. We show that high contrast to noise images representing standard and non-Gaussian diffusion measures are obtainable in 1 to 4 minutes. QDI entropy maps show pathological contrast similar to Diffusional Kurtosis Imaging (DKI) in stroke and brain tumours. In addition, QDI overcomes the b-value limitations of DKI.

Marc Golub¹, R. N. Henriques², and R. G. Nunes^1
1ISR-Lisboa/LARSyS and Department of Bioengineering, Instituto Superior Técnico – Universidade de Lisboa, Lisbon, Portugal, Lisbon, Portugal, ²Champalimaud Neuroscience Programme, Champalimaud Centre for the Unknown, Lisbon, Portugal, Lisbon, Portugal

Free water elimination (FWE) for diffusion tensor imaging (DTI) aims to decouple the free water contribution from tissue's diffusion. Recent studies suggest that FWE-DTI is essential to characterize degenerative processes since it suppresses free water partial volume effects from DTI analysis and provides an index sensitive to edema and tissue degeneration. In this work, the state-of-the-art of FWE-DTI estimation is explored and several fitting procedures are tested on in vivo data to which simulated lesions were added. We found that, unlike the multi-shell approach, the single-shell implementation was unable to discriminate between changes in tissue diffusion and free water increases.

Fan Zhang¹, Anna Breger², Lipeng Ning¹, Carl-Fredrik Westin¹, Lauren J O'Donnell¹, and Ofer Pasternak^1
1Harvard Medical School, Boston, MA, United States, ²University of Vienna, Wien, Austria

Brain tissue segmentation is important in many diffusion MRI (dMRI) visualization and quantification tasks. We propose a deep learning tissue segmentation method that relies only on dMRI data. We leverage diffusion kurtosis imaging (DKI) and a recently proposed mean-kurtosis-curve (MK-curve) method to create a feature set that is highly discriminative between different types of tissues. We train a Unet model with a recently developed augmented target loss function on dMRI data from the Human Connectome Project. We show improved segmentation performance compared to several other methods and reliable segmentation results when applied on data with a different acquisition.

Qiqi Tong¹, Ting Gong¹, Hongjian He¹, Yi-Cheng Hsu², and Jianhui Zhong^1,3
1Center for Brain Imaging Science and Technology, Department of Biomedical Engineering, Zhejiang University, Hangzhou, China, ²MR Collaboration, Siemens Healthcare, Shanghai, China, ³Department of Imaging Sciences, University of Rochester, Rochester, NY, United States

Deep learning–based harmonization for diffusion imaging data with high efficiency and low cost is gaining popularity. However, the performance of the training-required network depends on the training data, which lack the diversity of the large sets of data in more substantial multicenter projects. We proposed a leave-one-tissue-out training strategy to evaluate the validity and reliability across scanners of a deep learning–based diffusion kurtosis imaging harmonization method. The results confirm that the deep learning–based network can still reconstruct the untrained tissue with validity, although the reliability would be higher when the tissue is trained.

Ernst Christiaanse^1,2, Alexander Leemans², Patrik Wyss¹, and Alberto de Luca^2
1Department of Radiology, Swiss Paraplegic Centre, Nottwil, Switzerland, ²Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands

It is essential for longitudinal studies to validate the reproducibility of the applied methods. Therefore, we assessed the test-retest reliability of diffusion kurtosis imaging of the brain in 10 healthy participants and showed a good reproducibility with some variation in different white matter regions.

Qiuyun Fan¹, Qiyuan Tian¹, Chanon Ngamsombat¹, and Susie Y. Huang^1
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, United States

Conventional diffusion imaging protocols may require tens or hundreds of samples in the q-space to generate reliable maps. Knowing that the k-q joint space is highly redundant and given the tradeoffs between k, q and SNR, we trained a deep convolutional neural network using a HIgh B-value and high Resolution Integrated Diffusion (HIBRID) sampling scheme, dubbed DeepHIBRID. We show DeepHIBRID outperforms conventional sampling schemes, and is capable of outputting 14 synthesized diffusion metric maps simultaneously with only 10 input images, without sacrificing the quality of the output maps, using 30x angular downsampling. 

Magdoom Kulam Najmudeen¹, Dario Gasbarra², and Peter J Basser^1
1SQITS/NICHD, National Institute of Health, Bethesda, MD, United States, ²University of Helsinki, Helsinki, Finland

A new signal model is introduced for diffusion tensor distribution imaging which is monotonically decreasing for all b-values unlike the cumulant and kurtosis models. A constrained multi-normal distribution is used as the tensor distribution which is fully characterized by the 2^nd order mean and 4^th order covariance tensors. A theoretical framework is presented showing the richness of covariance tensor, using synthetic gray and white matter voxels, and the ability to estimate the mean and covariance tensor from noisy MR signal.   

Alexis Reymbaut^1,2, Jeffrey Critchley³, Giuliana Durighel³, Tim Sprenger^4,5, Michael Sughrue⁶, and Daniel Topgaard^1,2
1Physical Chemistry, Lund University, Lund, Sweden, ²Random Walk Imaging AB, Lund, Sweden, ³Spectrum Medical Imaging, Sydney, Australia, ⁴Karolinska Institute, Stockholm, Sweden, ⁵GE Healthcare, Stockholm, Sweden, ⁶Charlie Teo Foundation, Sydney, Australia

Conventional $$$T_1$$$-mapping techniques are only sensitive to voxel-averaged $$$T_1$$$ values, which hinders the study of fiber-specific myelination changes in the developing, aging or diseased brain. While recent works have focused on combining diffusion- and $$$T_1$$$- weightings to access orientation-resolved $$$T_1$$$ values, they rely on assumptions regarding the voxel content. This work combines a prototype diffusion-relaxation MR acquisition and a Monte-Carlo inversion method to extract intra-voxel nonparametric 5D distributions of diffusion tensors and longitudinal relaxation rates $$$R_1=1/T_1$$$ without the use of limiting assumptions. Estimated $$$R_1$$$ values are then mapped onto nonparametric orientation distribution functions, thereby yielding fiber-specific longitudinal relaxation rates.

Alexis Reymbaut^1,2
1Physical Chemistry, Lund University, Lund, Sweden, ²Random Walk Imaging AB, Lund, Sweden

Either on the voxel scale or within sub-voxel diffusion compartments, tissue microstructure can be described using a diffusion tensor distribution $$$\mathcal{P}(\mathbf{D})$$$. One way to resolve microstructural heterogeneity relies on choosing a plausible parametric functional form to approximate $$$\mathcal{P}(\mathbf{D})$$$. However, such a high-dimensional mathematical object is usually intractable. Here, we define matrix moments enabling the computation of diffusion metrics for any arbitrary functional choice approximating $$$\mathcal{P}(\mathbf{D})$$$. Applying these general tools to the matrix-variate Gamma distribution on the voxel scale, we obtain a new signal representation, the matrix-variate Gamma approximation, that we validate in vivo and in silico.

Alexis Reymbaut^1,2, João Pedro de Almeida Martins^1,2, Chantal M. W. Tax³, Filip Szczepankiewicz^2,4,5, Derek K. Jones³, and Daniel Topgaard^1,2
1Physical Chemistry, Lund University, Lund, Sweden, ²Random Walk Imaging AB, Lund, Sweden, ³Cardiff University Brain Research Imaging Centre (CUBRIC), School of Psychology, Cardiff University, Cardiff, United Kingdom, ⁴Harvard Medical School, Boston, MA, United States, ⁵Radiology, Brigham and Women’s Hospital, Boston, MA, United States

Due to their cubic-millimeter scale, white-matter diffusion MRI voxels are often heterogeneous, comprising not only multiple fibre bundles but also grey matter, cerebrospinal fluid, or pathological tissue. To tackle this problem, conventional approaches rely on assumptions regarding tissue properties. This work combines state-of-the-art diffusion-relaxation MR acquisition and processing methods to extract intra-voxel nonparametric 5D distributions of diffusion tensors and transverse relaxation rates $$$R_2$$$ without the use of limiting assumptions. Orientation-resolved (fibre-specific) means of isotropic diffusivities, diffusion anisotropies and $$$R_2$$$ values are then obtained via clustering within the orientation subspace of these 5D distributions.

Alexis Reymbaut^1,2, Paolo Mezzani^1,3, João Pedro de Almeida Martins^1,2, and Daniel Topgaard^1,2
1Physical Chemistry, Lund University, Lund, Sweden, ²Random Walk Imaging AB, Lund, Sweden, ³Physics, Università degli Studi di Milano, Milan, Italy

In first approximation, the diffusion signal writes as the Laplace transform of an intra-voxel diffusion tensor distribution (DTD). Several algorithms have been introduced to estimate the DTD’s statistical descriptors (mean diffusivity, variance of isotropic diffusivities, mean squared diffusion anisotropy, etc.) by inverting data obtained from tensor-valued diffusion encoding schemes. However, the trueness and precision of these estimations have not been systematically assessed and compared across methods. Here, we compare such estimations in silico for a 1D Gamma fit, a generalized two-term cumulant approach, and 2D and 4D Monte-Carlo inversion techniques, using a common and clinically feasible tensor-valued acquisition scheme.

Ezequiel Farrher¹, Chia-Wen Chiang², Kuan-Hung Cho², Richard Buschbeck¹, Ming-Jye Chen², Zaheer Abbas¹, Kuo-Jen Wu³, Yun Wang³, Farida Grinberg¹, Chang-Hoon Choi¹, N. Jon Shah^1,4,5,6, and Li-Wei Kuo^2,7
1INM-4, Forschungszentrum Jülich, Jülich, Germany, ²Institute of Biomedical Engineering and Nanomedicine, National Health Research Institutes, Miaoli, Taiwan, ³Center for Neuropsychiatric Research, National Health Research Institutes, Miaoli, Taiwan, ⁴Department of Neurology, RWTH Aachen University, Aachen, Germany, ⁵JARA – BRAIN – Translational Medicine, RWTH Aachen University, Aachen, Germany, ⁶Institute of Neuroscience and Medicine 11, JARA, Forschungszentrum Jülich, Jülich, Germany, ⁷Institute of Medical Device and Imaging, National Taiwan University College of Medicine, Taipei, Taiwan

It is known that excess fluid as a result of vasogenic oedema formation following stroke onset obscures the microstructural characterisation of ischemic tissue by diffusion MRI. DTI-based free water elimination and mapping (FWE) has been proposed as a technique to potentially reduce the partial-volume effect. However, FWE estimation is ill-conditioned, leading to inaccurate results. More recently, it has been shown that the addition of a second dimension spanned by transverse relaxation weighting, mitigates the ill-conditioned problem. We aim here to investigate the latter model in a longitudinal study of MCAo stroke animal models.

Michael Paquette¹, Chantal M.W. Tax², Cornelius Eichner¹, and Alfred Anwander^1
1Neuropsychology, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ²CUBRIC, School of Physics, Cardiff University, Cardiff, United Kingdom

We investigate the effect of gradient non-linearities (GNL) on free gradient waveform used for B-tensor diffusion encoding. We show the magnitude of the GNL-bias for strong gradients of $$$300 m\text{T}/\text{m}$$$. We derive a closed-form formula of the voxelwise B-tensor under GNL, independent of the choice of gradient waveform used to encode the B-tensor.

Lipeng Ning^1,2, Filip Szczepankiewicz^1,2, Borjan Gagoski^2,3, Carl-Fredrik Westin^1,2, and Yogesh Rathi^1,2
1Brigham and Women's Hospital, Boston, MA, United States, ²Harvard Medical School, Boston, MA, United States, ³Boston Children's Hospital, Boston, MA, United States

Multi-dimensional imaging techniques can provide complementary information to investigate the microscopic organizations of biological tissue. Deriving meaningful indices from these multi-dimensional imaging data is important to make these techniques relevant to clinical research. In this work, we introduce a general framework to estimate the moments of the joint probability distribution of T₁, T₂ relaxation and diffusion coefficients. This framework is an extension of our previously introduced REDIM approach which only focused on joint moments of T₂ relaxation and diffusion. We show the cross coupling between different parameters using three datasets acquired using different protocols.

Qianqian Yang¹ and Viktor Vegh^2
1School of Mathematical Sciences, Queensland University of Technology, Brisbane, Australia, ²Centre for Advanced Imaging, University of Queensland, Brisbane, Australia

Diffusion MRI measures of the human brain provide key insight into microstructural variations across individuals and into the impact of central nervous system diseases and disorders. One approach to extract information from diffusion signals has been to use biologically relevant analytical models to link millimetre scale diffusion MRI measures with microscale influences. The other approach has been to represent diffusion as an anomalous process, and infer information from the different anomalous diffusion equation parameters. Here, we show how parameters of three established anomalous diffusion equations change with age, in a microstructurally complex tissue, the human corpus callosum.  

Jordan A. Chad^1,2 and Ofer Pasternak^3
1Rotman Research Institute, Baycrest Health Sciences, Toronto, ON, Canada, ²Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, ³Departments of Psychiatry and Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States

Fitting diffusion MRI signal models with the standard weighted linear least squares (WLLS) approach necessarily places lower weight on data with lower SNR, therefore placing lower weight on shells with higher b-values. This can be non-optimal for fitting signal models that rely on information from high b-shells. In this work, we propose a “nested” WLLS approach where each shell is assigned a relative weight, with standard WLLS applied within each shell. We demonstrate that weighting shells equally may be beneficial for fitting signal models dependent on multiple shells. 

Zihan Zhou¹, Qiuping Ding¹, Hongjian He¹, and Jianhui Zhong^1,2
1Center for Brain Imaging Science and Technology, Key Laboratory for Biomedical Engineering of Ministry of Education, College of Biomedical Engineering and Instrumental Science, Zhejiang University, Hangzhou, China, ²School of Medicine and Dentistry, University of Rochester Medical Center, New York, NY, United States

White Matter Tract Integrity (WMTI) is a biophysical model with specificity to underlying tissue microstructures. However, recent work has suggested that inter-compartmental water exchange may affect outcomes of the model metrics. In this work, we analytically relate the WMTI-derived dMRI metrics to membrane permeability, and validate our predictions using Monte Carlo simulations. Our results show that the water exchange has a non-trivial effect on the metrics and needs to be carefully considered in WMTI.

BO LI¹, YANG GAO¹, SHAOYU WANG², YAN XU², GUANG YANG³, and HUAPENG ZHANG^2
1Department of Radiology, The Affiliated Hospital of Inner Mongolia Medical University, Hohhot, China, ²MR Sicentific Marketing, Siemens Healthineers, Shanghai, China, ³Shanghai Key Laboratory of Magnetic Resonance, East China Normal University, Shanghai, China

This study aimed at using mean-apparent-propagator (MAP) diffusion model in differential diagnosis of high-grade glioma and single brain metastatic tumor.the findings show that multiple parameters quantified by DSI quantitative model MAP MRI, especially MSD and QIV has high applicatical value and provides more information for preoperative diagnosis of patients.

 

Diwei Shi¹, Xuesong Li², Ziyi Pan², Hua Guo², and Quanshui Zheng^1
1Center for Nano & Micro Mechanics, Tsinghua University, Beijing, China, ²Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China

We propose a novel definition of orientation distribution function (ODF), which also represents diffusion coefficients for non-Gaussian situations, thus ODF has physical meaning. Then an expression is derived to calculate ODF values through diffusion-weighted signals. Next a new HARDI (high angular resolution diffusion imaging) method called “g-OPDT” (grids orientation probability density transform) is designed for practical use. Numerical simulations are performed to verify its accuracy. ISMRM-2015-Tracto-challenge data are used to quantitatively compare g-OPDT with other methods. Results show that g-OPDT has superior performance on most indicators. In-vivo experiments are conducted to show that its glyph representations have less redundant lobes.

Bianca Freytag¹, Vicky Y Wang², Alistair A Young³, and Martyn P Nash^1
1Auckland Bioengineering Institute, The University of Auckland, Auckland, New Zealand, ²Veterans Affairs Medical Center, San Francisco, CA, United States, ³Biomedical Engineering Department, King's College London, London, United Kingdom

Electro-mechanical models of heart function rely upon accurate representation of microstructural orientations in the myocardium. We assessed the sensitivity of a finite element parameterisation of the myofibre orientation field derived from high-resolution DWI data of a canine heart. Eigenanalysis of the indifference region in the neighbourhood of the optimal fit of the myofibre field enabled quantification of helix angle uncertainty consistent with the DWI data. This method can be used to propagate the uncertainty in myofibre parameterisation to electro-mechanics simulation results.