View Presentation Video

Deneb Boito^1,2 and Evren Özarslan^1,2
1Biomedical Engineering, Linköping University, Linköping, Sweden, ²Center for Medical Image Science and Visualization, Linköping University, Linköping, Sweden

Keywords: Data Processing, Diffusion/other diffusion imaging techniques

Q-space trajectory imaging (QTI) is a diffusion MRI framework which access features of the microstructure through the statistical moments of the diffusion tensor distribution. To overcome unreliable estimates obtained with standard fitting methods, a constrained estimation framework named QTI+ was recently proposed. Constrained optimization however typically requires sophisticated fitting routines which introduce a heavy computational burden. In this work we thus explore the possibility of speeding up the QTI parameter estimation, while retaining strict positivity constraints, using artificial intelligence. Results are shown on synthetic datasets as well as for healthy subjects and data from brain tumor patients.

View Presentation Video

Rakibul Hafiz¹, Amritha Nayak^1,2, M. Okan Irfanoglu¹, Leighton Chan³, and Carlo Pierpaoli^1
1National Institute of Biomedical Imaging and Bioengineering (NIBIB), National Institutes of Health (NIH), Bethesda, MD, United States, ²Henry Jackson Foundation for Advancement of Military Medicine, Bethesda, MD, United States, ³Rehabilitation Medicine Department, National Institutes of Health (NIH), Bethesda, MD, United States

Keywords: Data Analysis, Diffusion/other diffusion imaging techniques, Quantitative Medical Imaging

We propose a novel quantity to correctly assess the extent individuals deviate from the median of a heavy-tailed distribution. We compute a percentile-based score, we call ‘P-score’, which normalizes the deviation of an individual from the sample median by incorporating the individual’s position in the left/right tail of the sample distribution and the corresponding length between the sample median and the 5th/95th percentile edge values of the sample distribution, respectively. We demonstrate the skewness present in diffusion MRI data and the bias introduced when Z-scores are used and further show the control of this bias using the proposed ‘P-scores’ approach.

View Presentation Video

Chu-Yu Lee¹ and Merry Mani^1
1Department of Radiology, The University of Iowa, Iowa City, IA, United States

Keywords: Data Acquisition, Diffusion/other diffusion imaging techniques, Kurtosis; Multi-slab; 3D

The fast mean kurtosis measurements using 2D DWI has reduced the scan time substantially but remains subject to low SNR, particularly at b-value = 2500 s/mm². A larger voxel (≥ 2 mm³) is normally used to increase SNR. This study presents the navigator-free 3D multi-slab DWI for fast mean kurtosis measurements with a resolution of 1.5 mm³. By using a short TR of 2.5 sec, the work demonstrates the promise of using the navigator-free 3D multi-slab DWI for whole-brain fast mean kurtosis measurements with an improved SNR efficiency.

View Presentation Video

Tianshu Zheng¹, Ruicheng Ba¹, Xiaoli Wang², Xizhen Wang³, Chuyang Ye⁴, and Dan Wu^1
1Department of Biomedical Engineering, College of Biomedical Engineering & Instrument Science, Zhejiang University, Hangzhou, Zhejiang, China, Hangzhou, China, ²School of Medical Imaging, Weifang Medical University, Weifang, Shandong, China, Weifang, China, ³Medical imaging center, Affiliated hospital of Weifang Medical University, Weifang, Shandong, China, Weifang, China, ⁴School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing, China, Beijing, China

Keywords: Machine Learning/Artificial Intelligence, Diffusion/other diffusion imaging techniques

Time-dependent diffusion magnetic resonance imaging (TDDMRI) is useful for non-invasive characterization of tissue microstructure. The TDDMRI models require both densely sampled q-space (b-value and diffusion direction) and t-space (diffusion time) data for microstructural fitting, leading to very time-consuming acquisition protocols. In this work, we presented a tDKI-Net to estimate diffusion kurtosis at multiple diffusion times, which was fed into the Karger model to obtain K₀ and transmembrane exchange time, using downsampled q-space and t-space data. We tested the proposed network in the normal rat brains, as well as those in a rat model of Middle Cerebral Artery Occlusion.

View Presentation Video

Guangqi Li¹, Yuancheng Jiang¹, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China

Keywords: Data Acquisition, Diffusion Tensor Imaging

Single-shot acquisition techniques are commonly used to acquire diffusion-weighted images due to their high sampling efficiency. Single-shot uniform-density spiral sampling combined with in-plane under-sampling has been used to achieve diffusion imaging. In this study, we investigated simultaneous multi-slice (SMS) -accelerated single-shot spiral imaging (SMS-SSH-Spiral) to further improve the scan efficiency of DWI. The in vivo results demonstrate the feasibility of SMS-SSH-Spiral acquisitions for diffusion tensor imaging with 2-fold slice acceleration and 3-fold in-plane acceleration. Diffusion data with 1.5mm isotropic resolution is acquired using our multi-band single-shot acquisition strategy.

View Presentation Video

Jieying Zhang¹, Simin Liu¹, and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China

Keywords: Image Reconstruction, Diffusion Tensor Imaging

3D simultaneous multi-slab imaging (SMSlab) can achieve high-resolution DWI with high SNR efficiency. Multi-band acceleration can also achieve less SNR reduction. Recently, we integrated SMSlab DWI with blipped-CAIPI gradients to reduce the g-factor penalty and proposed a 4D k-space framework (k_x-k_y-k_z-k_m) to model the signal encoding, with k_m representing the multi-band encoding. Because the blipped-CAIPI gradients are applied along the slice direction, they introduce k_z deviations from the nominal k-space location. This study proposed a hybrid-space reconstruction algorithm, REACH, to solve the phase interferences introduced by the blipped-CAIPI gradients.

View Presentation Video

Benjamin Ades-Aron¹, Santiago Coelho¹, Jelle Veraart¹, Timothy M. Shepherd¹, Dmitry S. Novikov¹, and Els Fieremans^1
1Radiology, Center for biomedical imaging, NYU Grossman School of Medicine, New York, NY, United States

Keywords: Data Processing, Diffusion/other diffusion imaging techniques, Harmonization Denoising

To address the ongoing reproducibility crisis in quantitative diffusion MRI (dMRI), efforts are underway to harmonize and improve precision of diffusion parameter estimation. Using inter- and intra-scanner test-retest higher order dMRI, we compare the reproducibility of two popular harmonization methods, ComBat and linear-RISH, to that of denoising using MPPCA on complex-valued dMRI. We find that denoising combined with harmonization improves voxel-wise test-retest ICC by up to 60% compared to harmonization alone. Using dMRI at different voxel sizes, we find that denoising reduces the bias due to varying noise floors more accurately than harmonization. Denoising appears essential to harmonize dMRI datasets.

View Presentation Video

Zihan Li¹, Berkin Bilgic^2,3, Hong-Hsi Lee^2,3, Kui Ying⁴, Hongen Liao¹, Susie Huang^2,3, and Qiyuan Tian^2,3
1Department of biomedical engineering, Tsinghua University, Beijing, China, ²Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States, ⁴Department of Engineering Physics, Tsinghua University, Beijing, China

Keywords: Machine Learning/Artificial Intelligence, Diffusion/other diffusion imaging techniques

The accurate estimation of diffusion model parameter values using non-linear optimization is time-consuming. Supervised learning methods using neural networks (NNs) are faster and more accurate but require external ground-truth data for training. A unified and self-supervised learning-based diffusion model estimation method DIMOND is proposed. DIMOND maps diffusion data to model parameter values using NNs, synthesizes the input data from the predictions using the forward model, and minimizes the difference between the raw and synthetic data. DIMOND outperforms conventional ordinary least square regression (OLS) and has a high potential to improve and accelerate data fitting for more complicated diffusion models.

View Presentation Video

Hamza Kebiri^1,2, Ali Gholipour², Davood Karimi², and Meritxell Bach Cuadra^1
1Center for Biomedical Imaging & Lausanne University Hospital, Lausanne, Switzerland, ²Harvard Medical School & Boston Children's Hospital, Boston, MA, United States

Keywords: Machine Learning/Artificial Intelligence, Diffusion/other diffusion imaging techniques, Fiber Orientation Distribution functions

Diffusion MRI of fetal and newborn brains is constrained by short scanning time allowing only a small number of diffusion measurements to be acquired. Methods going beyond the diffusion tensor model require multi-shell and multiple gradient directions in order to unveil more accurate white matter properties. We propose a learning based framework to reconstruct fiber orientation distribution functions from only six diffusion measurements by leveraging existing high-quality datasets. Quantitative evaluation on 15 newborn subjects show that our framework achieves competitive results with state-of-the-art methods. Qualitative evaluation on a fetus shows the model ability to translate to this challenging population.

View Presentation Video

Oscar Jalnefjord^1,2, Amina Warsame¹, and Louise Rosenqvist^1
1Department of Medical Sciences, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden, ²Department of Medical Physics and Biomedical Engineering, Sahlgrenska University Hospital, Gothenburg, Sweden

Keywords: Data Processing, Diffusion/other diffusion imaging techniques

Signal drift has been identified as a confounding factor in diffusion MRI (dMRI), causing increases variance and potential bias in derived parameter maps. In this work, we show that temporal signal drift is spatially dependent in human brain images for dMRI, thus calling for spatiotemporal corrections. We also show that signal drift can have a substantial effect on short-term repeatability of ADC and IVIM f obtained from data acquired with a protocol ordered by b-value as is commonly done in clinical practice.

View Presentation Video

Jiqing Huang¹, Benjamin Leporq¹, Olivier Beuf¹, and Hélène Ratiney^1
1Univ Lyon, INSA Lyon, CNRS, Inserm, CREATIS UMR 5220, U1206, F-69621, Lyon, Villeurbanne, France

Keywords: Data Processing, Diffusion/other diffusion imaging techniques

Liver fibrosis is one of the leading features in chronic liver disease (CLD) since it conditions the prognosis and guides the treatment strategy. In this work, estimated parameters from various diffusion-weighted MRI models fitted by the Bayesian method were analyzed for the relationship with liver fibrosis through spearman’s correlation and t-test. Four parameters (D_s, σ, D^*_F, D_app) were selected for fibrosis classification and achieved the best result based on the decision tree. Our result suggested that the statistical model and a hybrid IVIM-DKI model are promising models and confirmed the confounding effect of fat for diffusivity to assess liver fibrosis. 

View Presentation Video

Joseph Bartlett^1,2,3, Catherine Davey^1,2, Leigh Johnston^1,2, and Jinming Duan^3,4
1Department of Biomedical Engineering, University of Melbourne, Melbourne, Australia, ²Melbourne Brain Centre Imaging Unit, University of Melbourne, Melbourne, Australia, ³School of Computer Science, University of Birmingham, Birmingham, United Kingdom, ⁴Alan Turing Institute, London, United Kingdom

Keywords: Machine Learning/Artificial Intelligence, Diffusion/other diffusion imaging techniques

SDNet utilises the learning ability of deep neural networks with the robustness of model-based approaches to produce high quality fibre orientation distributions (FODs) from a reduced set of multi-shell diffusion weighted images (DWI). The cascaded architecture, with data consistency layers throughout, makes use of model based prior knowledge and spatial correlations within the DWI signal to achieve state-of-the-art performance in both sum of squared errors and angular correlation coefficient. Our model also shows competitive results with respect to apparent fibre density error and peak amplitude error over a range of regions of interest. 

View Presentation Video

Zhigang Wu¹, Yajing Zhang², Gilbert Guillaume³, Wengu Su⁴, Yan Zhao⁵, and Jiazheng Wang^6
1Philips Healthcare, Shenzhen, Ltd., Shenzhen, China, ²Philips Health Technology, Suzhou, China, ³MR Clinical Science, Philips Healthcare, Mississauga, ON, Canada, ⁴BU MR Application, Philips Health Technology, Suzhou, China, ⁵BU MR R&D, Philips Health Technology, Suzhou, China, ⁶Philips Healthcare, Beijing, China

Keywords: Pulse Sequence Design, Diffusion/other diffusion imaging techniques, Pulse Sequence Design, Diffusion/other diffusion imaging techniques, Diffusion, 3D

3D Diffusion imaging (3D DWI) has showed great potential in probing tissue microstructure and brain structural connectivity. However, motion-induced phase errors introduced by diffusion gradients will cause severe artifacts in 3D DWI. Multi-slab can be used to overcome this limitation, but it will introduce slab boundary artifacts. We propose a method which utilizes motion-compensated diffusion gradients for 3D DWI to mitigate the phase error between shots. Results from in vivo data demonstrate the proposed method can improve  image quality and realize an isotropic high resolution and whole brain 3D DWI in a single slab.

View Presentation Video

Guangqi Li¹, Xinyu Ye¹, Yuan Lian¹, Yajing Zhang², and Hua Guo^1
1Center for Biomedical Imaging Research, Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing, China, ²MR Clinical Science, Philips Health Technology (China), Beijing, China

Keywords: Image Reconstruction, Diffusion Tensor Imaging

Single-shot spiral acquisitions allow shorter TE, thus provide higher SNR compared to EPI acquisitions for DWI. However, spiral acquisitions are sensitive to field inhomogeneity. Parallel imaging techniques can be used to alleviate static B0 off-resonance effects. In this study, single-shot spiral acquisitions with a large acceleration factor of 5 or 6 were used to achieve sub-millimeter diffusion tensor imaging at 3T. The in vivo results demonstrate that the single-shot spiral sampling strategy can be adopted to achieve whole-brain diffusion tensor imaging with an in-plane resolution of 0.77×0.77mm².

View Presentation Video

Jenny Chen¹, Benjamin Ades-Aron¹, Saurabh Maithani¹, Yvonne Lui¹, Dmitry S. Novikov¹, Jelle Veraart¹, and Els Fieremans^1
1New York University Grossman School of Medicine, New York, NY, United States

Keywords: Data Acquisition, Diffusion/other diffusion imaging techniques, Clinical Translation, DKI

Multi-shell diffusion MRI (dMRI) opens up the opportunity for microstructure mapping, but is not routinely acquired in clinics. This study retrospectively analyzed a large diffusion kurtosis imaging (DKI) dataset (N=7984) acquired as multi-series in the clinic and proposes a workflow to inspect the data before preprocessing. Automated analysis of dicom headers shows about 36% with incomplete acquisition, likely due to time constraints and patient discomfort, and 32% with image discrepancies. Finally, we show that signal variations between series occur and need to be corrected for in preprocessing. Our results indicate higher-order dMRI in the clinical setting is feasible.

View Presentation Video

J-Donald Tournier^1,2, Ben Jeurissen^3,4, and Daan Christiaens^5,6
1Biomedical Engineering, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom, ²Centre for the Developing Brain, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom, ³imec-Vision Lab, Department of Physics, University of Antwerp, Antwerp, Belgium, ⁴Lab for Equilibrium Investigations and Aerospace, Department of Physics, University of Antwerp, Antwerp, Belgium, ⁵Medical Imaging Research Center, KU Leuven, Leuven, Belgium, ⁶Department of Electrical Engineering, ESAT/PSI, KU Leuven, Leuven, Belgium

Keywords: Data Analysis, Diffusion/other diffusion imaging techniques

Rician bias is a long-standing problem in diffusion MRI. We propose a simple iterative strategy to mitigate the problem when a voxel-wise model of the signal is available, and demonstrate its efficacy when fitting spherical harmonics and when included in a denoising strategy. 

View Presentation Video

Zhigang Wu¹, Yajing Zhang², Wengu Su³, Masami Yoneyama⁴, Peng Sun⁵, Jing Zhang⁶, Yan Zhao⁷, and Jiazheng Wang^5
1Philips Healthcare, Shenzhen, Ltd., Shenzhen, China, ²Philips Health Technology, Suzhou, China, ³BU MR Application, Philips Health Technology, Suzhou, China, ⁴BU MR Clinical Science, Philips Healthcare, Tokyo, Japan, ⁵Philips Healthcare, Beijing, China, ⁶MR Clinical Application, Philips Healthcare, Beijing, China, ⁷BU MR R&D, Philips Health Technology, Suzhou, China

Keywords: Data Acquisition, Diffusion/other diffusion imaging techniques, Reduced FOV, Compressed SENSE

DWI is very important for MRI examination, but it has limited resolution due to distortion, blurring, and signal loss caused by B0 inhomogeneity. Reduced FOV imaging could decrease these impacts. However, due to the coil geometry penalty, it’s hard to combine it with parallel imaging to further improve the image quality, it will suffer from noise breakthrough issues and unfolding artifacts. We propose a framework that combines reduced FOV imaging, Compressed SENSE framework simultaneously to overcome these issues. This framework allows a new solution for reduced FOV based diffusion imaging with high resolution, low distortion, and without noise breakthrough issue.

View Presentation Video

Wenjuan Wang¹, Wenjing Zhao¹, and Dmytro Pylypenko^2
1Department of Radiology, Weifang People’s Hospital, Weifang, China, ²MR Research China, GE Healthcare, Beijing, China

Keywords: Artifacts, Diffusion/other diffusion imaging techniques, multiplexed sensitivity encoding, reversed polarity gradients

The aim of this study is to investigate the application of multiplexed sensitivity encoding diffusion weighted imaging with reversed polarity gradients in improving the image quality of uterine tumors. 10 patients with uterine lesions in our hospital were enrolled. Conventional SS-EPI DWI, MUSE DWI, and RPG-MUSE DWI sequences were employed. The subjective image quality assessment and objective data measurement of uterine lesions scanned by different DWI sequence was evaluated by three radiologists using double-blind method. The results of this study showed that RPG-MUSE DWI can greatly improve the image quality of the uterus.

View Presentation Video

Pablo Stack-Sanchez¹, Donald Gross², and Christian Beaulieu^1
1Biomedical Engineering, University of Alberta, Edmonton, AB, Canada, ²Neurology, University of Alberta, Edmonton, AB, Canada

Keywords: Data Processing, Diffusion Tensor Imaging

A hippocampus-focused, 1 mm isotropic, diffusion tensor imaging 5.5 minutes 3T protocol has yielded high-quality diffusion images and maps. Post-processing with denoising can potentially reduce the scan time considerably. Compared to the previously used 10 direction/10 average protocol, denoising enabled the use of 4 averages for a scan time of only 2.2 minutes to yield acceptable 1 mm isotropic diffusion image quality and mean diffusivity maps (MD) of the whole hippocampus in healthy controls. This rapid protocol was then able to delineate focal and small hippocampal lesions with elevated MD in temporal lobe epilepsy patients.

View Presentation Video

Xiongpeng He¹, Xiaoan Zhang¹, Xin Zhao¹, Yongbin Sun², Pengfei Geng¹, and Kaiyu Wang^3
1Third Affiliated Hospital of Zhengzhou University, zhengzhou, China, ²Zhengzhou University People’s Hospital, zhengzhou, China, ³MR Research China, GE Healthcare, Beijing 100000, PR China, Beijing, China

Keywords: Machine Learning/Artificial Intelligence, Diffusion/other diffusion imaging techniques, Tract-Based Spatial Statistics

Early and accurate diagnosis of pediatric autism was difficult for clinicians, which hindered the timely treatment of patients.This study aimed to explore the applicability of machine learning models of diffusion kurtosis imaging (DKI) based on Tract-Based Spatial Statistics to diagnose pediatric autism. Results showed that DKI parameter was potential for differentiating early autism from the normal. And the machine learning models can be used for early detection of pediatric autism with high accuracy and sensitivity.