View Presentation Video

Santiago Coelho¹, Gregory Lemberskiy¹, Ante Zhu², Nastaren Abad², Thomas K. F. Foo², Els Fieremans¹, and Dmitry S. Novikov^1
1Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University School of Medicine, New York, NY, United States, ²GE Research, Niskayuna, NY, United States

Keywords: Data Processing, Diffusion/other diffusion imaging techniques

A naive answer to this question is to loop over all voxels re-do the training and apply machine learning estimators of your favorite model. This is grossly computationally inefficient. We propose a matrix pseudoinversion-based method that can estimate nonlinear biophysical model parameters from a large set of voxels with independent acquisition protocols in a few minutes. Our framework can be tailored to any convolution-based model. Furthermore, the protocols are not required to have shells and there are no limits to the protocol differences among voxels. This method is readily extendable for simultaneously varying diffusion times, B-tensor shapes, TE, etc.

View Presentation Video

Nagesh Adluru¹, Douglas Dean III¹, Molly Prigge², Jace King², Nicholas Lange³, Erin Bigler⁴, Brandon Zielinski⁵, Janet Lainhart¹, and Andrew Alexander^1
1University of Wisconsin, Madison, WI, United States, ²University of Utah, Salt Lake City, UT, United States, ³Harvard Medical School, Cambridge, MA, United States, ⁴Brigham Young University, Provo, UT, United States, ⁵University of Florida, Gainesville, FL, United States

Keywords: Data Analysis, Brain

Normative modeling offers a promising approach for better characterization of brain differences in heterogeneous populations. In this study, empirical normative modeling was applied to characterize white matter regional heterogeneity of DTI measurements in adolescents and young adults with more cognitively-and-verbally able autism spectrum disorder (ASD) in comparison to an age matched cohort of typically developing controls. The results of this study demonstrated that the individual differences in DTI measurements in many JHU white matter regions are extremely heterogeneous across the ASD cohort. Hierarchical clustering analysis revealed three groups of white matter regions with similarity patterns across six subgroups within ASD.

View Presentation Video

Keywords: Data Analysis, Multimodal

Multiparametric MRI (mpMRI) has shown a marked impact on prostate cancer (PCa) diagnosis, but the understanding of the spatial distribution of tumors and the correlation between tumor location and mpMRI performance is still limited. The purpose of this study was to determine the performance of mpMRI for PCa detection and visualize the results on a prostate sector map described by Prostate Imaging Reporting and Data System (PI-RADS) v2.1. Also, this study aimed to investigate the association between mpMRI performance and tumor location in the prostate.

Jia Yuping¹ and Dou Weiqiang^2
1Lixia District, Jinan city, The First Affiliated Hospital of Shandong First Medical University & Shandong Provincial Qianfoshan, Jinan, China, ²MR Research, GE Healthcare, Beijing, China

Keywords: Data Analysis, Cancer

This study aimed to use intravoxel incoherent motion (IVIM) DWI with different mathematical models to predict rectal adenomas with and without canceration. The parameters of different IVIM-DWI models, including apparent diffusion coefficient (ADC) from mono-exponential model, true diffusion coefficient, pseudo-diffusion coefficient and perfusion fraction from bi-exponential model, and the distributed diffusion coefficient and water molecular diffusion heterogeneity index from stretched-exponential model were compared between 31 adenoma and 29 adenoma with canceration. We found that mono-exponential derived ADC can easily predict rectal adenomas with canceration, and bi-exponential model has a better combination of sensitivity and specificity for diagnosing rectal adenoma canceration.

Run Xu¹, Guanwu Li¹, Dan YU², Yongming Dai², and Xinyue Liang^2
1Department of Radiology, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China, ²Central Research Institute, United Imaging Healthcare, Shanghai, China

Keywords: Data Analysis, Cancer

In this work, multiparametric MRI was applied to reconstruct spatial habitats and validate the association with the molecular subtypes of breast cancer. By combining perfusion and diffusion characteristics, three habitats were constructed and assigned: hypervascular habitat, hypovascular cellular habitat, and nonviable habitat. In triple-negative breast cancer (TNBC), the volume fraction is lower for hypervascular habitat and higher for nonviable habitat, with respect to non-TNBC. 

Junying Cheng¹, Qian Zheng², Man Xu¹, Liang Liu¹, Yan Cui¹, Biaoshui Liu³, Yong Zhang¹, Yanqiu Feng⁴, and Jingliang Cheng^1
1Department of MRI, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China, ²College of Software Engineering, Zhengzhou University of Light Industry, Zhengzhou, China, ³Department of Radiation Oncology, Sun Yat-sen University CancerCenter, Guangzhou, China, ⁴School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, China

Keywords: Data Processing, Data Processing, phase imaging, phase unwrapping, Dixon technique

In this work, a novel robust and accelerated phase-unwrapping method is presented. The proposed method firstly introduces an artificial volume compartmentalization to break down the large-scale unwrapping problems, and then uses the phase partition method to cluster the phase into blocks to be paralleled unwrapped first, and residual-voxel to be unwrapped later. The simulated and in vivo datasets experiments have demonstrated that the proposed method allows for a reduction of the unwrapping problem size, a speed-up for handling large datasets, and obtains the accurate phase results under different SNRs and phase-change levels. 

View Presentation Video

Zihao Tang^1,2, Xinyi Wang², Mariano Cabezas¹, Lihaowen Zhu², Dongnan Liu^1,2, Michael Barnett^1,3, Weidong Cai², and Chenyu Wang^1,3
1Brain and Mind Centre, University of Sydney, Camperdown, Australia, ²School of Computer Science, University of Sydney, Camperdown, Australia, ³Sydney Neuroimaging Analysis Centre, Camperdown, Australia

Keywords: Data Analysis, Diffusion Tensor Imaging

Diffusion Weighted Imaging (DWI) can be disrupted due to acquisition constraints or imaging artifacts, which can lead to unreliable scalar metrics calculated and valuable scans discarded as a result. To reduce the impact of disrupted brain regions in DTI, we adapted a deep learning DTI inpainting network to reconstruct the disrupted  ROIs. We evaluated the performance of the method by calculating the Fractional Anisotropy errors according to each individual brain region. Experimental results show that the inpainting method can reconstruct the relevant clinical imaging information by mitigating the Fractional Anisotropy differences overall and in individual disrupted brain regions.

View Presentation Video

Daiki Tamada¹, Rianne A Van der Heijden¹, Diego Hernando^1,2,3,4, and Scott B Reeder^1,2,4,5,6
1Radiology, University of Wisconsin-Madison, Madison, WI, United States, ²Medical Physics, University of Wisconsin-Madison, Madison, WI, United States, ³Electrical and Computer Engineering, University of Wisconsin-Madison, Madison, WI, United States, ⁴Biomedical Engineering, University of Wisconsin-Madison, Madison, WI, United States, ⁵Medicine, University of Wisconsin-Madison, Madison, WI, United States, ⁶Emergency, University of Wisconsin-Madison, Madison, WI, United States

Keywords: Data Processing, Artifacts, CSE-MRI

In this work we develop a confidence map algorithm to identify reliable estimates of PDFF and R2* for CSE-MRI in the liver. Reliability of estimates is evaluated based on a normalized root-mean-square-error (NRMSE) between measured and model signals. Threshold values of the NRMSE are determined by using the Cramér–Rao Lower Bound. Monte-Carlo simulations and clinical CSE-MRI successfully validate the proposed method.

View Presentation Video

Shoujin Huang¹, Sixing Liu¹, Lifeng Mei¹, Chenhui Tang¹, Ed X Wu^2,3, and Mengye Lyu^1
1College of Health Science and Environmental Engineering, ShenZhen Technology University, Shenzhen, China, ²Laboratory of Biomedical Imaging and Signal Processing, The University of Hong Kong, Hong Kong, China, ³Department of Electrical and Electronic Engineering, The University of Hong Kong, Hong Kong, China

Keywords: Data Processing, Modelling, Deep learning, Denoise

In this work, we propose a new denoising method named Cross-Subject Transformer Denoising (CSTD), which transfers the texture of a reference image retrieved from a large database to the noisy image with soft attention mechanisms. The experiments on the fastMRI dataset with various noise levels show that our method is likely superior to many competing denoising algorithms including current the state-of-the-art NAFNet. Moreover, our method exhibits excellent generalizability when directly applied to in-vivo low-field data without retraining. Due to the flexibility, the method is expected to have a wide range of applications. 

View Presentation Video

Risavarshni Thevakumaran^1,2, Marcus Couch^2,3,4, Sridar Narayanan^2,3, Douglas Arnold ^2,3, and David Rudko^1,2,3
1Department of Biological and Biomedical Engineering, McGill University, Montreal, QC, Canada, ²McConnell Brain Imaging Centre, Montreal Neurological Institute and Hospital, Montreal, QC, Canada, ³Department of Neurology and Neurosurgery, McGill University, Montreal, QC, Canada, ⁴Siemens Healthcare Limited, Montreal, QC, Canada

Keywords: Data Processing, High-Field MRI, Quantitative MRI

We designed an MRI protocol for rapid, high-resolution human cortical MTsat (csMTsat) imaging at 7 T. Our approach leveraged the PTx capability of the 7 T Siemens Terra MRI system. Further, we developed a post-processing scheme that reduces residual B1+ field bias and noise in csMTsat maps for imaging cortical myelin in the human cortex at high-resolution, which can be applied to study cortical demyelination in Multiple Sclerosis. 

View Presentation Video

Istvan N Huszar¹, Silei Zhu², Adele Smart², Saad Jbabdi², Karla L Miller², and Amy F.D. Howard^2
1Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, ²University of Oxford, Oxford, United Kingdom

Keywords: Data Processing, Data Processing, registration, microscopy, histology, postmortem, macaque

We built a novel pipeline to register 290 high-resolution multi-contrast microscopy sections in the BigMac dataset to post-mortem MRI with sub-millimetre accuracy. The MRI-microscopy alignment is optimised over all microscopy sections jointly, to achieve high-quality co-registration whilst ensuring the slide order and separation is consistent with how the tissue was sectioned. We demonstrate the accuracy of the registration by overlaying FreeSurfer-estimated grey-white matter contours on orthogonal sections of 3D volumetric reconstructions of each microscopy modality. Our pipeline provides an integrated workflow for diffusion-microscopy studies using the BigMac dataset, which is openly available via the Oxford Digital Brain Bank.

View Presentation Video

M. Okan Irfanoglu¹, Amritha Nayak^1,2, Paul Taylor³, and Carlo Pierpaoli^1
1QMI/NIBIB, National Institutes of Health, Bethesda, MD, United States, ²Henry Jackson Foundation, Bethesda, MD, United States, ³Scientific and Statistical Computing Core, NIMH, National Institutes of Health, Bethesda, MD, United States

Keywords: Data Processing, Diffusion/other diffusion imaging techniques

The processing needs for diffusion MRI data have evolved over the years with data sizes getting larger, diffusion sensitization going higher.  Large multi-site studies, especially on "uncooperative subjects" such as young children or patients with movement disorders increased the necessity for dMRI processing pipelines that are fast, robustly capable of handling a variety of artifacts/distortions, and that have summary reporting capabilities that can pinpoint problematic data. The NIH Diffusion MRI processing pipeline, TORTOISE,  has been reimagined, redesigned and and significantly enriched to satisfy these processing needs.

View Presentation Video

Kadharbatcha S Saleem¹, Alexandru V Avram¹, Daniel Glen², Cecil Chern-Chyi Yen³, Vincent Schram⁴, and Peter J Basser^1
1Section on Quantitative Imaging and Tissue Sciences (SQITS), Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD-NIH), Bethesda, MD, United States, ²Scientific and Statistical Computing Core, National Institute of Mental Health (NIMH-NIH), Bethesda, MD, United States, ³Neuroimaging core, National Institute of Neurological Disorders and Stroke (NINDS-NIH), Bethesda, MD, United States, ⁴Microscopy and Imaging Core (MIC), Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD-NIH), Bethesda, MD, United States

Keywords: Visualization, Ex-Vivo Applications, MAP-MRI, subcortical, histology, Marmoset, and diffusion MRI

Despite its importance as a model for human brain development and neurological disorders, the marmoset lacks a comprehensive MRI-histology-based parcellation of subcortical regions. Here, we mapped the subcortical areas of the marmoset brain in three dimensions (3D) using ultra-high resolution MAP-MRI, T2w, and MTR imaging at 7T, combined with histological stains of the same brain. Our results demonstrate that MAP-MRI can delineate cytoarchitectonic subregions of many deep brain structures observed with histology. Tracing and validating these important brain regions in 3D are imperative for neurosurgical planning, navigation of deep brain stimulation probes, and establishing brain structure-function relationships.

Junshen Xu¹, Daniel Moyer², Borjan Gagoski³, P. Ellen Grant^3,4, Polina Golland⁵, Juan Eugenio Iglesias^4,5,6,7, and Elfar Adalsteinsson^1
1Massachusetts Institute of Technology, Cambridge, MA, United States, ²Computer Science, Vanderbilt University, Nashville, TN, United States, ³Fetal-Neonatal Neuroimaging and Developmental Science Center, Boston Children’s Hospital, Boston, MA, United States, ⁴Harvard Medical School, Boston, MA, United States, ⁵Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA, United States, ⁶Centre for Medical Image Computing, Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ⁷Athinoula A. Martinos Center for Biomedical Imaging, Cambridge, MA, United States

Keywords: Image Reconstruction, Machine Learning/Artificial Intelligence

Volumetric reconstruction of fetal brains from multiple stacks of MR slices is challenging due to severe subject motion and image artifacts. We propose a deep learning method to solve the slice-to-volume reconstruction problem in two stages. First, a Transformer network is used to correct motion between slices by registering the input slices to a 3D canonical space. Second, an implicit neural network reconstructs the 3D volume by learning a continuous 3D representation of the fetal brain from the 2D observations. Results show that our method achieves high reconstruction quality and outperforms existing state-of-the-art methods in presence of severe fetal motion.

View Presentation Video

Sneha Vaishali Senthil^1,2, Brenden Toshihide Kadota¹, Peter Truong¹, and Jamie Patrick Near^1
1Medical Biophysics, Sunnybrook Research Institute, Toronto, ON, Canada, ²Integrated Program in Neuroscience, McGill University, Montreal, QC, Canada

Keywords: Data Processing, Spectroscopy, Frequency drift correction, MRSI

Field drift correction is an important post-processing step in MRS and MRSI that yields considerable improvements in spectral quality and metabolite quantification. While several drift correction methods have been proposed in single-voxel MRS, drift correction in MRSI is more challenging due to the application of phase encoding gradients. In such cases, separately acquired navigator echoes are needed. Here, we demonstrate the use of self-navigating Rosette-MRSI trajectories for characterising scanner magnetic field drift using time-domain spectral registration in-vivo. Results from the study showed substantial increases in SNR and spectral quality of the acquired data.