View Presentation Video

Xiaole Zhong^1,2 and J. Jean Chen^1,2,3
1Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, ²Rotman Research Institute, Baycrest Hospital, Toronto, ON, Canada, ³Institute of Biomedical Engineering, University of Toronto, Toronto, ON, Canada

Keywords: Signal Modeling, Simulations

The R₂’ effect is the foundation of BOLD fMRI contrast. R₂’ is disproportionately sensitive to the presence of macrovasculature, which disrupts the homogeneity of the main magnetic field (B₀), resulting in BOLD signal intensities that may heavily depend on macrovascular orientation and volume but also on vascular position. We simulate the BOLD signal strength in a voxel containing macrovasculature at various vascular positions within the voxel. This simulation highlights an additional source of variability in the macrovascular contributions to R₂’-weighted MRI.

View Presentation Video

Matteo Ferrante¹, Tommaso Boccato², and Nicola Toschi^3,4
1Biomedicine and prevention, University of Rome Tor Vergata, Roma, Italy, ²Biomedicine and prevention, University of Rome Tor Vergata, Rome, Italy, ³BioMedicine and prevention, University of Rome Tor Vergata, Rome, Italy, ⁴Department of Radiology,, Athinoula A. Martinos Center for Biomedical Imaging and Harvard Medical school, Boston, MA, USA, Boston, MA, United States

Keywords: Machine Learning/Artificial Intelligence, Neuroscience, brain decoding, fMRI

In vision, the brain is a feature extractor which works from images. We hypothesize that fMRI can  mimic the latent space of a classifier, and employ deep diffusion models with BOLD data from the occipital cortex to generate images which are plausible and semantically close to the visual stimuli administered during fMRI. To this end, we mapped BOLD signals onto the latent space of a pretrained classifier and used its gradients to condition a generative model to reconstruct images. The semantic fidelity of our BOLD response to visual stimulus reconstruction model is superior to the state of the art.

View Presentation Video

Emma Brouwer^1,2, Wietske van der Zwaag^1,2, and Nikos Priovoulos^1,2
1Spinoza Centre for Neuroimaging, Amsterdam, Netherlands, ²Computational Cognitive Neuroscience and Neuroimaging, Netherlands Institute for Neuroscience, Amsterdam, Netherlands

Keywords: Data Acquisition, Shims, B1 shim

The human cerebellum is underexplored in-vivo due to a lack of acquisition methods that effectively portray it. Its dense architecture necessitates high resolution, which benefits from 7T imaging. Unfortunately, at higher field strengths, images suffer from severely destructive B₁-interference with typical coil geometries. Subject-optimized parallel-transmit approaches can mitigate B₁ artifacts but increase scan-time. We designed a universal B₁-shim specific to the cerebellum and immediately applicable across individuals for high-resolution 7T fMRI. We demonstrate improvements in tSNR and functional responses in cerebellar ROIs during a hand-motor task compared to a quadrature mode shim setting.

View Presentation Video

Dilmini Wijesinghe¹, Danny JJ Wang¹, and Kay Jann^1
1USC Mark and Mary Stevens Neuroimaging and Informatics Institute, Keck School of Medicine at USC, Los Angeles, CA, United States

Keywords: Software Tools, fMRI (resting state), Complexity, Higuchi Fractal Dimension, Hurst Exponent, Lempel-Ziv Complexity, Approximate Entropy, Multiscale Sample Entropy, Fuzzy Entropy, Permutation Entropy

Complexity measures of rs-fMRI signals based on non-linear timeseries analyses have been proposed for quantifying the predictability of fMRI signals. There are multiple mathematical methods for evaluating complexity in timeseries signals. This study evaluates the optimal parameter settings in seven complexity tests by analyzing mean complexity measures of grey matter (GM) and CSF across multiple complexity tests with different parameter settings and different fMRI acquisition protocols. Furthermore, complexity evaluation was performed on surrogate timeseries data. Overall, Multiscale Sample Entropy, Fuzzy Entropy, and Hurst Exponent consistently showed an increase in mean complexity of GM compared to CSF in original data. 

View Presentation Video

Abdoljalil Addeh^1,2, Fernando Vega^1,2, Rebecca J. Williams^2,3,4, Ali Golestani⁵, G. Bruce Pike^2,3,4, and M. Ethan MacDonald ^1,2,3,6
1Department of Biomedical Engineering, Schulich School of Engineering, University of Calgary, Calgary, AB, Canada, ²Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada, ³Department of Radiology, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada, ⁴Department of Clinical Neurosciences, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada, ⁵Department of Medical Physics, Alberta Heath Services, Calgary, AB, Canada, ⁶Department of Electrical & Software Engineering, Schulich School of Engineering, University of Calgary, Calgary, AB, Canada

Keywords: Data Processing, fMRI, Physiological Noise

In many fMRI studies, respiratory signals are unavailable or do not have acceptable quality.  Consequently, the direct removal of low-frequency respiratory variations from BOLD signals is not possible. This study proposes a one-dimensional CNN model for reconstruction of two respiratory measures including RV and RVT. Results show that a CNN can capture informative features from the BOLD signals and reconstruct accurate RV and RVT timeseries.  It is expected that application of the proposed method will lower the cost of fMRI studies, reduce complexity, and decrease the burden on participants because they will not be required to wear a respiratory bellows

View Presentation Video

Pierre-Antoine Comby¹, Zaineb Amor¹, Alexandre Vignaud¹, and Philippe Ciuciu^1
1Neurospin (CEA), Gif-sur-Yvette, France

Keywords: Data Processing, fMRI (task based), Denoising, Benchmak

Local-low-rank denoising in task-based fMRI increases sensitivity to statistical detection of neural activity, without harming specificity.

We compared 5 methods (NORDIC, MP-PCA, Hybrid-PCA, Optimal-Threshold, Hybrid-OT) in 4 preprocessing configuration (denoising on magnitude/complex data, before/after motion correction) and their effect on downstream analysis. For best performance, the denoising shoud be done prior to motion correction, and using complex-valued data is only valuable in some settings.

In average (n=6), up to 8 times more activations can be detected (p < 0.05, controlling for FDR). We also provide open-source implementations for broader use of  Local-Low-Rank denoising methods in fMRI.

View Presentation Video

Scott Peltier¹, Michelle Karker², Kuan Han², Doug Noll², and Zhongming M Liu^2
1Functional MRI Laboratory, University of Michigan, Ann Arbor, MI, United States, ²Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States

Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence

In this work, we consider whether a VAE model trained with volumetric fMRI data (rather than a cortical subset of the data) is capable of encoding fMRI into low-dimensional representations, decoding these representations back into volumetric fMRI space, and also generating new fMRI patterns from the latent space. For 3D VAE model training, validation, and testing, volumetric resting-state fMRI data was used from the Human Connectome Project minimally preprocessed pipeline. We find the 3D VAE is able to accurately represent the spatial and temporal information in the data. In addition, it is able to synthesize realistic resting-state networks.
 

View Presentation Video

Tianyun Zhao¹, Philip Tubiolo^1,2, Thomas Hagan¹, John C. Williams^1,2, Jared Van Snellenberg^2,3,4, and Chuan Huang^1,4
1Biomedical Engineering, Stony Brook University, Stony Brook, NY, United States, ²Psychiatry and Behavioral Health, Renaissance School of Medicine at Stony Brook University, Stony Brook, NY, United States, ³Psychology, Stony Brook Univeristy, Stony Brook, NY, United States, ⁴Radiology, Renaissance School of Medicine at Stony Brook University, Stony Brook, NY, United States

Keywords: Machine Learning/Artificial Intelligence, fMRI (task based)

Deep learning, especially convolutional neural networks (CNN), has been shown to be able to identify the non-linear relation between functional magnetic resonance imaging (fMRI) and task performance. CNN can generate an interpretable result called saliency map highlighting regions that are important for task performance. It can uncover other neural processes that linear modeling cannot due to the high dimensionality nature of the fMRI. The CNN result can be presented as a saliency  Previously, we developed a pipeline to produce the saliency map for working memory tasks. In this work, we further evaluated the repeatability of our pipeline.

View Presentation Video

Nayana Menon¹, Jonathan Polimeni², and J. Jean Chen^1,3,4
1Rotman Research Institute, Baycrest Health Sciences, Toronto, ON, Canada, ²Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Boston, MA, United States, ³Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, ⁴Department of Biomedical Engineering, University of Toronto, Toronto, ON, Canada

Keywords: Data Analysis, White Matter, fMRI, functional connectivity, Fourier analysis, Human Connectome Project

In this study, we illustrate the BOLD fMRI signal characteristics underlying functional correlations in the white matter of healthy adults. We use a group-level white-matter only Pearson’s correlation matrix to identify several high-connectivity regions of interest. Our results show that the BOLD signal spectra of connected white-matter regions contain a low-frequency peak unseen in unconnected regions.

View Presentation Video

Imam Ahmed Shaik¹, Mukund Balasubramanian^2,3, Avery J.L. Berman^4,5, Jonathan R. Polimeni^2,6,7, and William Grissom^1
1Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, ²Harvard Medical School, Boston, MA, United States, ³Boston Children's Hospital,, Boston, MA, United States, ⁴Carleton University, Ottawa, ON, Canada, ⁵University of Ottawa Institute of Mental Health Research, Ottawa, ON, Canada, ⁶5Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital,, Charlestown, MA, United States, ⁷Harvard-MIT Division of Health Sciences and Technology,Massachusetts Institute of Technology, Cambridge, MA, United States

Keywords: Data Acquisition, fMRI

Multishot EPI-based fMRI can achieve high spatial resolutions to resolve functional activity at the level of layers and columns, but suffers sensitivity to motion and phase changes between shots. Reordering the shots in a multislice stack so that each slice’s shots are acquired sequentially (VFA-FLEET-EPI) reduces this sensitivity but requires specialized RF pulses to maintain consistent signal between shots as longitudinal magnetization evolves. In this work we show that designing these pulses jointly using an autodiff optimal control algorithm yields more consistent signal across shots which reduces ghosting compared to VFA sinc and recursively designed SLR pulses. 

View Presentation Video

Wen-Ju Pan¹, Nmachi Anumba¹, Nan Xu¹, Lisa Meyer-Baese¹, and Shella Keilholz^1
1Emory University/Georgia Institute of Technology, Atlanta, GA, United States

Keywords: Data Analysis, Brain

Rodent EPI image qualities may vary across coil types, coil positioning and different animals that challenge atlas registration. We proposed an accurate registration with study-group customized EPI template and initially manual registration with an assistance of the newly-introduced tissue-boundary atlas.  Our studies demonstrated some visible mismatching in local anatomic structures by the standard registration methods for rodent data which were effectively corrected by the presented method.

View Presentation Video

Sina Ghaffarzadeh¹, Vahid Malekian², Faeze Makhsousi³, and Seyyed Ali Seyyedsalehi^3
1Biomedical Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran (Islamic Republic of), ²University College London, London, United Kingdom, ³Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran (Islamic Republic of)

Keywords: Data Processing, Brain

In this study a novel method for sampling the active and noisy areas is proposed by using the purification of gray and non-gray matter areas of fMRI data. Also, a data-driven network is proposed in a parallel, multi-step and integrated manner for optimal noise reduction of t-fMRI data. Besides, the proposed method reduces substantially physiological noise without considering the specific noise source and only by using the ROI of noise and activity. Based on the results, the proposed method provides a more accurate and improved activity map than previous methods, which increases the power of activity analysis in fMRI data.

View Presentation Video

Guoxiang Liu^1,2, Takashi Ueguchi^1,2, and Seiji Ogawa^1,3
1NICT, Osaka, Japan, ²Graduate School of Frontier Biosciences, Osaka University, Osaka, Japan, ³Tohoku Fukushi University, Sendai, Japan

Keywords: Image Reconstruction, fMRI (task based), Highe resolution

Recently, a new noise reduction technique NOise Reduction with Distribution Corrected (NORDIC) PCA has been proposed for High resolution fMRI, and the improvements in keys including temporal-SNR, BOLD detectability have been demonstrated using this technique. In this work, we applied this technique to 0.4-mm isotropic human fMRI data acquired with our Block-interleaved segmented EPI, and compare the BOLD detectability of the data with and without NORDIC-performed.
 

View Presentation Video

Maria Guidi^1,2,3, Giovanni Giulietti^4,5, Harald E. Moeller², David G. Norris³, and Federico Giove^1,4
1MARBILab, Enrico Fermi Research Center, Rome, Italy, ²Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ³Donders Centre for Cognitive Neuroimaging, Radboud University, Nijmegen, Netherlands, ⁴Fondazione Santa Lucia IRCCS, Rome, Italy, ⁵SAIMLAL Department, Sapienza University, Rome, Italy

Keywords: Data Processing, Data Analysis, Denoising, Layers

In this study, we evaluated the effect of common denoising steps (NORDIC, regression for motion parameters, RETROICOR and aCompCor) on a high-resolution resting-state BOLD fMRI dataset. We extracted the temporal standard deviation and the spectral power density at different cortical depths in the primary motor cortex and found that each denoising algorithm had a distinct signature on the profile shape. We further estimated the effect of denoising by calculating the temporal signal-to-noise ratio and delta variation signal (DVARS) for different tissue types and found that NORDIC and aCompCor had the largest impacts on the metrics considered.

View Presentation Video

Viktor Pfaffenrot¹ and David Norris^1,2
1Erwin L. Hahn Institute for Magnetic Resonance Imaging, University of Duisburg-Essen, Essen, Germany, ²Donders Institute for Brain, Cognition and Behavior, Radboud Iniversity Nijmegen, Nijmegen, Netherlands

Keywords: Data Processing, fMRI, denoising, laminar fMRI, NORDIC PCA

High-resolution fMRI is dominated by thermal noise. The NORDIC PCA filter is a promising method to remove thermal noise but its effect on the spatial fidelity of layer-dependent profiles is sparsely investigated. We evaluated the performance of NORDIC with several parameter setting for laminar fMRI at 7 T and sought to investigate the effect NORDIC has on the shape of the profiles. Our results suggest that NORDIC indeed affects high-resolution profiles but the overall effect can be seen to be small and can be further reduced with careful parameter settings.

View Presentation Video

Sung Suk Oh¹, Min Cheol Chang², Soyoung Kwak², Jong-ryul Choi³, N. Jon Shah^4,5,6,7, and Seong Dae Yun^4
1Medical Device Development Center, K-MEDI hub, Daegu, Korea, Republic of, ²Department of Rehabilitation Medicine, Yeungnam University, Daegu, Korea, Republic of, ³K-MEDI hub, Daegu, Korea, Republic of, ⁴Institute of Neuroscience and Medicine 4, INM-4, Forschungszentrum Juelich, Juelich, Germany, ⁵Institute of Neuroscience and Medicine 11, INM-11, Forschungszentrum Juelich, Juelich, Germany, ⁶JARA - BRAIN - Translational Medicine, Aachen, Germany, ⁷Department of Neurology, RWTH Aachen University, Aachen, Germany

Keywords: Data Acquisition, fMRI (task based)

The tactile stimulus of fingers in fMRI can effectively reveal the somatosensory functional region. However, this research requires a high-resolution fMRI technique, enabling a distinct delineation of functional areas for each finger. This work demonstrates the use of high-resolution EPIK (1.25 x 1.25 mm²) for enhanced characterization of somatosensory areas at 3T. EPIK provides improved spatial resolution than EPI (1.72 x 1.72 mm²) with increased brain coverage while keeping the same temporal resolution as EPI. The activation region from EPIK was shown to be more locally specific, thereby enabling a clearer recognition of functional areas for each finger.

View Presentation Video

Salvatore John Torrisi^1,2,3, Congyu Liao⁴, Jennifer Townsend^1,3, and An (Joseph) Vu^1,3
1Radiology, SF VA Medical Center, San Francisco, CA, United States, ²Northern California Institute of Research and Education, San Francisco, CA, United States, ³Radiology, University of California, San Francisco, San Francisco, CA, United States, ⁴Division of Radiological Sciences Laboratory, Stanford University, Palo Alto, CA, United States

Keywords: Pulse Sequence Design, Contrast Mechanisms, Neuro

We demonstrate sub-millimeter generalized Slice Dithered Enhanced Resolution (gSLIDER) for high-resolution spin-echo (SE) fMRI at 3T. Activations were significantly greater than standard SE fMRI, demonstrating the suitability of this method for high-resolution, mesoscale fMRI.

Menghan Feng¹, Yue Zhang², Zeying Wen^3,4, Yuchi Wu², Chengwei Fu³, Kan Deng⁵, Qizhan Lin², and Bo Liu^2
1Xin-Huangpu Joint Innovation Institute of Chinese Medicine in Guangdong Province, Guangzhou, China, ²The Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China, ³Guangzhou University of Chinese Medicine, Guangzhou, China, ⁴The First Affiliated Hospital of Henan University of Chinese Medicine, Zhengzhou, China, ⁵Philips Healthcare, Guangzhou, China

Keywords: fMRI (resting state), Kidney

The prevalence of SD is high in chronic kidney disorder patients who suffering maintenance hemodialysis whereas little is known about the neuropathologic mechanism. We investigated the ALFF alterations between MHDSD patients and HCs and those meaningful features were used for constructing discriminating model based the SVM algorithm to classify the MHDSD patients. We found the aberrant spontaneous activities in the DMN, AN, CEN, and VIN in MHDSD patients. Additionally, the classifier indicated the discriminative ALFF features in the above regions demonstrated good performance. This will contribute to well understanding the neuropathological mechanism and seeking biomarkers for discrimination in MHDSD patients. 

View Presentation Video

Qiu Ge^1,2,3, MinLiang Yao⁴, Juan Yue^1,2,3,4, Xue Yang^1,2,3, YueJiao Ding^1,2,3, Yong Zhang⁵, and Yufeng Zang^1,2,3,4
1Center for Cognition and Brain Disorders, The Affiliated Hospital of Hangzhou Normal University, Hangzhou, China, ²Zhejiang Key Laboratory for Research in Assessment of Cognitive Impairments, HangZhou, China, ³Institute of Psychological Sciences, Hangzhou Normal University, HangZhou, China, ⁴Hangzhou Normal University Affiliated Deqing Hospital, HangZhou, China, ⁵Yong Zhang, GE Healthcare, Shanghai, China

Keywords: Data Acquisition, fMRI

FDA cleared a TMS treatment protocol with fMRI-guided individualized targeting for major depressive disorder. This protocol was based on 3T scanner. However, 1.5 T MRI is more widely available in hospitals. We comprehensively tested the reliability and sensitivity of fMRI metrics  related to TMS treatment on 1.5T and 3T scanners. In general, the sensitivity of 3T is larger than 1.5T. However,  the reproducibility of peak FC location and the activation location, the reliability of RS-fMRI local metrics, are similar for 1.5T and 3T. In conclusion, 1.5 fMRI meets the needs for guiding individualized precise rTMS treatment.

View Presentation Video

R. Allen Waggoner¹, Chisato Suzuki¹, Ken-ichi Ueno¹, and Keiji Tanaka^1
1RIKEN Center for Brain Science, Wako-shi, Saitama, Japan

Keywords: Artifacts, fMRI, Multi-Echo, Multi-Shot

In regions suffering from magnetic susceptibility effects, the use of BOLD optimized echo combination for averaging Single-Shot/Multi-Echo EPI data across echos, leads to voxels with improves signal but reduced BOLD sensitivity.  The use of Multi-shot/Multi-Echo EPI can suppress the susceptibility effects allowing BOLD optimized echo averaging that retains BOLD sensitivity.