ISMRM 2024

Computer #
2793.	17	Mixed-sequence training for deep subspace learning image reconstruction of T1-T2-T2-FF CMR Multitasking data Zheyuan Hu^1,2,3, Tianle Cao^1,2,3, Zihao Chen^1,2,3, Yibin Xie¹, Debiao Li^1,3, and Anthony Christodoulou^1,2,3 ¹Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Department of Radiological Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA, United States, ³Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States Keywords:* Machine Learning/Artificial Intelligence, Cardiovascular Motivation: Multi-parametric mapping using T1-T2-T2-fat fraction (FF) MR Multitasking is promising but is hindered by lengthy reconstruction times. Goal(s): To improve T1-T2-T2-FF Multitasking reconstruction time with deep subspace learning, overcoming challenges in training data scarcity and network scalability to high-dimensional spatial factors. Approach: A component-by-component (CBC) network structure was evaluated for three training strategies: 1) large T1 data, 2) limited T1-T2-T2-FF data, and 3) multi-domain, mixed-sequence learning on both T1 and T1-T2-T2-FF data. Results: Mixed-domain learning demonstrated superior image reconstruction quality, achieving the lowest normalized root mean squared error, displaying fewer structural artifacts, and narrowing Bland-Altman limits of agreement. Impact: Component-by-component deep-subspace-learning image reconstruction with mixed-sequence training can dramatically speed up T1-T2-T2*-fat fraction (FF) MR Multitasking image reconstruction by approximately 600 times, potentially overcoming a major barrier to clinical translation.
2794.	18	RANGR: Deep Learning Autonavigation of Free-Breathing Golden-Angle Radial Abdominal MRI Joel Jose Quitlong Nario^1,2, Victor Murray³, Anthony Mekhanik³, and Ricardo Otazo^1,2,3,4 ¹Weill Cornell Graduate School of Medical Sciences, New York, NY, United States, ²Department of Radiology, NewYork-Presbyterian/Weill Cornell Medical Center, New York, NY, United States, ³Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, United States, ⁴Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, NY, United States Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence Motivation: Current autonavigation methodology for free-breathing MRI methods lacks reliability. Goal(s): Develop deep learning methodology to estimate a motion signal directly from the acquired data without manually tuned filtering or PCA transformation. Approach: RANGR uses an encoder network based on the popular VGG architecture to estimate a 1-D respiratory navigator signal from 1-D projections extracted directly from the data. Results: RANGR improved motion estimation and results on motion-resolved images with reduced artifacts, and was even able to detect motion even in cases where filtering+PCA completely failed. Impact: The improved robustness and automation presented by RANGR can promote the use of free-breathing motion-resolved imaging for both diagnostic and treatment guidance purposes.
2795.	19	Deep Learning with Spatio-Channel Regularization for Accelerated Cardiac Cine Omer Burak Demirel¹, Fahime Ghanbari¹, Manuel A Morales¹, Patrick Pierce¹, Scott Johnson¹, Jennifer Rodriguez¹, Jordan A Street¹, Warren J Manning^1,2, and Reza Nezafat¹ ¹Department of Medicine (Cardiovascular Division), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States, ²Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States Keywords: Machine Learning/Artificial Intelligence, Cardiovascular Motivation: Evaluation of cardiac function with cine imaging remains long and requires repeated breath-holds that are sometimes corrupted with artifacts if patients have non-sinus rhythm or difficulty in breath-holding. Goal(s): To develop a deep learning method with spatio-channel regularization with multi-channel k-space reconstruction for accelerated cine imaging. Approach: Coil-self consistency based deep learning (DL) was implemented with 3D regularization across spatial and channel dimensions in contrast to single coil-combined image used in sensitivity encoding (SENSE). Results: Our approach at 5-fold acceleration showed quantitative improvements over SENSE-based DL on retrospectively accelerated data and showed good agreement with left ventricular (LV) measurements on prospectively accelerated data. Impact: The spatio-channel regularized DL reconstruction shortens the scan time by a factor of 5, leading to fewer breath-holds and 2–3-minute scans. This can greatly benefit patients struggling with breath-holding and accelerate the overall scan time.
2796.	20	Simultaneously Multi-slice Imaging by the Fusion of Reconstruction and Collecting Under-sampled Signal with Deep Learning (FoCUS) Yuki SATO¹, Naoya ENDO¹, Shohei OUCHI², and Satoshi ITO¹ ¹Utsunomiya University, Utsunomiya, Japan, ²National Institute of Technology, Oyama College, Oyama, Japan Keywords: Machine Learning/Artificial Intelligence, New Trajectories & Spatial Encoding Methods Motivation: Simultaneous multi-slice imaging (SMS) can obtain multiple slice images simultaneously, but it requires the sensitivity distribution of the receiver coils. Goal(s): Our goal was to separate slice images using deep learning reconstruction without coil sensitivity. Approach: Different amplitude modulation is given to each slice, and the CNN separates each slice from the focal image based on the value of the amplitude modulation. Results: Simulation experiments showed that image separation was successfully achieved not only for real-valued images but also for complex-valued images. Image quality decreases when the number of excitation images was increased. Impact: SMS can be used much more easily because it does not require coil sensitivity distribution for image separation. No non-uniform residual noise will be generated. The proposed method may expand the application of SMS.
2797.	21	Deep image-pass filter for dynamic MRI reconstruction Yinghao Zhang¹ and Yue Hu¹ ¹Harbin Institute of Technology, Harbin, China, China Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, deep image prior, unsupervised learning, deep image-pass filter, dynamic MRI Motivation: For accelerating dynamic MRI, the gradient flow in the unsupervised deep image prior (DIP) methods lacks a direct pathway from the image domain to the network parameters, resulting in suboptimal performance. Goal(s): To propose a novel approch to address DIP's drawback for robust dynamic MRI reconstruction. Approach: Deep image-pass filter is proposed, replacing the random noise input of DIP with learnable image and constraint input consistent with output to establish an efficient gradient pathway from image domain. Results: Experimental results in the reconstruction of both long-axis and short-axis dynamic cardiac cine MRI demonstrate that DIPF outperforms DIP and other state-of-the-art unsupervised methods. Impact: We have introduced a new formulation for unsupervised MRI reconstruction, which will drive a series of research around this paradigm, including network architecture design, reconstuction models combining DIPF with other data priors, and more.
2798.	22	Attention-based two-stage network for non-cartesian multi-coil ASL MRI reconstruction Yanchen Guo¹, Shichun Chen¹, Zhao Li², Manuel Taso³, David C. Alsop³, and Weiying Dai¹ ¹Computer Science, State University of New York at Binghamton, Vestal, NY, United States, ²Zhejiang University, Zhejiang, China, ³Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence Motivation: High-resolution arterial spin labeling (ASL) imaging is time-consuming, limiting its clinical applications in studying small brain structures. Goal(s): To reconstruct high-resolution ASL images from 8-time accelerated ASL image acquisition, an under-sampled non-Cartesian k-space sampling. Approach: We proposed an attention-based deep learning (DL) model. Results: The proposed DL model can successfully reconstruct 8-fold under-sampled, non-cartesian, multi-coil data from k-space. Impact: Our proposed attention-based deep learning model can reconstruct under-sampled non-cartesian multi-coil data in k-space and thereby significantly decrease long MRI acquisition time required for high-resolution ASL MRI imaging, which may enable clinical applications in studying small brain structures.
2799.	23	Image reconstruction performance using mixed real and synthetic MR phase training data Nikhil Deveshwar^1,2, Erin Argentieri¹, Abhejit Rajagopal¹, Sharmila Majumdar¹, and Peder E.Z. Larson¹ ¹Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States, ²Electrical Engineering and Computer Science, University of California, Berkeley, Berkeley, CA, United States Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction Motivation: Prospectively developing MRI datasets with raw-kspace for MRI reconstruction is difficult, expensive and time-consuming. Generating synthetic k-space with synthetic phase as training data has been shown to work comparably for training MRI reconstruction models but its hard to access paired synthetic data. Goal(s): How does training data consisting of mixed real and synthetic k-space (including synthetic phase) affect image reconstruction performance. Approach: Five variational networks were trained with varying amounts of mixed real and synthetic training data. Image quality metrics were used to evaluate the quality of reconstructed images. Results: Adding small amounts of real training data helps increase reconstruction performance. Impact: The results suggest that small addition of real training data in addition to using mostly synthetic training data can help reconstruction performance. This could be useful clinically where synthetic data can augment models trained with small amounts of real data.
2800.	24	Robust Magnetic Resonance Reconstruction by Alternating Deep Low-Rank Approach Yihui Huang¹, Zi Wang¹, Xinlin Zhang², Meijin Lin³, Di Guo⁴, and Xiaobo Qu^1,5 ¹Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen, China, ²College of Physics and Information Engineering, Fuzhou University, Fuzhou, China, ³Department of Applied Marine Physics and Engineering, Xiamen University, Xiamen, China, ⁴School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, China, ⁵Institute of Artificial Intelligence, Xiamen University, Xiamen, China Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction Motivation: Magnetic resonance reconstruction by deep learning is heavily compromised due to the mismatch between the training and target data, such as the sampling rate of undersampling, the organ and the contrast of imaging. Goal(s): Reliablely reconstruct magnetic resonance signal in multiple scenes by one trained deep learning model Approach: Alternating Deep Low-Rank, which combines deep learning solvers and classic low-rank optimization solvers. Results: Compared with state-of-the-art deep learning methods HDSLR and ODLS, one ADLR trained by coronal PDw knee can provide a lower reconstruction error by about 10% in coronal PDw knees, 15% in sagittal PDw knees, and 30% in axial T2w brains. Impact: The proposed ADLR can effectively alleviate the drop in reconstruction quality due to the mismatches of attributes between training and target signals of the MR imaging or MR spectroscopy.
2801.	25	Denoising very low field magnetic resonance images using native noise modeling and deep learning Tonny Ssentamu¹, Ronald Omoding², Atamba Edgar ¹, Pius kabanda Mukwaya ¹, Jjuuko George William¹, Alvin Bagetuuma Kimbowa ², and Sairam Geethanath³ ¹Department of physiology, Makerere University, Kampala, Uganda, ²Department of Electrical and Computer Engineering, Makerere University, Kampala, Uganda, ³Accessible MR Laboratory, Icahn School of Medicine at Mount Sinai, New York, NY, United States Keywords: Machine Learning/Artificial Intelligence, Low-Field MRI, Denoising, Native noise Motivation: Low-field MRI (LF-MRI) can increase accessibility in low-income countries where high-field MRI is not available due to cost, power and siting requirements. However, noise significantly affects LF-MR image quality. Goal(s): This study aims to enhance the signal-to-noise ratio (SNR) in very LF-MRI (0.05T) images using native noise modeling and deep learning. Approach: We extracted noise from 0.05T phantom MRI images, modeled it, added it to high-field brain MRI (1.5T & 3T), trained two deep-learning algorithms, and evaluated them on in vivo brain MRI images. Results: Our approach improves the SNR of in-vivo LF images by a factor of approximately two. Impact: Using native noise while developing deep-learning denoising algorithms for LF-MRI images is better than using synthetic random noise. As a result, the developed algorithms are more explainable and follow domain knowledge on noise in LF-MRI improving trust in the models.
2802.	26	Latent-Optimized Adversarial Regularizers for accelerated MRI Huayu Wang¹, Jing Cheng², Chen Luo¹, Taofeng Xie^1,3, Qiyu Jin¹, Zhuo-xu Cui⁴, and Dong Liang⁴ ¹School of Mathematical Sciences, Inner Mongolia University, Hohhot, China, ²Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ³Inner Mongolia Medical University, Hohhot, China, ⁴Reasearch Center for Medical AI, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction, regularization method, interpretability, adversarial training Motivation: Introduce latent optimization techniques to enhance the interpretability of learnable regularization methods, thereby improving the performance of MRI acceleration reconstruction. Goal(s): Theoretically, we aim to elucidate the iterative direction of learnable regularization methods. Experimentally, we aim to achieve high-quality reconstruction of undersampled MRI data. Approach: Revise the optimization objective of the network by incorporating a stochastic gradient descent generator, training learnable regularizers that guide the latent process during iteration, and accomplish reconstruction using the projected gradient method. Results: Compared to other regularization methods, proposed method achieved a higher level of interpretability and accomplished higher-quality reconstruction. Impact: The method directly learns the distribution information of real data and guides the iteration towards the real data manifold. We believe that the method and its theoretical properties are undoubtedly inspiring for researchers seeking to further acquire data distribution information.
2803.	27	High-Resolution Deep Learning Reconstruction (HR-DLR) to Improve Sharpness in Diffusion Weighted Imaging Kensuke Shinoda¹, Shun Uematsu¹, Yuki Takai¹, and Hideaki Kutsuna¹ ¹MRI Systems Development Department, Canon Medical Systems Corporation, Tochigi, Japan Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, Super Resolution Motivation: Echo planar diffusion weighted imaging (EPI-DWI) often suffers from Gibbs ringing artifact and/or image blurring, because of limited matrix size. A recently proposed High-Resolution Deep Learning Reconstruction (HR-DLR) may bring a breakthrough to the limitation. Goal(s): Our goal was to test benefits of HR-DLR when applied to brain EPI-DWI. Approach: HR-DLR was compared to conventional reconstruction method (zero-filling interpolation[ZIP] and low-pass filtering) with regards to image sharpness and ringing artifact suppression, with a conventional and an accelerated scan conditions. Results: The advantage of HR-DLR over the conventional method was confirmed by measurements of edge slope width (ESW) and ringing variable mean (RVM). Impact: A recently proposed High-Resolution Deep Learning Reconstruction successfully improved the sharpness of single shot EPI-DWI while suppressing Gibbs artifacts. The method could help improve clinical confidence by increasing image resolution and gain examination throughput by shortening acquisition time.
2804.	28	Deep Learning-based Reconstruction of Accelerated MR Cholangiopancreatography Jinho Kim^1,2, Marcel Dominik Nickel², and Florian Knoll¹ ¹Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ²MR Application Pre-development, Siemens Healthineers AG, Erlangen, Germany Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction Motivation: We address the issue of long scan times in MR Cholangiopancreatography (MRCP), which often leads to poor image quality. Goal(s): We aim to leverage a Deep Learning-based model to accelerate MRCP acquisition. Approach: We acquired two-times parallel imaging accelerated MRCP data at 3T, trained a variational network with retrospective undersampling to a total acceleration factor of 6, and then tested the trained model with both retrospective and prospective 6-times accelerated data, acquired at both 3T and 0.55T. Results: The trained model shows potential to improve MRCP by reducing artifacts and enhancing distal ducts compared to parallel imaging and compressed sensing. Impact: The proposed method effectively removes artifacts in highly accelerated MRCP, shortening scan times from 303 seconds to 138 seconds. Moreover, the corresponding SNR enhancement enables MRCP acquisitions at 0.55T, where traditional image reconstruction methods face challenges.
2805.	29	Robust Partial Fourier Reconstruction with Zero-shot Deep Untrained Generative Prior So Hyun Kang¹, Jihoo Kim¹, JaeJin Cho^2,3, Clarissa Z. Cooley^2,3, Berkin Bilgic^2,3,4, and Tae Hyung Kim¹ ¹Department of Computer Engineering, Hongik University, Seoul, Korea, Republic of, ²Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ³Department of Radiology, Harvard Medical School, Boston, MA, United States, ⁴Harvard/MIT Health Sciences and Technology, Cambridge, MA, United States Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, Image Reconstruction, Partial Fourier, Brain, Multi-echo MRI, Low-field MR Motivation: We introduce a novel partial Fourier reconstruction method. Goal(s): The objective is to enhance partial Fourier reconstruction by integrating the traditional phase constraint with the recent zero-shot deep learning approach. Approach: The proposed method combines the virtual conjugate coils (VCC) phase constraint with zero-shot deep untrained generative prior (ZS-DUGP), assuming MRI can be nonlinearly represented by untrained networks, enabling simultaneous image reconstruction and prior learning without external training data. This approach enables robust partial Fourier reconstruction. Results: Evaluation across diverse datasets, including the fastMRI, the QALAS multi-echo data, and the low-field MR data, validates its enhanced performance compared to existing techniques. Impact: We propose a novel partial Fourier reconstruction combining virtual conjugate coils with a zero-shot untrained generative network prior. It provides robust reconstruction without external training dataset, evaluated across various scenarios (parallel imaging, multi-echo/contrast imaging, low-field MR) demonstrating its utility.
2806.	30	CAMERA-NET: Cascade Multi-Level Wavelet neural network with data consistency for MRI Reconstruction Gaojie Zhu^1,2, Xiongjie Shen², and Hua Guo¹ ¹Department of Biomedical Engineering, School of Medicine, Tsinghua University, Center for Biomedical Imaging Research, Beijing, China, ²Anke High-tech Co., Ltd, Shenzhen, China Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence Motivation: The U-net is widely used in deep learning-based MRI reconstruction. Its encoding-decoding component enlarges receptive field while the pooling and interpolation operation limits the ability to recover sparsely sampled MR signals. Goal(s): The wavelet transform and inverse wavelet transform are introduced to replace pooling and interpolation operations in order to maintain the spatial information of images during the encoding-decoding process within the neural network. Approach: A cascaded multi-level wavelet neural network with data consistency, termed as CAMERA-Net, is presented for under-sampled MRI reconstruction. Results: CAMERA-Net demonstrates significant enhancement in reconstructing quality with public fastMRI knee dataset. Impact: The improved reconstruction capabilities of CAMERA-Net have the potential to enhance precision and reliability when reconstructing under-sampled MRI data. This could result in more efficient clinical scans.
2807.	31	Model based rEconstruction by Deep Algorithm unrolLing (MEDAL) for fast 3D whole-heart T2 mapping Alberto Di Biase^1,2, Alina Schneider³, Rene Botnar^1,3,4, and Claudia Prieto^1,2,3 ¹MILLENNIUM INSTITUTE FOR INTELLIGENT HEALTHCARE ENGINEERING, Santiago, Chile, ²School of Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile, ³School of Biomedical Engineering, King’s College London, London, United Kingdom, ⁴Institute for Biological and Medical Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, model-based Motivation: T2 mapping provides quantitative myocardial tissue characterization. However, current approaches acquire several 2D contrast images which are then fitted to a model to estimate the T2 values, leading to limited coverage, and long acquisition and reconstruction times. Goal(s): Here we propose to speed up 3D whole-heart T2 mapping using a model-based deep learning unrolling network (MEDAL) that leverages the power of machine learning and physical knowledge. Approach: MEDAL reconstructs the T2 maps directly without reconstructing any intermediate contrast weighted images or fitting. Results: The proposed approach was evaluated in iNAV-based free-breathing 3D T2 mapping 4x accelerated showing promising results. Impact: A novel method for reconstructing parametric maps using a model-based deep learning unrolling network is presented. The method was demonstrated in a highly accelerated free breathing 3D whole-heart T2 mapping sequence allowing for fast and accurate T2 measurements.
2808.	32	DeepSepSTI: Improved Susceptibility Tensor Reconstruction by Anisotropic Susceptibility Source Separation Zhenghan Fang¹, Hyeong-Geol Shin^2,3, Blake E. Dewey⁴, Peter A. Calabresi⁴, Peter van Zijl^1,2,3, Jeremias Sulam¹, and Xu Li^2,3 ¹Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, United States, ²Department of Radiology and Radiological Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, United States, ³F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States, ⁴Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, MD, United States Keywords: Machine Learning/Artificial Intelligence, Brain, Susceptibility Tensor Imaging, Susceptibility Source Separation Motivation: Magnetic susceptibility source separation has potential for characterizing pathological tissue changes in disease. However, existing source separation methods assume isotropic susceptibility, ignoring anisotropy in white matter. Goal(s): To develop a method for anisotropic susceptibility source separation for better susceptibility tensor reconstruction. Approach: The paramagnetic susceptibility, modeled by an isotropic scalar, and the diamagnetic susceptibility, modeled by an anisotropic tensor, are jointly estimated in each voxel from local frequency and R2’ measurements using a deep learning model, named DeepSepSTI. Results: DeepSepSTI shows generally improved estimation of susceptibility tensors, anisotropy and PEV than DeepSTI. DeepSepSTI can better describe tissue characteristics in multiple sclerosis lesions. Impact: The proposed DeepSepSTI approach may help better measure changes in iron, myelin, and susceptibility anisotropy in various neurological diseases such as multiple sclerosis, potentially providing improved biomarkers for better characterization of disease stage and progression.

Computer #

2793.

Mixed-sequence training for deep subspace learning image reconstruction of T1-T2-T2*-FF CMR Multitasking data

Zheyuan Hu^1,2,3, Tianle Cao^1,2,3, Zihao Chen^1,2,3, Yibin Xie¹, Debiao Li^1,3, and Anthony Christodoulou^1,2,3
¹Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Department of Radiological Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA, United States, ³Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States

Keywords: Machine Learning/Artificial Intelligence, Cardiovascular

Motivation: Multi-parametric mapping using T1-T2-T2*-fat fraction (FF) MR Multitasking is promising but is hindered by lengthy reconstruction times.

Goal(s): To improve T1-T2-T2*-FF Multitasking reconstruction time with deep subspace learning, overcoming challenges in training data scarcity and network scalability to high-dimensional spatial factors.

Approach: A component-by-component (CBC) network structure was evaluated for three training strategies: 1) large T1 data, 2) limited T1-T2-T2*-FF data, and 3) multi-domain, mixed-sequence learning on both T1 and T1-T2-T2*-FF data.

Results: Mixed-domain learning demonstrated superior image reconstruction quality, achieving the lowest normalized root mean squared error, displaying fewer structural artifacts, and narrowing Bland-Altman limits of agreement.

Impact: Component-by-component deep-subspace-learning image reconstruction with mixed-sequence training can dramatically speed up T1-T2-T2*-fat fraction (FF) MR Multitasking image reconstruction by approximately 600 times, potentially overcoming a major barrier to clinical translation.

2794.

RANGR: Deep Learning Autonavigation of Free-Breathing Golden-Angle Radial Abdominal MRI

Joel Jose Quitlong Nario^1,2, Victor Murray³, Anthony Mekhanik³, and Ricardo Otazo^1,2,3,4
¹Weill Cornell Graduate School of Medical Sciences, New York, NY, United States, ²Department of Radiology, NewYork-Presbyterian/Weill Cornell Medical Center, New York, NY, United States, ³Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, United States, ⁴Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, NY, United States

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

Motivation: Current autonavigation methodology for free-breathing MRI methods lacks reliability.

Goal(s): Develop deep learning methodology to estimate a motion signal directly from the acquired data without manually tuned filtering or PCA transformation.

Approach: RANGR uses an encoder network based on the popular VGG architecture to estimate a 1-D respiratory navigator signal from 1-D projections extracted directly from the data.

Results: RANGR improved motion estimation and results on motion-resolved images with reduced artifacts, and was even able to detect motion even in cases where filtering+PCA completely failed.

Impact: The improved robustness and automation presented by RANGR can promote the use of free-breathing motion-resolved imaging for both diagnostic and treatment guidance purposes.

2795.

Deep Learning with Spatio-Channel Regularization for Accelerated Cardiac Cine

Omer Burak Demirel¹, Fahime Ghanbari¹, Manuel A Morales¹, Patrick Pierce¹, Scott Johnson¹, Jennifer Rodriguez¹, Jordan A Street¹, Warren J Manning^1,2, and Reza Nezafat¹
¹Department of Medicine (Cardiovascular Division), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States, ²Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States

Keywords: Machine Learning/Artificial Intelligence, Cardiovascular

Motivation: Evaluation of cardiac function with cine imaging remains long and requires repeated breath-holds that are sometimes corrupted with artifacts if patients have non-sinus rhythm or difficulty in breath-holding.

Goal(s): To develop a deep learning method with spatio-channel regularization with multi-channel k-space reconstruction for accelerated cine imaging.

Approach: Coil-self consistency based deep learning (DL) was implemented with 3D regularization across spatial and channel dimensions in contrast to single coil-combined image used in sensitivity encoding (SENSE).

Results: Our approach at 5-fold acceleration showed quantitative improvements over SENSE-based DL on retrospectively accelerated data and showed good agreement with left ventricular (LV) measurements on prospectively accelerated data.

Impact: The spatio-channel regularized DL reconstruction shortens the scan time by a factor of 5, leading to fewer breath-holds and 2–3-minute scans. This can greatly benefit patients struggling with breath-holding and accelerate the overall scan time.

2796.

Simultaneously Multi-slice Imaging by the Fusion of Reconstruction and Collecting Under-sampled Signal with Deep Learning (FoCUS)

Yuki SATO¹, Naoya ENDO¹, Shohei OUCHI², and Satoshi ITO¹
¹Utsunomiya University, Utsunomiya, Japan, ²National Institute of Technology, Oyama College, Oyama, Japan

Keywords: Machine Learning/Artificial Intelligence, New Trajectories & Spatial Encoding Methods

Motivation: Simultaneous multi-slice imaging (SMS) can obtain multiple slice images simultaneously, but it requires the sensitivity distribution of the receiver coils.

Goal(s): Our goal was to separate slice images using deep learning reconstruction without coil sensitivity.

Approach: Different amplitude modulation is given to each slice, and the CNN separates each slice from the focal image based on the value of the amplitude modulation.

Results: Simulation experiments showed that image separation was successfully achieved not only for real-valued images but also for complex-valued images. Image quality decreases when the number of excitation images was increased.

Impact: SMS can be used much more easily because it does not require coil sensitivity distribution for image separation. No non-uniform residual noise will be generated. The proposed method may expand the application of SMS.

2797.

Deep image-pass filter for dynamic MRI reconstruction

Yinghao Zhang¹ and Yue Hu¹
¹Harbin Institute of Technology, Harbin, China, China

Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, deep image prior, unsupervised learning, deep image-pass filter, dynamic MRI

Motivation: For accelerating dynamic MRI, the gradient flow in the unsupervised deep image prior (DIP) methods lacks a direct pathway from the image domain to the network parameters, resulting in suboptimal performance.

Goal(s): To propose a novel approch to address DIP's drawback for robust dynamic MRI reconstruction.

Approach: Deep image-pass filter is proposed, replacing the random noise input of DIP with learnable image and constraint input consistent with output to establish an efficient gradient pathway from image domain.

Results: Experimental results in the reconstruction of both long-axis and short-axis dynamic cardiac cine MRI demonstrate that DIPF outperforms DIP and other state-of-the-art unsupervised methods.

Impact: We have introduced a new formulation for unsupervised MRI reconstruction, which will drive a series of research around this paradigm, including network architecture design, reconstuction models combining DIPF with other data priors, and more.

2798.

Attention-based two-stage network for non-cartesian multi-coil ASL MRI reconstruction

Yanchen Guo¹, Shichun Chen¹, Zhao Li², Manuel Taso³, David C. Alsop³, and Weiying Dai¹
¹Computer Science, State University of New York at Binghamton, Vestal, NY, United States, ²Zhejiang University, Zhejiang, China, ³Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States

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

Motivation: High-resolution arterial spin labeling (ASL) imaging is time-consuming, limiting its clinical applications in studying small brain structures.

Goal(s): To reconstruct high-resolution ASL images from 8-time accelerated ASL image acquisition, an under-sampled non-Cartesian k-space sampling.

Approach: We proposed an attention-based deep learning (DL) model.

Results: The proposed DL model can successfully reconstruct 8-fold under-sampled, non-cartesian, multi-coil data from k-space.

Impact: Our proposed attention-based deep learning model can reconstruct under-sampled non-cartesian multi-coil data in k-space and thereby significantly decrease long MRI acquisition time required for high-resolution ASL MRI imaging, which may enable clinical applications in studying small brain structures.

2799.

Image reconstruction performance using mixed real and synthetic MR phase training data

Nikhil Deveshwar^1,2, Erin Argentieri¹, Abhejit Rajagopal¹, Sharmila Majumdar¹, and Peder E.Z. Larson¹
¹Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States, ²Electrical Engineering and Computer Science, University of California, Berkeley, Berkeley, CA, United States

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction

Motivation: Prospectively developing MRI datasets with raw-kspace for MRI reconstruction is difficult, expensive and time-consuming. Generating synthetic k-space with synthetic phase as training data has been shown to work comparably for training MRI reconstruction models but its hard to access paired synthetic data.

Goal(s): How does training data consisting of mixed real and synthetic k-space (including synthetic phase) affect image reconstruction performance.

Approach: Five variational networks were trained with varying amounts of mixed real and synthetic training data. Image quality metrics were used to evaluate the quality of reconstructed images.

Results: Adding small amounts of real training data helps increase reconstruction performance.

Impact: The results suggest that small addition of real training data in addition to using mostly synthetic training data can help reconstruction performance. This could be useful clinically where synthetic data can augment models trained with small amounts of real data.

2800.

Robust Magnetic Resonance Reconstruction by Alternating Deep Low-Rank Approach

Yihui Huang¹, Zi Wang¹, Xinlin Zhang², Meijin Lin³, Di Guo⁴, and Xiaobo Qu^1,5
¹Department of Electronic Science, Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance, Xiamen University, Xiamen, China, ²College of Physics and Information Engineering, Fuzhou University, Fuzhou, China, ³Department of Applied Marine Physics and Engineering, Xiamen University, Xiamen, China, ⁴School of Computer and Information Engineering, Xiamen University of Technology, Xiamen, China, ⁵Institute of Artificial Intelligence, Xiamen University, Xiamen, China

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction

Motivation: Magnetic resonance reconstruction by deep learning is heavily compromised due to the mismatch between the training and target data, such as the sampling rate of undersampling, the organ and the contrast of imaging.

Goal(s): Reliablely reconstruct magnetic resonance signal in multiple scenes by one trained deep learning model

Approach: Alternating Deep Low-Rank, which combines deep learning solvers and classic low-rank optimization solvers.

Results: Compared with state-of-the-art deep learning methods HDSLR and ODLS, one ADLR trained by coronal PDw knee can provide a lower reconstruction error by about 10% in coronal PDw knees, 15% in sagittal PDw knees, and 30% in axial T2w brains.

Impact: The proposed ADLR can effectively alleviate the drop in reconstruction quality due to the mismatches of attributes between training and target signals of the MR imaging or MR spectroscopy.

2801.

Denoising very low field magnetic resonance images using native noise modeling and deep learning

Tonny Ssentamu¹, Ronald Omoding², Atamba Edgar ¹, Pius kabanda Mukwaya ¹, Jjuuko George William¹, Alvin Bagetuuma Kimbowa ², and Sairam Geethanath³
¹Department of physiology, Makerere University, Kampala, Uganda, ²Department of Electrical and Computer Engineering, Makerere University, Kampala, Uganda, ³Accessible MR Laboratory, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Keywords: Machine Learning/Artificial Intelligence, Low-Field MRI, Denoising, Native noise

Motivation: Low-field MRI (LF-MRI) can increase accessibility in low-income countries where high-field MRI is not available due to cost, power and siting requirements. However, noise significantly affects LF-MR image quality.

Goal(s): This study aims to enhance the signal-to-noise ratio (SNR) in very LF-MRI (0.05T) images using native noise modeling and deep learning.

Approach: We extracted noise from 0.05T phantom MRI images, modeled it, added it to high-field brain MRI (1.5T & 3T), trained two deep-learning algorithms, and evaluated them on in vivo brain MRI images.

Results: Our approach improves the SNR of in-vivo LF images by a factor of approximately two.

Impact: Using native noise while developing deep-learning denoising algorithms for LF-MRI images is better than using synthetic random noise. As a result, the developed algorithms are more explainable and follow domain knowledge on noise in LF-MRI improving trust in the models.

2802.

Latent-Optimized Adversarial Regularizers for accelerated MRI

Huayu Wang¹, Jing Cheng², Chen Luo¹, Taofeng Xie^1,3, Qiyu Jin¹, Zhuo-xu Cui⁴, and Dong Liang⁴
¹School of Mathematical Sciences, Inner Mongolia University, Hohhot, China, ²Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China, ³Inner Mongolia Medical University, Hohhot, China, ⁴Reasearch Center for Medical AI, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction, regularization method, interpretability, adversarial training

Motivation: Introduce latent optimization techniques to enhance the interpretability of learnable regularization methods, thereby improving the performance of MRI acceleration reconstruction.

Goal(s): Theoretically, we aim to elucidate the iterative direction of learnable regularization methods. Experimentally, we aim to achieve high-quality reconstruction of undersampled MRI data.

Approach: Revise the optimization objective of the network by incorporating a stochastic gradient descent generator, training learnable regularizers that guide the latent process during iteration, and accomplish reconstruction using the projected gradient method.

Results: Compared to other regularization methods, proposed method achieved a higher level of interpretability and accomplished higher-quality reconstruction.

Impact: The method directly learns the distribution information of real data and guides the iteration towards the real data manifold. We believe that the method and its theoretical properties are undoubtedly inspiring for researchers seeking to further acquire data distribution information.

2803.

High-Resolution Deep Learning Reconstruction (HR-DLR) to Improve Sharpness in Diffusion Weighted Imaging

Kensuke Shinoda¹, Shun Uematsu¹, Yuki Takai¹, and Hideaki Kutsuna¹
¹MRI Systems Development Department, Canon Medical Systems Corporation, Tochigi, Japan

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

Motivation: Echo planar diffusion weighted imaging (EPI-DWI) often suffers from Gibbs ringing artifact and/or image blurring, because of limited matrix size. A recently proposed High-Resolution Deep Learning Reconstruction (HR-DLR) may bring a breakthrough to the limitation.

Goal(s): Our goal was to test benefits of HR-DLR when applied to brain EPI-DWI.

Approach: HR-DLR was compared to conventional reconstruction method (zero-filling interpolation[ZIP] and low-pass filtering) with regards to image sharpness and ringing artifact suppression, with a conventional and an accelerated scan conditions.

Results: The advantage of HR-DLR over the conventional method was confirmed by measurements of edge slope width (ESW) and ringing variable mean (RVM).

Impact: A recently proposed High-Resolution Deep Learning Reconstruction successfully improved the sharpness of single shot EPI-DWI while suppressing Gibbs artifacts. The method could help improve clinical confidence by increasing image resolution and gain examination throughput by shortening acquisition time.

2804.

Deep Learning-based Reconstruction of Accelerated MR Cholangiopancreatography

Jinho Kim^1,2, Marcel Dominik Nickel², and Florian Knoll¹
¹Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ²MR Application Pre-development, Siemens Healthineers AG, Erlangen, Germany

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction

Motivation: We address the issue of long scan times in MR Cholangiopancreatography (MRCP), which often leads to poor image quality.

Goal(s): We aim to leverage a Deep Learning-based model to accelerate MRCP acquisition.

Approach: We acquired two-times parallel imaging accelerated MRCP data at 3T, trained a variational network with retrospective undersampling to a total acceleration factor of 6, and then tested the trained model with both retrospective and prospective 6-times accelerated data, acquired at both 3T and 0.55T.

Results: The trained model shows potential to improve MRCP by reducing artifacts and enhancing distal ducts compared to parallel imaging and compressed sensing.

Impact: The proposed method effectively removes artifacts in highly accelerated MRCP, shortening scan times from 303 seconds to 138 seconds. Moreover, the corresponding SNR enhancement enables MRCP acquisitions at 0.55T, where traditional image reconstruction methods face challenges.

2805.

Robust Partial Fourier Reconstruction with Zero-shot Deep Untrained Generative Prior

So Hyun Kang*¹, Jihoo Kim*¹, JaeJin Cho^2,3, Clarissa Z. Cooley^2,3, Berkin Bilgic^2,3,4, and Tae Hyung Kim¹
¹Department of Computer Engineering, Hongik University, Seoul, Korea, Republic of, ²Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ³Department of Radiology, Harvard Medical School, Boston, MA, United States, ⁴Harvard/MIT Health Sciences and Technology, Cambridge, MA, United States

Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, Image Reconstruction, Partial Fourier, Brain, Multi-echo MRI, Low-field MR

Motivation: We introduce a novel partial Fourier reconstruction method.

Goal(s): The objective is to enhance partial Fourier reconstruction by integrating the traditional phase constraint with the recent zero-shot deep learning approach.

Approach: The proposed method combines the virtual conjugate coils (VCC) phase constraint with zero-shot deep untrained generative prior (ZS-DUGP), assuming MRI can be nonlinearly represented by untrained networks, enabling simultaneous image reconstruction and prior learning without external training data. This approach enables robust partial Fourier reconstruction.

Results: Evaluation across diverse datasets, including the fastMRI, the QALAS multi-echo data, and the low-field MR data, validates its enhanced performance compared to existing techniques.

Impact: We propose a novel partial Fourier reconstruction combining virtual conjugate coils with a zero-shot untrained generative network prior. It provides robust reconstruction without external training dataset, evaluated across various scenarios (parallel imaging, multi-echo/contrast imaging, low-field MR) demonstrating its utility.

2806.

CAMERA-NET: Cascade Multi-Level Wavelet neural network with data consistency for MRI Reconstruction

Gaojie Zhu^1,2, Xiongjie Shen², and Hua Guo¹
¹Department of Biomedical Engineering, School of Medicine, Tsinghua University, Center for Biomedical Imaging Research, Beijing, China, ²Anke High-tech Co., Ltd, Shenzhen, China

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

Motivation:

The U-net is widely used in deep learning-based MRI reconstruction. Its encoding-decoding component enlarges receptive field while the pooling and interpolation operation limits the ability to recover sparsely sampled MR signals.

Goal(s):

The wavelet transform and inverse wavelet transform are introduced to replace pooling and interpolation operations in order to maintain the spatial information of images during the encoding-decoding process within the neural network.

Approach:

A cascaded multi-level wavelet neural network with data consistency, termed as CAMERA-Net, is presented for under-sampled MRI reconstruction.

Results:

CAMERA-Net demonstrates significant enhancement in reconstructing quality with public fastMRI knee dataset.

Impact: The improved reconstruction capabilities of CAMERA-Net have the potential to enhance precision and reliability when reconstructing under-sampled MRI data. This could result in more efficient clinical scans.

2807.

Model based rEconstruction by Deep Algorithm unrolLing (MEDAL) for fast 3D whole-heart T2 mapping

Alberto Di Biase^1,2, Alina Schneider³, Rene Botnar^1,3,4, and Claudia Prieto^1,2,3
¹MILLENNIUM INSTITUTE FOR INTELLIGENT HEALTHCARE ENGINEERING, Santiago, Chile, ²School of Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile, ³School of Biomedical Engineering, King’s College London, London, United Kingdom, ⁴Institute for Biological and Medical Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile

Keywords: Machine Learning/Artificial Intelligence, Machine Learning/Artificial Intelligence, model-based

Motivation: T2 mapping provides quantitative myocardial tissue characterization. However, current approaches acquire several 2D contrast images which are then fitted to a model to estimate the T2 values, leading to limited coverage, and long acquisition and reconstruction times.

Goal(s): Here we propose to speed up 3D whole-heart T2 mapping using a model-based deep learning unrolling network (MEDAL) that leverages the power of machine learning and physical knowledge.

Approach: MEDAL reconstructs the T2 maps directly without reconstructing any intermediate contrast weighted images or fitting.

Results: The proposed approach was evaluated in iNAV-based free-breathing 3D T2 mapping 4x accelerated showing promising results.

Impact: A novel method for reconstructing parametric maps using a model-based deep learning unrolling network is presented. The method was demonstrated in a highly accelerated free breathing 3D whole-heart T2 mapping sequence allowing for fast and accurate T2 measurements.

2808.

DeepSepSTI: Improved Susceptibility Tensor Reconstruction by Anisotropic Susceptibility Source Separation

Zhenghan Fang¹, Hyeong-Geol Shin^2,3, Blake E. Dewey⁴, Peter A. Calabresi⁴, Peter van Zijl^1,2,3, Jeremias Sulam¹, and Xu Li^2,3
¹Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, United States, ²Department of Radiology and Radiological Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, United States, ³F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States, ⁴Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, MD, United States

Keywords: Machine Learning/Artificial Intelligence, Brain, Susceptibility Tensor Imaging, Susceptibility Source Separation

Motivation: Magnetic susceptibility source separation has potential for characterizing pathological tissue changes in disease. However, existing source separation methods assume isotropic susceptibility, ignoring anisotropy in white matter.

Goal(s): To develop a method for anisotropic susceptibility source separation for better susceptibility tensor reconstruction.

Approach: The paramagnetic susceptibility, modeled by an isotropic scalar, and the diamagnetic susceptibility, modeled by an anisotropic tensor, are jointly estimated in each voxel from local frequency and R2’ measurements using a deep learning model, named DeepSepSTI.

Results: DeepSepSTI shows generally improved estimation of susceptibility tensors, anisotropy and PEV than DeepSTI. DeepSepSTI can better describe tissue characteristics in multiple sclerosis lesions.

Impact: The proposed DeepSepSTI approach may help better measure changes in iron, myelin, and susceptibility anisotropy in various neurological diseases such as multiple sclerosis, potentially providing improved biomarkers for better characterization of disease stage and progression.