View Presentation Video

Tomoki Amemiya¹, Atsuro Suzuki¹, Yukio Kaneko¹, Suguru Yokosawa¹, and Toru Shirai^1
1Imaging Technology Center, FUJIFILM Corporation, Tokyo, Japan

Keywords: Parallel Imaging, Image Reconstruction

We propose an iterative reconstruction method of parallel imaging using convolutional neural network (CNN)-based denoising in the image domain and data-consistency processing in k-space. The proposed method reduces the noise and artifacts of a reconstructed image compared with the iterative method using sparsity of wavelet transform, suggesting that using CNN-based denoising in iterative reconstruction is effective in reducing noise in parallel imaging.

View Presentation Video

Sajad Mohammed Ali¹, Nirbhay Yadav², Ronnie Wirestam¹, Munendra Singh³, Hye-Young Heo³, Peter van Zijl², and Linda Knutsson^4
1Medical Physics, Lund University, Lund, Sweden, ²Radiology, F.M. Kirby Research Center, Johns Hopkins University, Kennedy Krieger Institute, Baltimore, MD, United States, ³Radiology, Johns Hopkins University, Baltimore, MD, United States, ⁴F.M. Kirby Research Center, Radiology, Medical Radiation Physics, Kennedy Krieger Institute, Johns Hopkins University, Lund University, Baltimore, MD, United States

Keywords: Data Analysis, Machine Learning/Artificial Intelligence, Lorentzian curve fitting

Water saturation shift referencing (WASSR) Z-spectra can be used to correct shifts due to B₀-field inhomogeneities, for magnetic susceptibility mapping and analysis of relaxation effects. The spectra follow a Lorentzian shape with discrete values. Hence, a Lorentzian fit to retrieve the shape parameters (amplitude A, line width LW and frequency shift Δf_H2O ) simplifies analysis. Conventionally, the least-squares (LS) method is used for such fitting despite being time consuming and sensitive to the unavoidable noise in vivo. We propose a deep learning-based Lorentzian-fitting neural network (LoFNet) that demonstrated improved robustness against noise and sampling density in combination with reduced time consumption.

View Presentation Video

Evan McNabb¹, Véronique Fortier^1,2,3,4, and Ives R. Levesque^3,4
1Medical Imaging, McGill University Health Centre, Montreal, QC, Canada, ²Diagnostic Radiology, McGill University, Montreal, QC, Canada, ³Gerald Bronfman Department of Oncology, McGill University, Montreal, QC, Canada, ⁴Medical Physics Unit, McGill University, Montreal, QC, Canada

Keywords: Data Acquisition, Data Analysis

Noise estimates in Deep Learning-based image reconstruction (DLR) from background regions lead to artificially low noise estimates and thus inaccurate SNR. Noise estimate accuracy was improved using the difference between two identical acquisitions. SNR increased in DLR compared to standard reconstruction in clinical sequences over a range of acquired signal averages and voxel sizes. With DLR, varying signal averages had little effect on the measured SNR in fast spin echo acquisitions, while SNR increased by more than three-fold in single-shot fast spin echo. However, with DLR, SNR no longer follows predicable behavior, therefore sequence optimizations need to be performed experimentally. 

Junhao Zhang^1,2, Zheyuan Yi^1,2,3, Yujiao Zhao^1,2, Linfang Xiao^1,2, Jiahao Hu^1,2,3, Vick Lau^1,2, Fei Chen³, Alex T.L.Leong^1,2, and Ed X. Wu^1,2
1Laboratory of Biomedical Imaging and Signal Processing, the University of Hong Kong, HongKong, China, ²Department of Electrical and Electronic Engineering, the University of Hong Kong, HongKong, China, ³Department of Electrical and Electronic Engineering, Southern University of Science and Technology, Shenzhen, China

Keywords: Parallel Imaging, Data Acquisition, Brain reconstruction, Cardiac reconstruction

We present a U-Net based deep learning model to estimate the multi-channel ESPIRiT maps directly from uniformly-undersampled multi-channel multi-slice MR data. The model is trained with a hybrid loss function using fully-sampled multi-slice axial brain datasets from the same MR receiving coil system. The proposed model robustly predicted ESPIRiT maps from uniformly-undersampled k-space brain and cardiac MR data, yielding highly comparable performance to reconstruction using to acquired reference ESPIRiT maps. Our proposed method presents a general strategy for calibrationless parallel imaging reconstruction through learning from coil and protocol specific data.

View Presentation Video

Xia Li¹, Hui Zhang², and Tie-Qiang Li^3
1Information Engineering, China Jiliang University, Hangzhou, China, ²Information Engineering, China Jiliang University, HANGZHOU, China, ³Karolinska Institute, Stockholm, Sweden

Keywords: Image Reconstruction, Brain, Compressed sensing MRI

Generative adversarial network (GAN) has emerged as one of the most prominent approaches for fast CS-MRI reconstruction. However, most deep-learning models achieve performance by increasing the depth and width of the networks, leading to prolonged reconstruction time and difficulty to train. We have developed an improved GAN-based model to achieve quality performance without increasing complexity by implementing the following: 1) dilated-residual structure with different dilation rates at different depth of the networks; 2) CAM to adjust the allocation of network resources; 3) multi-scale information fusion module to achieve feature fusion. Experiment data have confirmed the validity for the modules.

View Presentation Video

Marcelo V. W. Zibetti¹ and Ravinder R. Regatte^1
1Radiology, NYU Grossman School of Medicine, New York, NY, United States

Keywords: Data Acquisition, Image Reconstruction

This work proposes a stochastic variation of the bias-accelerated subset selection (BASS) algorithm to learn an efficient sampling pattern (SP) for accelerated MRI. This algorithm is used in the joint learning of an SP and neural network reconstruction. We apply the proposed approach to two different 3D Cartesian parallel MRI problems. The proposed stochastic approach, when used for joint learning, improves the learning speed from 2.5X to 5X, obtaining SPs with similar properties as the non-stochastic approach with nearly the same RMSE and SSIM.

View Presentation Video

Jinho Kim^1,2, Thomas Benkert², Bruno Riemenschneider¹, Marcel Dominik Nickel², and Florian Knoll^1
1Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ²MR Application Pre-development, Siemens Healthcare GmbH, Erlangen, Germany

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction, Magnetic Resonance Cholangiopancreatography

MR cholangiopancreatography (MRCP) is a special MRI technique to visualize the biliary systems. Deep Learning-based (DL) reconstruction models have shown to reduce scan time from many anatomical regions. However, they generally require large training datasets. This is challenging for applications like MRCP, where public datasets are not available. This work analyzes two approaches to training a DL model for highly accelerated MRCP reconstruction: training from scratch using a small MRCP dataset and fine-tuning a model pretrained on a public knee dataset. Results show that despite of the substantial data domain shift between training and testing, fine-tuning outperformed training from scratch. 

View Presentation Video

Satoshi ITO¹ and Keitaro TAKAHASHI^1
1Graduate Program in Information, Electrical and Electronic Systems Engineering, Utsunomiya University, Utsunomiya, Japan

Keywords: Data Processing, Data Processing

In noise reduction, the lower the amount of noise, the lower the image degradation associated with the noise reduction. Using this property, we propose a new denoising scheme that can improve denoising performance using trained CNN. Noise suppressed image by low-pass filter is inputted to denoising CNN and then output image is enhanced by high-pass filter. Experimental results show that noise suppression of the input image improves both image structure preservation and noise processing performance. The proposed method was applied to a parallel blind image denoising method. As a results, further improvement in performance was shown.

View Presentation Video

Natthanan Ruengchaijatuporn^1,2, Siddharth Srinivasan Iyer^3,4,5, Sophie Schauman^3,4, Quan Chen^3,4, Xiaozhi Cao^3,4, Itthi Chatnuntawech⁶, and Kawin Setsompop^3,4
1Center of Excellence in Computational Molecular Biology, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand, ²Center for Artificial Intelligence in Medicine, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand, ³Department of Radiology, Stanford University, Stanford, CA, United States, ⁴Department of Electrical Engineering, Stanford University, Stanford, CA, United States, ⁵Department of Electrical Engineering and Computer Science, MIT, Cambridge, MA, United States, ⁶National Nanotechnology Center, National Science and Technology Development Agency, Pathum Thani, Thailand

Keywords: Sparse & Low-Rank Models, Machine Learning/Artificial Intelligence

Recent advances in spatio-temporal subspace reconstruction has enabled accurate reconstruction from highly accelerated scans. Nevertheless, such methods suffer from being computationally intensive due to their iterative nature coupled with the large dimensionality of the problem, especially when imperfection correction is incorporated into the formulation.  This abstract proposes deep-learning-initialized compressed sensing (Deli-CS) to accelerate such spatio-temporal reconstruction by providing it with a deep-learning-reconstructed initial solution, reducing the number of iterations required. Using MRF as an example, Deli-CS reconstructs data from a rapid 1-mm isotropic whole-brain TGAS-SPI-MRF with time-segmented B₀ correction at 3x faster speed compared to FISTA.

View Presentation Video

Muhammad Ahmad Sultan¹, Chong Chen¹, Yuchi Han¹, and Rizwan Ahmad^1
1The Ohio State University, Columbus, OH, United States

Keywords: Image Reconstruction, Data Processing, Image Reconstruction, Unsupervised Learning, Deep Image Prior

We propose an unsupervised learning method for dynamic MRI reconstruction. Our method, Deep Image prior with StruCtUred Sparsity (DISCUS), is an extension of Deep Image Prior (DIP) and employs joint optimization of network parameters and latent code vectors to recover image series. We enforce group sparsity on code vectors to reveal the underlying low-dimensional manifold of the image series. Using two sets of in vivo measurements and digital phantom simulation, we show that our approach significantly improves the reconstruction quality in terms of Normalized Mean Square Error (NMSE) and Structure Similarity Index Measure (SSIM) compared to compressed sensing and DIP.

View Presentation Video

Muhammad Shafique^1,2, Sohaib Ayyaz Qazi³, Faisal Najeeb¹, and Hammad Omer^1
1Electrical and Computer Engineering, COMSATS University Islamabad, Islamabad, Pakistan, ²Electrical Engineering, University of Poonch Rawalakot, Rawalakot AJK, Pakistan, ³Center for Medical Image Science and Visualization, Linköping University, Linköping, Sweden

Keywords: Image Reconstruction, Artifacts, Parallel MRI, Deep Learning, Attention Mechanism

Deep learning has made significant progress in recent years, however the local receptive field in CNN raises questions about signal synthesis and artefact compensation. This paper proposes a deep learning Spatial-Channel Attention U-Net (SCA-U-Net) to solve the folding problems arising in MR images because of undersampling. The main idea is to enhance local features in the images and restrain the irrelevant features at the spatial and channel levels. The output from SCA-U-Net is further refined by adding a small number of originally acquired low-frequency k-space data. Experimental results show a better performance of the SCA-U-Net model than classical U-Net model.

View Presentation Video

Thomas Benkert¹, Elisabeth Weiland¹, Simon Arberet², Majd Helo¹, Fasil Gadjimuradov^1,3, Karl Engelhard⁴, Gregor Thoermer¹, and Dominik Nickel^1
1MR Application Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany, ²Digital Technology & Innovation, Siemens Medical Solutions USA, Princeton, NJ, United States, ³Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, ⁴Institute of Radiology, Martha-Maria Hospital, Nuremberg, Germany

Keywords: Diffusion/other diffusion imaging techniques, Translational Studies

Diffusion weighted imaging (DWI) has found widespread use in daily clinical routine but can still be limited by long acquisition times and low spatial resolution. In this work, combining deep learning-based k-space to image reconstruction with super resolution processing tailored to support partial Fourier acquisitions is demonstrated to efficiently mitigate these obstacles. The approach is shown for various applications, including liver, breast, prostate, and brain DWI at 0.55T, 1.5T, and 3T.

Qin Zhao¹, Song Zhang¹, Liyun Zheng², and Yongming Dai^3
1Department of Radiology, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and Therapy, Guangzhou, China, ²Shenzhen United Imaging Research Institute of Innovative Medical Equipment, Shenzhen, China, ³MR Collaboration, Central Research Institute, United Imaging Healthcare, Shanghai, China

Keywords: Data Acquisition, Machine Learning/Artificial Intelligence

To improve the scanning efficiency of magnetic resonance imaging (MRI) for nasopharyngeal carcinoma, this study investigated the value of accelerating technique, artificial intelligence-assisted compressed sensing (ACS), in comparison to conventional sequences without accelerating technique and accelerating MRI using parallel imaging (PI). Eleven patients diagnosed with nasopharyngeal carcinoma were prospectively enrolled. As a result, ACS achieved the shortest acquisition time, with similar or even better image quality and SNR than conventional sequences. ACS has the potential to provide sufficient image quality for T1- and T2-weighted imaging in nasopharyngeal carcinoma and could be an alternative to conventional sequences in clinical practice.

View Presentation Video

Siddhartha Satpathi¹, Jacinta E. Browne¹, AbdulRahman Alfayad¹, Jared T. Verdoorn¹, and James G. Pipe^1
1Mayo Clinic, Rochester, MN, United States

Keywords: Artifacts, Spinal Cord

Upon investigating a dataset of 804 studies for spinal fractures from two major vendors, the authors observed that 11% of the water-fat images in the studies are mis-labelled. This motivated the development of an automated algorithm to correct the mis-labelling. We used a 2D CNN based deep learning model to classify the images correctly with the aim of reducing error and fatigue in clinical diagnosis caused by such mis-labeling, as well as providing correct labels for further AI workflow. We also demonstrated the use of active learning in this problem by achieving the same test-error with fewer labels than using the entire training data.

View Presentation Video

Ganeshkumar M¹, Devasenathipathy Kandasamy², Raju Sharma², and Amit Mehndiratta^1,3
1Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi, India, ²Department of Radio Diagnosis, All India Institute of Medical Sciences, New Delhi, India, ³Department of Biomedical Engineering, All India Institute of Medical Sciences, New Delhi, India

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction, Deep Learning, Fat-Water seperation

In this study, we propose a novel lightweight encoder-decoder architecture for deep learning based Fat-Water separation in multi-echo MRI data. The architecture's performance is evaluated in the biophysical model-guided deep learning-based Fat-Water separation task and compared against the widely used U-Net. This biophysical model-guided deep learning-based Fat-Water separation requires no training data and ground truths, but it involves time-consuming loss minimization for thousands of epochs. Despite having significantly fewer training parameters, the proposed architecture performed equally well in generating the Fat-Water maps compared to the U-Net. So, our proposed architecture aids in the faster generation of Fat-Water maps.

Jing Cheng¹, Zhuo-Xu Cui¹, Qingyong Zhu¹, and Dong Liang^1
1Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

Keywords: Machine Learning/Artificial Intelligence, Image Reconstruction

Existing deep learning-based methods for MR reconstruction mainly use MSE as loss function to train the network under the assumption that MR images follow the sub-Gaussian distribution, without considering the real distribution of the images. In this work, we propose a new DL-based method that models the image distribution with equilibrium Langevin dynamic to converge the distribution, and trains the network with Wasserstein distance to approach the real distribution. Experimental results on highly undersampled MR data demonstrate the superior performance of the proposed method.

View Presentation Video

Yuxiang Zhou¹, Riti Paul², Pak Lun Kevin Ding², Leland Hu¹, Ameet C Patel¹, and Baoxin Li^2
1Radiology, Mayo Clinic at Arizona, Phoenix, AZ, United States, ²CIDSE, Arizona State University, Tempe, AZ, United States

Keywords: Machine Learning/Artificial Intelligence, Data Processing, Reconstruction

Artificial intelligence (AI) applications in the field of magnetic resonance imaging have been implemented in routine clinical practice. However, MRI still faces a practical and persistent challenge: its long acquisition time. This has led to two prominent issues in health care: high cost and poor patient experience. Long acquisition time is also a source of degraded imaging quality (e.g., motion artifacts).  In the study, we propose to develop novel Deep Learning (DL) architectures combined with transfer learning capabilities to address the above challenge and apply this newly developed AI technique for image reconstruction in MRI with very fast imaging acquisition.

View Presentation Video

Mohammad Zalbagi Darestani¹, Vishwesh Nath², Wenqi Li², Yufan He², Holger Reinhard Roth², Ziyue Xu², Daguang Xu², Reinhard Heckel^1,3, and Can Zhao^2
1Rice University, Houston, TX, United States, ²NVIDIA, Santa Clara, CA, United States, ³Technical University of Munich, Munich, Germany

Keywords: Software Tools, Machine Learning/Artificial Intelligence, MRI Reconstruction

Deep learning models outperform traditional methods in terms of quality and speed for numerous medical imaging applications. A critical application is the acceleration of magnetic resonance imaging (MRI) reconstruction, where a deep learning model reconstructs a high-quality MR image from a set of undersampled measurements. For this application, we present the MONAI Recon Module to facilitate fast prototyping of deep-learning-based models for MRI reconstruction. Our free and open-source software is pre-equipped with a baseline and a state-of-the-art deep-learning-based reconstruction model and contains the necessary tools to develop new models. The developed open-source software covers the entire MRI reconstruction pipeline.

View Presentation Video

Tiebao Meng¹, Huiming Liu¹, Chuanmiao Xie¹, Jialu Zhang², and Long Qian^2
1Department of Radiology, Sun Yat-Sen University Cancer Center, Guangzhou, China, ²MR Research, GE Healthcare, Beijing, China

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

As the most common malignant tumor in women, the differentiation of benign lesion from malignant breast tumors is the most essential step in early diagnosis. Recent study demonstrated that DWI could offer great help in differential diagnosis of breast tumors. The novel deep learning-based reconstruction (DLR) technique is able to increase SNR of MRI images. Using DLR, DWI images can be acquired with less NEX (fast DWI) and still maintain the image quality. This study indicated the feasibility of fast DWI protocol with DLR in the differentiation of breast benign lesions and malignant tumors.

View Presentation Video

Arjun Narula¹, Uday Patil², Anand SH³, Harikrishnan Raveendran⁴, Shailaja Muniraj⁵, Pallavi Rao⁶, Anurita Menon⁶, Allison Nicole Garza⁷, Santosh Kumar⁷, Gautam Kumar⁷, Srinivas NR⁷, Syam Babu⁷, Ravi Jaiswal⁷, Rajagopalan Sundaresan⁷, Ashok Kumar Reddy⁷, Sajith Rajamani⁷, Rajdeep Das⁷, Nitin Jain⁷, Sudhir Ramanna⁷, Sundar V⁷, Florintina Charlaas⁷, Sudhanya Chatterjee⁷, Rohan Patil⁷, Megha Goel⁷, Dattesh Shanbhag⁷, Vikas Kumar Anand⁷, Abhishek Galagali⁷, Sathish KV⁷, Preetham Shankpal⁷, Harsh Kumar Agarwal⁷, Suresh Emmanuel Joel⁷, and Ramesh Venkatesan^7
1Narula Diagnostics, Rohtak, India, ²Manipal Hospitals, Bangalore, India, ³Jivaa Diagnostics, Tumkur, India, ⁴Core Clinico PET Imaging, Thane, India, ⁵Prima Diagnostics, Bangalore, India, ⁶Image Core Lab, Bangalore, India, ⁷GE Healthcare, Bangalore, India

Keywords: Machine Learning/Artificial Intelligence, Low-Field MRI, Accessible MRI

We validate the clinical viability of a 0.5T scanner to reduce cost and improve access to quality MRI using AI based IQ enhancement to compensate for IQ reduction due to lower field and other lower hardware specifications. We obtained data from 65 patients from the re-ramped 0.5T and a commercially available 1.5T MRI system for brain and cervical spine. Radiologists compared image quality between the two and rated the image on the ability to perform diagnosis. We observed that more than 90% of the images were rated to be above diagnostic levels and AI reconstruction significantly improved the image quality.