Oliver C. Kiersnowski¹, Patrick Fuchs¹, Stephen J. Wastling^2,3, John S. Thornton^2,3, and Karin Shmueli^1
1Department of Medical Physics & Biomedical Engineering, University College London, London, United Kingdom, ²UCL Queen Square Institute of Neurology, University College London, London, United Kingdom, ³Lysholm Department of Neuroradiology, National Hospital for Neurology and Neurosurgery, London, United Kingdom

Simultaneous multi-slice (SMS) acquisition is increasingly used to accelerate echo planar imaging (EPI). EPI acquisitions have been used for quantitative susceptibility mapping (QSM) but, to utilise SMS, an investigation into the effect of SMS on EPI-QSM accuracy is necessary. Here, we show that SMS has no significant effect on magnetic susceptibility maps and values, and can, therefore, provide accurate QSM within a short TR. We also show, for the first time, that multi-echo phase images can be acquired using an EPI sequence (highly) accelerated using SMS and parallel imaging, leading to more accurate QSM reconstruction compared to standard single-echo EPI. 

Beata Bachrata^1,2, Steffen Bollmann^3,4,5, Günther Grabner^1,6,7, Siegfried Trattnig^1,2, and Simon Daniel Robinson^1,3,8
1High Field Magnetic Resonance Centre, Department of Biomedical Imaging and Image-guided Therapy, Medical University Of Vienna, Vienna, Austria, ²Karl Landsteiner Institute for Clinical Molecular MR in Musculoskeletal Imaging, Vienna, Austria, ³Centre of Advanced Imaging, University of Queensland, Brisbane, Australia, ⁴University of Queensland, Centre for Innovation in Biomedical Imaging Technology, Brisbane, Australia, ⁵School of Information Technology and Electrical Engineering, University of Queensland, Brisbane, Australia, ⁶Department of Medical Engineering, Carinthia University of Applied Sciences, Klagenfurt, Austria, ⁷Department of Neurology, Medical University Of Vienna, Vienna, Austria, ⁸Department of Neurology, Medical University of Graz, Graz, Austria

We propose a method to generate a super-resolution QSM from three orthogonal 2D EPI acquisitions with anisotropic voxels. Using distortion correction and non-linear co-registration of the individual EPI images with thick slices, a super-resolution EPI image with isotropic voxels was generated and used to compute a QSM. The super-resolution 2D EPI susceptibility maps, as well as the susceptibility values within deep grey matter structures, showed close correspondence to a standard GRE QSM. The net acquisition time was, however, reduced from several minutes to several seconds, allowing QSM in problematic patient cohorts and clinical routine.

Jingjia Chen¹ and Chunlei Liu^1,2
1Electrical Engineering and Computer Sciences, University of California, Berkeley, Berkeley, CA, United States, ²Helen Wills Neuroscience Institute, University of California, Berkeley, Berkeley, CA, United States

Phase filtering is a very important step in obtaining a good susceptibility tensor. We compare and illustrate the effects of different phase processing methods on the quality of susceptibility tensor reconstruction with different numbers of orientations. In general, tensor reconstructed from variations of the SHARP phase filtering methods show the most robustness against the large field inhomogeneity and produce tensor elements that are spatially and anatomically more consistent. Compared to the original STI reconstruction, asymmetric STI reduces some of the variations induced by phase filtering methods.

Hwihun Jeong¹, Sung Suk Oh², Jongho Lee¹, and Hyeong-Geol Shin^1
1Department of Electrical and Computer Engineering, Seoul National University, Seoul, Korea, Republic of, ²Medical Device Development Center, K-MEDI hub, Daegu, Korea, Republic of

We develop pipelines for reconstructing susceptibility source separation (χ-separation) maps, which requires a T₂ map, from UK Biobank protocol and routine clinical protocol data that have no T₂ map but have various T₂-weighted contrasts (e.g., FLAIR and T₂-weighted images). Using these and additional contrast-weighted images, we propose a deep neural network framework that generates an R₂ (=1/T₂) map, with which χ-separation is conducted. The proposed pipelines successfully generated positive and negative susceptibility maps that are highly similar to gold standard results. The results suggest that χ-separation is applicable to various clinical routine protocols and open-source data.

Nestor Munoz^1,2,3, Carlos Milovic⁴, and Cristian Tejos^1,2,3
1Biomedical Imaging Center, Pontificia Universidad Católica de Chile, Santiago, Chile, ²Department of Electrical Engineering, Pontificia Universidad Católica de Chile, Santiago, Chile, ³Nucleo Milenio Cardio MR, Santiago, Chile, ⁴Department of Medical Physics and Biomedical Engineering, University College, London, United Kingdom

Susceptibility tensor imaging (STI) is a recent MRI technique that is able to quantify anisotropic magnetic susceptibilities in the human tissues. It needs at least 6 acquisitions with different rotation angles to robustly reconstruct each component of a second order tensor. Several methods have been proposed to reconstruct STI, including least squares methods or optimization algorithms. Since there is not a ground truth, it is difficult to evaluate the effectiveness of these algorithms. In this study, we propose a realistic brain human phantom to evaluate and compare different STI reconstruction algorithms.

Jingwen Yao¹, Melanie A. Morrison¹, Angela Jakary¹, Sivakami Avadiappan¹, Yicheng Chen^1,2,3, Julia Glueck⁴, Theresa Driscoll⁴, Michael Geschwind⁴, Alexandra Nelson⁴, Christopher P. Hess^1,4, and Janine M. Lupo^1,2
1Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States, ²Graduate Program in Bioengineering, UCSF/UC Berkeley, San Francisco, CA, United States, ³Facebook Inc., Mountain View, CA, United States, ⁴Department of Neurology, University of California, San Francisco, San Francisco, CA, United States

Quantitative susceptibility mapping (QSM) is a promising tool to investigate iron dysregulation in neurodegenerative diseases. A diverse range of methods has been proposed to generate accurate and robust QSM images. In this study, we evaluated the performance of different dipole inversion algorithms for brain iron imaging at 7T, including iLSQR, iterative methods with regularization (STAR-QSM, FANSI, HD-QSM, MEDI), single-step methods (QSIP, SSTV, SSTGV), and deep learning methods (QSMGAN, QSMnet+). We found that SSTV/SSTGV provided the best performance in terms of correlation with age, correlation with iron, and the differentiation between healthy control and premanifest Huntington’s disease individuals.

Carlos Milovic^1,2, Cristian Tejos^2,3,4, Pablo Irarrazaval^2,3,4,5, and Karin Shmueli^1
1Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ²Department of Electrical Engineering, Pontificia Universidad Catolica de Chile, Santiago, Chile, ³Biomedical Imaging Center, Pontificia Universidad Catolica de Chile, Santiago, Chile, ⁴Millennium Nucleus for Cardiovascular Magnetic Resonance, Santiago, Chile, ⁵Institute for Biological and Medical Engineering, Pontificia Universidad Catolica de Chile, Santiago, Chile

The Structural Similarity Index (SSIM) has become a popular quality metric to evaluate Quantitative Susceptibility Mapping (QSM) in a way that is closer to human perception than the Root-Mean-Squared-Error (RMSE). However, SSIM may over-penalize errors in diamagnetic tissues and under-penalize them in paramagnetic tissues. Extreme susceptibility artifacts may also compress the dynamic-range, resulting in unrealistically high SSIM scores (hacking). To overcome these problems we propose XSIM: SSIM implemented in the native QSM ppm range with new susceptibility-optimized internal parameters. We validated XSIM using data from both QSM challenges. XSIM avoids bias and metric-hacking, promoting sharp susceptibility maps and preventing over-regularization.

Chen Shuang Zhu^1,2, Alexander Mark Weber^1,2,3, Ruth E Grunau^1,3,4, and Natalie Chan^1,3,4
1Institute of Research, BC Children’s Hospital, Vancouver, BC, Canada, ²School of Biomedical Engineering, University of British Columbia, Vancouver, BC, Canada, ³Department of Pediatrics, University of British Columbia, Vancouver, BC, Canada, ⁴Division of Neonatology, BC Women’s Hospital, Vancouver, BC, Canada

Cerebral metabolic rate of oxygen (CMRO2) is a measurement of oxygen metabolism that may help determine outcomes in preterm neonates and improve treatments. However, limitations exist in current methods for measuring CMRO2. Advanced MRI techniques such as quantitative susceptibility mapping (QSM) may be able to overcome these challenges. GE 3T MRI data of preterm neonates at term equivalent age (TEA) (n=9) was obtained and analyzed using a standard equation to determine the CMRO2. CMRO2 values agreed with literature values and correlated strongly with birth age, but not with birth weight or head circumference at birth.

Antonio Ricciardi¹, Anita Karsa², Carmen Tur^1,3, Alberto Calvi¹, Sara Collorone¹, Francesco Grussu^1,4, Marco Battiston¹, Rebecca S Samson¹, Marios Yannakas¹, Jon Stutters¹, Baris Kanber^1,5, Ferran Prados^1,5,6, Karin Shmueli², and Claudia AM Gandini Wheeler-Kingshott^1,7,8
1NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Queen Square Institute of Neurology, University College London, London, United Kingdom, ²Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ³Multiple Sclerosis Centre of Catalonia (Cemcat), Vall d’Hebron Institute of Research, Vall d’Hebron Hospital Campus, Barcelona, Spain, ⁴Radiomics Group, Vall d’Hebron Institute of Oncology, Vall d’Hebron Hospital Campus, Barcelona, Spain, ⁵Centre for Medical Image Computing (CMIC), Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, ⁶Universitat Oberta de Catalunya, Barcelona, Spain, ⁷Department of Brain and Behavioral Sciences, University of Pavia, Pavia, Italy, ⁸Brain Connectivity Centre, IRCCS Mondino Foundation, Pavia, Italy

Quantitative susceptibility mapping (QSM) from multi-echo acquisitions provides more accurate results than single echo, but the optimal choice of echo train length and echo time (TE) range requires investigation. In this work, we showed that QSM values depend on the TE range sampled, so QSM data computed over different TE ranges should not be mixed. Dropping later echoes might help resolve echo-to-echo phase inconsistency artifacts, at the expense of QSM accuracy. Finally, in multi-centre studies using data acquired with different TE settings, covering the same total TE range is preferable to simply matching the number of echoes.

Dimitrios G. Gkotsoulias¹, Roland Müller ¹, Torsten Schlumm¹, Niklas Alsleben¹, Carsten Jäger¹, Jennifer Jaffe^2,3, André Pampel¹, Catherine Crockford^2,3, Roman Wittig^2,3, and Harald Möller^1
1Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, ²Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany, ³Tai Chimpanzee Project, Centre Suisse de Recherches Scientifiques en Cote d'Ivoire, Abidjan, Cote D'ivoire

Calculation Of Susceptibility through Multiple Orientation Sampling (COSMOS) is assessed comparing the optimal, a clinically feasible and multiple-orientation schemes. The optimal COSMOS estimation is used as a gold standard and is compared to the other schemes using the similarity index (SSIM), mean absolute error (MAE) and Pearson’s coefficient (PC). Further comparisons include Thresholded K-space Division (TKD) quantitative susceptibility mapping. For selected white-matter regions, linear regression is used to assess the similarities between the different estimations.

Patrick Fuchs¹ and Karin Shmueli^1
1Medical Physics and Biomedical Engineering, University College London, London, United Kingdom

Direct solutions of the dipole inversion problem in quantitative susceptibility mapping (QSM) are computationally efficient but plagued by streaking artifacts. Here, we have shown that non-uniform sampling of frequency space can achieve additional streaking artifact reduction compared to QSM with thresholded k-space division and state-of-the-art regularisation. By avoiding sampling areas in frequency space where the solution is not well defined, the solution of the ill-posed inverse problem is made more robust and noise amplification is reduced. This approach could be combined with compressed sensing techniques to further improve the QSM reconstruction. This research uses open-source tools from the MR community.