Relaxometry
 

Tuesday 2 June 2015

Noam Ben-Eliezer^1,2, Daniel K Sodickson^1,2, Timothy M Shepherd^1,2, Graham C Wiggins^1,2, and Kai Tobias Block^1,2
1Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States, ²Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University School of Medicine, New York, NY, United States

Accurate mapping of T₂ relaxation values in vivo is highly challenging in clinical settings. This work integrates two techniques: (1) a recently-developed T₂ mapping technique – the echo-modulation curve (EMC) algorithm – which is immune to the T₂ bias caused by indirect coherence pathways in multi spin-echo protocols, and (2) non-linear model-based reconstruction of radially sampled datasets allowing to overcome undersampling-related aliasing artifacts. The synergistic combination of the two provides accelerated scanner- and parameter-invariant quantification of T₂ relaxation, and morphological proton-density information at sub-millimeter spatial resolutions and reduced motion sensitivity.

Bruno Madore¹, W. Scott Hoge¹, Tai-Hsin Kuo², and Cheng-Chieh Cheng^1
1Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ²Philips Healthcare, Taipei, Taiwan

The present method aims to obtain in little more than a single scan all of the main physical MR parameters: T1, T2, T2*, M0, B0 and B1. In principle at least, based on these maps one could generate images of essentially any desired contrast, through well-known signal equations. The approach treats the flip angle as a parameter to be evaluated rather than a known quantity, and all parameters are calculated one at a time and/or through linear equations, as opposed to numerically solving larger and non-linear systems of equations. The method was validated on phantom data, with in vivo results forthcoming.

Kathryn E Keenan¹, Karl A Stupic¹, Elizabeth Horneber², Michael Boss¹, and Stephen E Russek^1
1National Institute of Standards and Technology, Boulder, CO, United States, ²University of Colorado, Boulder, CO, United States

Method to independently tune T1 and T2 relaxation times using the paramagnetic ions Ni++ and Mn++ in aqueous solution. These phantoms, when sealed and stored in the MR laboratory exhibit long term stability with 10% change.

Alexander Ruh¹, Philipp Emerich¹, Harald Scherer², Dmitry S. Novikov³, and Valerij G. Kiselev^1
1Dept. of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, ²Dept. of Inorganic and Analytical Chemistry, University Freiburg, Freiburg, Germany,³Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States

Transverse relaxation in biological tissues is sensitive to the structural organization of magnetic inhomogeneities on the cellular level. A recently developed theory predicts a reflection of this structural organization in the long-time behavior of the induced relaxation rate as a power law approach to the asymptote. Here we present the first direct experimental verification of such dependence in differently composed media. Specifically, we observe the theoretically expected behavior in a microbead phantom with two different types of structural disorder.

Christoph Birkl¹, Christian Langkammer², Nicole Golob-Schwarzl³, Marlene Leoni³, Johannes Haybaeck³, Walter Goessler⁴, Franz Fazekas¹, and Stefan Ropele^1
1Department of Neurology, Medical University of Graz, Graz, Austria, ²MGH/HST Martinos Center for Biomedical Imaging, Harvard Medical School, Boston, MA, United States,³Department of Neuropathology, Institute of Pathology, Medical University of Graz, Austria, ⁴Institute of Chemistry, Analytical Chemistry, University of Graz, Austria

Post mortem MRI is commonly used to validate quantitative MRI methods and to improve our understanding of relaxation in tissue. However, formalin which is frequently used to preserve tissue samples for degradation, can substantially affect MR relaxation times, which hampers comparison with in vivo conditions. In this study, we systematically evaluated the influence of formalin fixation on MR relaxation times by combining water content measurements, electrophoresis, and temperature dependent relaxation time measurements. Our results suggest that formalin induced cross-linking is the main reason for the shortening of relaxation times while changes of water content do not play a major role.

Kyoko Fujimoto^1,2, Trent V. Robertson¹, Vanessa Douet², David G. Garmire¹, and V. Andrew Stenger^1,2
1Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, HI, United States, ²Department of Medicine, John A. Burns School of Medicine, University of Hawaii, Honolulu, HI, United States

Techniques for better Magnetic Resonance imaging are being continuously developed. Before applying new techniques on subjects and patients in a scanner, they are tested with cylindrical or spherical phantoms. However, results are often not realistic since the human cerebrum has complex structure with multi-contrast tissues and gyrifications. Some phantoms model electrical and functional properties but a phantom with gray and white matter structure does not exit. The purpose of this study is to show the development of an anthropomorphic phantom to obtain calibration data with simulated cerebral tissues to reduce cost and time by not necessitating in-vivo subjects.

Xiaoqing Wang¹, Volkert Roeloffs¹, Klaus-Dietmar Merboldt¹, Dirk Voit¹, Sebastian Schaetz¹, and Jens Frahm^1
1Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fuer biophysikalische Chemie, Göttingen, Germany

Low spatial resolution and long acquisition time have been the two major limitations for T1 mapping’s routine application in clinical MRI. Recently several image space and k-space based reconstruction methods have been proposed to estimate T1 maps from undersampled data, however, most of them dealing with single-shot single-slice T1 mapping. In this work, a single-shot sequential multi-slice acquisition scheme together with a new relaxation model is proposed. After combining it with radial undersampling and iterative reconstruction, our method facilitates accurate, high-resolution multi-slice T1 mapping.

Andreas Petrovic¹ and Rudolf Stollberger^2
1Institute of Medical Engineering, Universtiy of Technology Graz, Graz, -, Austria, ²Institute of Medical Engineering, University of Technology Graz, -, Austria

In multi-parametric MRI several MR parameters are estimated within one single scan, which reduces acquisition time. In this work we modified a multi-echo spin-echo sequence for simultaneous estimation of M₀, T₁, and T₂ using the Generating Functions formalism. Reducing the refocusing angle allows longitudinal magnetization recovery during the sequence, which is used to create a final “T₁ weighted echo”. Simulations, measurements and fitted T₂ values are in excellent agreement; however, accuracy of fitted T₁ values is limited, requiring a thorough sensitivity assessment. The robust spin echo nature and reduced SAR of this sequence render it applicable at high field scanners.

Gopal Nataraj¹, Jon-Fredrik Nielsen^2,3, and Jeffrey A. Fessler^1,2
1Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, United States, ²Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States,³Functional MRI Laboratory, University of Michigan, Ann Arbor, MI, United States

We describe a CRLB-inspired min-max optimization problem to guide scan design for relaxometry. In essence, our method minimizes the theoretical worst-case (i.e., maximum) standard deviations of T₁ and T₂ estimates. As an example, we first optimize two DESS acquisitions for T₂ relaxometry in the brain. Our results show that predicted and empirical T₂standard deviations over WM/GM ROIs recommend similar scan parameter combinations for precise T₂ estimation. We then compare a regularized T₂ estimate from our suggested scan protocol against one from many acquisitions and find that much T₂ content in DESS is well captured with only two scans.

Sofia Chavez^1,2
1Centre for Addiction and Mental Health, Toronto, Ontario, Canada, ²Psychiatry, University of Toronto, Toronto, Ontario, Canada

The purpose of this work is to present a simplified method (eMoS) that can be used to compute T1 maps from two low flip angle SPGR signal acquisitions if B1 maps are known. A study of the effects of B1 errors and noise on eMoS in contrast to the effect of these on the standard variable flip angle (VFA) method can be used to explain the observed differences in human brain, in vivo, T1 maps. Simulations and human scans are used to demonstrate the improved accuracy and noise robustness of the eMoS relative to the VFA method for T1 mapping.

Eamon K Doyle^1,2, Jonathan M Chia³, Krishna Nayak^1,4, and John C Wood^1,2
1Biomedical Engineering, University of Southern California, Los Angeles, CA, United States, ²Cardiology, Children's Hospital of Los Angeles, Los Angeles, CA, United States, ³Philips Healthcare, Cleveland, OH, United States, ⁴Electrical Engineering, University of Southern California, Los Angeles, CA, United States

Evaluation of iron-containing tissues at 3T presents challenges related to ultra-fast, susceptibility-mediated signal decay. We propose a low flip angle spin echo sequence as a method of acquiring data to diagnose liver iron overload when traditional pulse sequences are too slow.

Sampada Bhave¹, Sajan Goud Lingala², Casey P Johnson¹, Vincent A Magnotta¹, and Mathews Jacob^1
1University of Iowa, Iowa City, Iowa, United States, ²Electrical Engineering, University of Southern California, Los Angeles, California, United States

Quantification of multiple tissue parameters is emerging as a powerful tool in diagnosing various neurological and psychiatric diseases. However the major bottleneck in its routine clinical use is the long acquisition time. In addition, long acquisition times are likely to result in motion induced artifacts. In this work, we propose a dictionary learning based scheme to simultaneously recover T1ρ and T2 maps. The proposed method models the data as a weighted linear combination of basis functions from a dictionary, which is learned from the measured data.

Andreas Lesch¹, Andreas Petrovic¹, Tilman Johannes Sumpf², Christoph Stefan Aigner¹, and Rudolf Stollberger^1
1Department for Medical Engineering, Graz University of Technology, Graz, Styria, Austria, ²Biomedizinische NMR Forschungs GmbH, Max-Planck-Institut für biophysikalische Chemie, Göttingen, Germany

This work describes multiple parameter quantification using the IR-bSSFP sequence with single coil measurements and parallel imaging. An advanced implementation is shown that address quantification accuracy by slice profile correction, high acceleration by undersampling and integrated parameter determination by model based image reconstruction.

Issac Y Yang¹, Kai-Ho Fok¹, Bernd J Wintersperger^2,3, and Marshall S Sussman^2,3
1Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada, ²Department of Medical Imaging, University of Toronto, Toronto, Ontario, Canada, ³Joint Department of Medican Imaging, University Health Network & Mt. Sinai Hospital, Toronto, Ontario, Canada

Modified Look-Locker Inversion Recovery (MOLLI) is a common technique for T1 mapping of the heart. However, it requires a lengthy rest period in between inversion groupings to allow for complete magnetization recovery. A new MOLLI fitting algorithm with a new fitting model is presented in this work to allow for arbitrary inversion groupings and rest periods, including no rest period. Phantom and in-vivo experiments verified the technique. T1 values measured using several different inversion grouping/rest period combinations yielded consistent results with the new algorithm, but with a slight loss of precision.

Shivaprasad Ashok Chikop¹, Antharikshanagar Bellappa Sachin Anchan¹, Shaikh Imam¹, Amaresha Shridhar Konar¹, Rashmi Rao¹, Arush Honnedevasthana Arun¹, and Sairam Geethanath^1
1Medical Imaging Research Center, Dayananda Sagar Institutions, bangalore, Karnataka, India

Plug-n-Play (PnP) MRF is a novel technique based on MRF and provides an opportunity to MR researchers who do not have access to pulse sequence designing to use a comprehensive framework like MRF on clinical scanner using readily available pulse sequence. PnP MRF is implemented using PSIF sequence and is modelled using a data driven model which uses an analytical equation obtaining optimum Contrast to Noise Ratio (CNR) between brain tissues, enabling fast computation of dictionary. The number of MR images required for MRF has been reduced to forty eight in PnP MRF.

Gwendolyn Van Steenkiste¹, Dirk H.J. Poot^2,3, Ben Jeurissen¹, Arnold J. den Dekker^1,4, and Jan Sijbers^1
1iMinds-Vision Lab, University of Antwerp, Antwerp (Wilrijk), Antwerp, Belgium, ²BIGR (Medical informatics and Radiology), Erasmus Medical Center Rotterdam, Rotterdam, Netherlands, ³Imaging Science and Technology, Delft University of Technology, Delft, Netherlands, ⁴Delft Center for Systems and Control, Delft University of Technology, Delft, Netherlands

The spin-lattice relaxation time (T1) of tissues can be estimated from a set of T1 weighted images. Even though T1 mapping has a broad range of potential applications, T1 maps are not routinely used in clinical practice. The high acquisition time of the necessary set of high resolution T1 weighted images is not clinically feasible. We propose to use super-resolution in combination with T1 estimation to provide a better trade off between the acquisition time, SNR and spatial resolution of the T1 maps.

Tobias Charles Wood¹, Stephen J Wastling¹, and Gareth J Barker^1
1Neuroimaging, King's College London, London, London, United Kingdom

We combine the Geometric Solution to bSSFP band-removal with DESPOT2 to speed up calculation of T2 by 4 orders of magnitude, at the expense of increased scan time.

Vasileios Zampetoulas¹, Lionel M. Broche¹, and David J. Lurie^1
1Aberdeen Biomedical Imaging Centre,School of Medicine&Dentistry, University of Aberdeen,Foresterhill, AB25 2ZD, Aberdeen, United Kingdom

Fast Field-Cycling (FFC) NMR is a technique that provides information about the molecular dynamics of a range of materials by using the T1-dispersion curve. After the compensation of the stray magnetic fields coming from external sources, the T1-dispersion curve can extend to magnetic fields in the region of μT allowing for the investigation of slow dynamic processes and determination of the local magnetic fields within the sample. With further development, differences in that region of the curve obtained from biological samples can be associated with the early stages of progress of diseases, leading to a new diagnostic tool.

Jennifer G. Whisenant¹, Lori R. Arlinghaus¹, Richard D. Dortch¹, William A. Grissom¹, Gregory S. Karczmar², and Thomas E. Yankeelov^1
1Vanderbilt University, Nashville, TN, United States, ²University of Chicago, Chicago, IL, United States

This study evaluated the effect of B₁⁺ correction on T₁ measurements calculated from variable flip angle (VFA) data in phantoms and in vivo. Additionally, the accuracy of these measurements was compared to T₁ values from inversion recovery (IR) data. On average, 31% error was observed between IR and uncorrected VFA data in both phantoms and in vivo, which was significantly reduced to 6% (P<0.001) when applying a B₁⁺ correction. IR measurements of T₁ are very robust, yet the long acquisition times limit clinical utility. Therefore, these data suggest that the VFA method with B₁⁺ correction is a suitable alternative.

Kelly C McPhee¹ and Alan H Wilman^2
1Physics, University of Alberta, Edmonton, Alberta, Canada, ²Biomedical Engineering, University of Alberta, Edmonton, Alberta, Canada

It is common for researchers to skip the first echo, or skip all odd echoes in the echo train in order to determine T2 from exponential fitting multi-echo spin echo data. Using both simulations and in vivo experiments of slice selective muti-echo spin echo experiments, we find that these alternate exponential fitting schemes do not avoid T2-misestimation. Skipping the first echo may reduce errors to less than 10%, if very wide refocusing pulses are used, provided that the refocusing angles are >150 degrees.

Aviv Mezer¹, Ariel Rokem², Trevor Hastie², and Brian Wandell^2
1Edmond and Lily Safra Center for Brain Sciences, The Hebrew University, Jerusalem, Israel, ²Stanford university, CA, United States

Proton density is the most basic MRI measure, representing the amount of water protons in each voxel. Water content differs between tissue types, and changes during development and in disease. Using simulation we find that in the presence of noise, combining parallel imaging, smoothness assumptions, and the biophysical regularization together generates the most accurate estimates of both coil sensitivity maps and PD. We confirm the high accuracy when using multiple coils and T1-regularization on data from a phantom and a living human brain.

Mustapha Bouhrara¹ and Richard G. Spencer^1
1National Institute on Aging, NIH, BALTIMORE, Maryland, United States

Stochastic region contraction (SRC) has been proposed as an efficient approach for extracting system parameters from mcDESPOT data. However, the SRC algorithm exhibits a high degree of sensitivity to initial parameter conditions, especially at low-to-moderate signal-to-noise ratios. In this study, we investigated the accuracy and precision of component fraction determination in a bicomponent mcDESPOT model using two Bayesian methods, based respectively on maximum posterior probability and means, and compared the results with those derived using the SRC algorithm. Results show that the estimation of component fractions was markedly improved through use of Bayesian analysis.

Dushyant Kumar^1,2, Susanne Siemonsen^1,3, Jens Fiehler¹, and Jan Sedlacik^1
1Neuroradiology, Universitätsklinikum Hamburg-Eppendorf, Hamburg, Hamburg, Germany, ²Multiple Sclerosis Imaging Section (SeMSI), Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany, Germany, ³Multiple Sclerosis Imaging Section (SeMSI), Universitätsklinikum Hamburg-Eppendorf, Hamburg, Hamburg, Germany

Problem: The voxelwise T2-decay deviate significantly from multi-exponential (ME) model due to the stimulated emission (SE) resulting from the imperfect slice profile of refocusing pulses and the B1-error and so, following ME model would lead to serious quantification inaccuracy. Methods: A novel post-processing method, combining extended phase graph model and multi-exponential model with nonnegative least-square, is being proposed to compensate for B1 error. The method requires using only QT2R data. Results & Conclusions: Both experimental and simulated results are presented to demonstrate the improvement. A significant improvement in myelin quantification is demonstrated when flip angle error is significant (≥8-10%).

Amaresha Shridhar Konar¹, Rashmi R Rao¹, Shaik Imam¹, Shivaprasad Chikop¹, Sachin Anchan¹, and Sairam Geethanath^1
1Medical Imaging Research Center, Dayananda Sagar College of Engineering, Bangalore, Karnataka, India

Magnetic Resonance Fingerprinting (MRF) provides simultaneous MR multi-parametric maps. Purpose of this work is to reduce the number of TR/FA and to obtain optimal combinations. PSIF sequence was used to obtain 3D plots for different brain tissue types and a difference plot of GM and WM were generated. 48 TR/FA (through sample size analysis) values were picked from difference plot to achieve high CNR and T1 and T2 maps were obtained by pattern matching algorithm. Mean and Standard deviation of T1 and T2 values for WM, GM, and CSF were calculated that lie in the standard range.

Kelly C McPhee¹ and Alan H Wilman^2
1Physics, University of Alberta, Edmonton, Alberta, Canada, ²Biomedical Engineering, University of Alberta, Edmonton, Alberta, Canada

Recent works have implemented methods to fit for single component T2 from slice selective multi-echo spin echo experiments with indirect and stimulated echo compensation. These works have made use of either the Extended Phase Graph with Fourier Transform slice approximations, or Shinnar-Le Roux and Bloch simulations to model complete echo pathways. Here, we compare the recently demonstrated SLR based fitting technique with prior flip angle measurement to the EPG based method which is a commonly used correction technique for multi-echo spin echo based T2 measurements. We use simulations and human in vivo experiments to compare these two methods of indirect and stimulate echo compensated T2 fitting.

Yang Liu¹, John R. Buck¹, Shaokuan Zheng², and Vasiliki N. Ikonomidou^3
1Electrical and Computer Engineering, University of Massachusetts Dartmouth, North Dartmouth, MA, United States, ²Department of Radiology, University of Massachusetts Medical School, Worcester, MA, United States, ³Bioengineering, George Mason University, Fairfax, VA, United States

Quantitative MRI estimates MR quantities such as T1 and T2 as precisely and accurately as possible within a relatively short scan time. Although there is extensive relaxometry research for T1/T2 mapping, little effort has been made to statistically quantify T1/T2 estimate efficiencies. This paper establishes a framework to compare different relaxometry sequences on their T1/T2 estimate precisions per unit time. The new metric T1/T2-to-noise ratio (TNR) efficiency defines from the Cramer-Rao bound, a statistical lower bound on the parameter estimate variance. This framework fairly predicts the T1/T2 mapping performances of any relaxometry approaches before phantom or in vivo experiments.

Mark Bydder¹, Gavin Hamilton², Ajinkya Desai², Elhamy R Heba², Tanya Wolfson², and Claude B Sirlin^2
1CRMBM UMR 7339, CNRS / Aix-Marseille Université, Marseille, France, ²University of California San Diego, CA, United States

This study investigates several sources of systematic errors affecting fat quantification by standard methods (chemical shift imaging). The influence from these different error parameters was assessed.

Jan Sedlacik¹, Dushyant Kumar¹, and Jens Fiehler^1
1University Medical Center Hamburg-Eppendorf, Hamburg, Hamburg, Germany

The qBOLD model describes the effect of deoxygenated blood of the capillary network on the signal decay. However, the oxygenation extraction fraction (OEF) and the deoxygenated blood volume (DBV) affect the signal decay very similarly, which allows reliable OEF and DBV estimation only for very high signal to noise ratios (SNR>500). We improved the fitting of the qBOLD model by obscuring the divergent global minimum root mean squared errors (RMSE) by adding noise and finding an effective global minimum by analyzing the local RMSE minima.

Feng Qi^1,2, Limiao Jiang^1,2, Quek Swee Tian¹, and Ng Thian C.^1,2
1Diagnostic Radiology, National University of Singapore, Singapore, Singapore, Singapore, ²Clinical Imaging Research Cente, A*STAR-NUS, Singapore, Singapore, Singapore

MR Fingerprint approach can combine multiple de/rephsing mechanisms to resolve microvascular parameters in high spatial resolution. We propose two new biomarkers, quadratic coefficient q and falling down parameter f, to enhance r-v assessment accuracy by building MRF dictionary in q-f feature space (qf-MRF). Feature parameters q-f are acquired by SE-EPI before echo time, which preserve more microstructural information than traditional acquisition of relaxation rate at echo time. qf-MRF requires only 4 EPI readout, costing less than tenth of the scanning time of tradition MRF, while the estimation accuracy is improved by 59% as compared with traditional identical weighting MRF (iw-MRF)

Pierre-Yves Baudin¹, Benjamin Marty^2,3, Ericky C.A. Araujo^2,3, Noura Azzabou^2,3, Pierre G. Carlier^2,3, and Paulo Loureiro de Sousa^4
1Consultants for Research in Imaging and Spectroscopy, Tournai, Belgium, ²NMR Laboratory, Institute of Myology, Paris, France, ³NMR Laboratory, CEA/I2BM/MIRCen, Paris, France,⁴ICube, Université de Strasbourg, CNRS, Strasbourg, France

T2 mapping from Multi-Slice Multi-Echo data is generally performed using a mono-exponential regression. However, the presence of stimulated echoes in the signal severely limits the performance of this approach and better signal models have been proposed, using the Bloch equations or the Extended-Phase-Graph (EPG) formalism. The case of mixed T2 decays – such as water and fat components in fatty infiltrated muscles - has been addressed via a non-negative least-square fitting of an EPG model, at the cost of long processing times. We propose a simpler and faster approach where a 2-component EPG model is explicitly fitted by dictionary searching.

Martina F. Callaghan¹, Shaihan J Malik², and Nikolaus Weiskopf^1
1Wellcome Trust Centre for Neuroimaging, UCL Institute of Neurology, London, London, United Kingdom, ²Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom

Quantitative MRI is attractive because of its standardised nature and higher specificity to underlying tissue microstructure. Acquiring spoiled gradient echo datasets with two different flip angles is an efficient approach to mapping the longitudinal relaxation rate with high resolution and whole brain coverage. However, this approach is biased when the transverse magnetisation is not fully spoiled. Correcting this effect relies on simulating the steady state signal taking account of the full suite of sequence parameters, including diffusion effects. Long simulations need to be recomputed for every sequence change. Here we demonstrate a rapid alternative using the extended phase graph approach.

Neil Peter Jerome¹, Vasia Papoutsaki¹, James A d'Arcy¹, Harold G Parkes¹, Nandita deSouza¹, Martin O Leach¹, and David J Collins^1
1Radiotherapy & Imaging, The Institute of Cancer Research, Sutton, London, United Kingdom

The inclusion of Dynamic Contrast Enhanced MRI into clinical trials, in particular those run across multiple centres, requires that quality assurance be performed, ensuring that functional parameters derived from different scanners may be usefully combined and compared. For DCE, the stability and noise characteristics are useful metrics for comparison; we present data from a multi-compartment test object with physiological T1s that allow comparison of DCE protocols with varying flip angles, as well as the same protocol acquired at two separate scanners, and demonstrate the usefulness of the test object for quality assurance of DCE-MRI.

Congyu Liao¹, Meng Chen¹, Darong Zhu², Hongjian He¹, Song Chen¹, Qiuping Ding¹, and Jianhui Zhong^1
1Center for Brain Imaging Science and Technology, Zhejiang University, Hangzhou, Zhejiang, China, ²Hangzhou First People's Hospital, Zhejiang, China

In this study, T1 mapping of a phantom with different concentration of Polyvinylpyrrolidone (PVP) solutions and a volunteer were acquired in two different 3T scanners using DESPOT1 method, and wild bootstrap analysis was used to assess both intra- and inter-site uncertainty of quantitative T1 measurements.

Busakol Ngammuang¹, Kittichai Wantanajittikul², Monruedee Tapanya¹, Suchaya Silvilairat³, Pimlak Charoenkwan³, and Suwit Saekho^1
1Department of Radiological Technology, Faculty of Associated Medical Sciences, Chiang Mai University, Chiang Mai, Thailand, ²Biomedical Engineering Center, Faculty of Engineering, Chiang Mai University, Chiang Mai, Thailand, ³Department of Pediatrics, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand

T2' is another MRI parameter that reflects solely to the magnetic inhomogeneity and shows high correlations to LIC, T2 and T2*. We compare liver T2' obtained from ROI and pixel based method by three curve-fitting models, mono-exponential, offset and truncation model. Fifteen β thalassemia major patients were study. Results showed that ROI based method with offset curve fitting model potentially provided no significantly difference outcomes to those of pixel based method for liver T2´ measurements.