Advertisement

Multi-centre and multi-vendor reproducibility of a standardized protocol for quantitative susceptibility Mapping of the human brain at 3T

Open AccessPublished:October 08, 2022DOI:https://doi.org/10.1016/j.ejmp.2022.09.012

      Highlights

      • Reproducibility of harmonized QSM protocol at 3T was tested in a multicentric study.
      • The traveling brain study yielded optimal structural similarity and reproducibility.
      • ROI-dependent χ quantification errors were 0.001–0.017 ppm for traveling brains.
      • In healthy subjects, ROI-dependent χ variability of 0.005–0.025 ppm can be expected.
      • The harmonized protocol yields reliable χ and it is ready for multicentric studies.

      Abstract

      Quantitative Susceptibility Mapping (QSM) is an MRI-based technique allowing the non-invasive quantification of iron content and myelination in the brain. The RIN – Neuroimaging Network established an optimized and harmonized protocol for QSM across ten sites with 3T MRI systems from three different vendors to enable multicentric studies. The assessment of the reproducibility of this protocol is crucial to establish susceptibility as a quantitative biomarker. In this work, we evaluated cross-vendor reproducibility in a group of six traveling brains. Then, we recruited fifty-one volunteers and measured the variability of QSM values in a cohort of healthy subjects scanned at different sites, simulating a multicentric study. Both voxelwise and Region of Interest (ROI)-based analysis on cortical and subcortical gray matter were performed.
      The traveling brain study yielded high structural similarity (∼0.8) and excellent reproducibility comparing maps acquired on scanners from two different vendors. Depending on the ROI, we reported a quantification error ranging from 0.001 to 0.017 ppm for the traveling brains. In the cohort of fifty-one healthy subjects scanned at nine different sites, the ROI-dependent variability of susceptibility values, of the order of 0.005–0.025 ppm, was comparable to the result of the traveling brain experiment.
      The harmonized QSM protocol of the RIN – Neuroimaging Network provides a reliable quantification of susceptibility in both cortical and subcortical gray matter regions and it is ready for multicentric and longitudinal clinical studies in neurological and pychiatric diseases.

      Graphical abstract

      Keywords

      Introduction

      The understanding of the pathological pattern of iron deposition and its causal relationship with neuronal loss is a crucial issue for the study of neurodegeneration [
      • Ndayisaba A.
      • Kaindlstorfer C.
      • Wenning G.K.
      Iron in Neurodegeneration – Cause or Consequence?.
      ,
      • Ward R.J.
      • Zucca F.A.
      • Duyn J.H.
      • Crichton R.R.
      • Zecca L.
      The role of iron in brain ageing and neurodegenerative disorders.
      ]. Non-invasive in-vivo quantification of iron stores can be performed using Quantitative Susceptibility Mapping (QSM), an advanced MRI technique [
      • de Rochefort L.
      • Liu T.
      • Kressler B.
      • Liu J.
      • Spincemaille P.
      • Lebon V.
      • et al.
      Quantitative susceptibility map reconstruction from MR phase data using bayesian regularization: validation and application to brain imaging.
      ,
      • Li L.
      • Leigh J.S.
      Quantifying arbitrary magnetic susceptibility distributions with MR.
      ,
      • Wharton S.
      • Schäfer A.
      • Bowtell R.
      Susceptibility mapping in the human brain using threshold-based k-space division.
      ]. QSM measures magnetic susceptibility (χ), which is an inherent magnetic property of tissues, from the phase signal obtained by Gradient Recalled Echo (GRE) acquisitions. The static magnetic field B0 of the MRI scanners induces a magnetization in tissues containing paramagnetic molecules, such as ferritin, or diamagnetic substances, like myelin or calcium, which in turn generate dipolar field perturbations. These perturbations lead to two main effects: on the one hand, transverse T2* relaxation is accelerated causing a faster decay of signal intensity; on the other hand, the phase of the signal accrues linearly with the dipolar field distortion. QSM aims at deconvolving the dipole field kernel which characterizes this perturbation in order to retrieve information on the susceptibility sources generating the perturbation, resulting in a quantitative and local map of magnetic susceptibility.
      The histological validation of QSM, showing the correlation of measured susceptibility with iron concentration and myelination, together with the ease of implementation as no dedicated hardware or sequence are required, increased QSM popularity and its application to a variety of disorders [
      • Ravanfar P.
      • Loi S.M.
      • Syeda W.T.
      • Van Rheenen T.E.
      • Bush A.I.
      • Desmond P.
      • et al.
      Systematic review: quantitative susceptibility mapping (QSM) of brain iron profile in neurodegenerative diseases.
      ,
      • Vinayagamani S.
      • Sheelakumari R.
      • Sabarish S.
      • Senthilvelan S.
      • Ros R.
      • Thomas B.
      • et al.
      Quantitative susceptibility mapping: technical considerations and clinical applications in neuroimaging.
      ], e.g. Parkinson’s disease [
      • Acosta-Cabronero J.
      • Cardenas-Blanco A.
      • Betts M.J.
      • Butryn M.
      • Valdes-Herrera J.P.
      • Galazky I.
      • et al.
      The whole-brain pattern of magnetic susceptibility perturbations in Parkinson’s disease.
      ,
      • Biondetti E.
      • Santin M.D.
      • Valabrègue R.
      • Mangone G.
      • Gaurav R.
      • Pyatigorskaya N.
      • et al.
      The spatiotemporal changes in dopamine, neuromelanin and iron characterizing Parkinson’s disease.
      ,
      • Pyatigorskaya N.
      • Sanz-Morère C.B.
      • Gaurav R.
      • Biondetti E.
      • Valabregue R.
      • Santin M.
      • et al.
      Iron imaging as a diagnostic tool for parkinson’s disease: a systematic review and meta-analysis.
      ,
      • Lancione M.
      • Donatelli G.
      • Del Prete E.
      • Campese N.
      • Frosini D.
      • Cencini M.
      • et al.
      Evaluation of iron overload in nigrosome 1 via quantitative susceptibility mapping as a progression biomarker in prodromal stages of synucleinopathies.
      ], Amyotrophic Lateral Sclerosis [
      • Acosta-Cabronero J.
      • Machts J.
      • Schreiber S.
      • Abdulla S.
      • Kollewe K.
      • Petri S.
      • et al.
      Quantitative susceptibility MRI to detect brain iron in amyotrophic lateral sclerosis.
      ,
      • Costagli M.
      • Symms M.R.
      • Angeli L.
      • Kelley D.A.C.
      • Biagi L.
      • Farnetani A.
      • et al.
      Assessment of silent T1-weighted head imaging at 7 T.
      ,
      • Wang C.
      • Foxley S.
      • Ansorge O.
      • Bangerter-Christensen S.
      • Chiew M.
      • Leonte A.
      • et al.
      Methods for quantitative susceptibility and R2* mapping in whole post-mortem brains at 7T applied to amyotrophic lateral sclerosis.
      ] or Alzheimer’s diseases [
      • Acosta-Cabronero J.
      • Williams G.B.
      • Cardenas-Blanco A.
      • Arnold R.J.
      • Lupson V.
      • Nestor P.J.
      • et al.
      In Vivo quantitative susceptibility mapping (QSM) in Alzheimer’s disease.
      ,
      • Ayton S.
      • Fazlollahi A.
      • Bourgeat P.
      • Raniga P.
      • Ng A.
      • Lim Y.Y.
      • et al.
      Cerebral quantitative susceptibility mapping predicts amyloid-β-related cognitive decline.
      ,
      • Cogswell P.M.
      • Wiste H.J.
      • Senjem M.L.
      • Gunter J.L.
      • Weigand S.D.
      • Schwarz C.G.
      • et al.
      Associations of quantitative susceptibility mapping with Alzheimer’s disease clinical and imaging markers.
      ,
      • Kim H.-G.
      • Park S.
      • Rhee H.Y.
      • Lee K.M.
      • Ryu C.-W.
      • Rhee S.J.
      • et al.
      Quantitative susceptibility mapping to evaluate the early stage of Alzheimer’s disease.
      ], and pioneered the use of QSM as a quantitative MRI biomarker. However, an MRI biomarker should be fully quantitative to make an impactful contribution to the diagnosis, prognosis, and follow-up of patients, both during the natural history of pathology and during treatments. The assessment of its reliability in time and space is fundamental to evaluate the reproducibility of results in longitudinal and in multi-center/multi-vendor studies.
      The impact of mono-centric studies is limited by the small cohorts recruited for research purposes, especially for low-prevalence diseases such as atypical Parkinsonism (e.g. Multiple System Atrophy), Friedreich ataxia or Huntington’s disease. The optimization and the harmonization of acquisition protocols across a wide network of hospitals and research centers would enable the enrollment of larger populations of patients and a more comprehensive study of risk factors and other relevant clinical variables. However, susceptibility measured via QSM was shown to depend on several experimental factors including both fundamental physical properties of susceptibility and acquisition parameters. Magnetic susceptibility could be more rigorously modeled as a second-order tensor and its estimation would require at least six acquisitions with the patient's head oriented differently with respect to the main static magnetic field B0 [
      • Li W.
      • Wu B.
      • Avram A.V.
      • Liu C.
      Magnetic susceptibility anisotropy of human brain in vivo and its molecular underpinnings.
      ,
      • Liu C.
      • Li W.
      • Johnson G.A.
      • Wu B.
      High-field (9.4T) MRI of brain dysmyelination by quantitative mapping of magnetic susceptibility.
      ,
      • Wharton S.
      • Bowtell R.
      Effects of white matter microstructure on phase and susceptibility maps.
      ]. As this is not feasible in clinical practice due to patient discomfort and long scanning time, QSM approaches approximate susceptibility to a scalar quantity. This leads to χ dependence on head orientation, especially in highly organized structures such as white matter fiber bundles where myelin sheath microstructure is responsible for susceptibility anisotropy, while this should not represent a confounding factor in gray matter [
      • Lancione M.
      • Tosetti M.
      • Donatelli G.
      • Cosottini M.
      • Costagli M.
      The impact of white matter fiber orientation in single-acquisition quantitative susceptibility mapping.
      ]. Acquisition parameters affecting χ quantification are spatial resolution [
      • Karsa A.
      • Punwani S.
      • Shmueli K.
      The effect of low resolution and coverage on the accuracy of susceptibility mapping.
      ,
      • Zhou D.
      • Cho J.
      • Zhang J.
      • Spincemaille P.
      • Wang Y.
      Susceptibility underestimation in a high-susceptibility phantom: dependence on imaging resolution, magnitude contrast, and other parameters.
      ], brain coverage [
      • Karsa A.
      • Punwani S.
      • Shmueli K.
      The effect of low resolution and coverage on the accuracy of susceptibility mapping.
      ,
      • Elkady A.M.
      • Sun H.
      • Wilman A.H.
      Importance of extended spatial coverage for quantitative susceptibility mapping of iron-rich deep gray matter.
      ], echo time [
      • Biondetti E.
      • Karsa A.
      • Thomas D.L.
      • Shmueli K.
      Investigating the accuracy and precision of TE-dependent versus multi-echo QSM using Laplacian-based methods at 3 T.
      ,
      • Cronin M.J.
      • Wang N.
      • Decker K.S.
      • Wei H.
      • Zhu W.-Z.
      • Liu C.
      Exploring the origins of echo-time-dependent quantitative susceptibility mapping (QSM) measurements in healthy tissue and cerebral microbleeds.
      ,
      • Lancione M.
      • Cencini M.
      • Costagli M.
      • Donatelli G.
      • Tosetti M.
      • Giannini G.
      • et al.
      Diagnostic accuracy of quantitative susceptibility mapping in multiple system atrophy: the impact of echo time and the potential of histogram analysis.
      ,
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ,
      • Sood S.
      • Urriola J.
      • Reutens D.
      • O’Brien K.
      • Bollmann S.
      • Barth M.
      • et al.
      Echo time-dependent quantitative susceptibility mapping contains information on tissue properties.
      ] and field strength [
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ,
      • Spincemaille P.
      • Anderson J.
      • Wu G.
      • Yang B.
      • Fung M.
      • Li K.e.
      • et al.
      Quantitative susceptibility mapping: MRI at 7T versus 3T.
      ,
      • Emmerich J.
      • Bachert P.
      • Ladd M.E.
      • Straub S.
      On the separation of susceptibility sources in quantitative susceptibility mapping: theory and phantom validation with an in vivo application to multiple sclerosis lesions of different age.
      ]. For this reason, multicentric studies require careful optimization and standardization of the GRE sequence used to perform QSM across clinical systems from different vendor platforms and the assessment of the reproducibility of the biomarkers of interest.
      Previous studies assessed QSM reproducibility in multi-centric studies at 1.5 T, 3T and 7 T using scanner platforms from one [
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ,
      • Spincemaille P.
      • Anderson J.
      • Wu G.
      • Yang B.
      • Fung M.
      • Li K.e.
      • et al.
      Quantitative susceptibility mapping: MRI at 7T versus 3T.
      ,
      • Hinoda T.
      • Fushimi Y.
      • Okada T.
      • Fujimoto K.
      • Liu C.
      • Yamamoto A.
      • et al.
      Quantitative susceptibility mapping at 3 T and 1.5 T.
      ,
      • Voelker M.N.
      • Kraff O.
      • Goerke S.
      • Laun F.B.
      • Hanspach J.
      • Pine K.J.
      • et al.
      The traveling heads 2.0: multicenter reproducibility of quantitative imaging methods at 7 Tesla.
      ,
      • Wang R.
      • Xie G.
      • Zhai M.
      • Zhang Z.
      • Wu B.
      • Zheng D.
      • et al.
      Stability of R2* and quantitative susceptibility mapping of the brain tissue in a large scale multi-center study.
      ] or two vendors [
      • Deh K.
      • Nguyen T.D.
      • Eskreis-Winkler S.
      • Prince M.R.
      • Spincemaille P.
      • Gauthier S.
      • et al.
      Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
      ,
      • Lin P.-Y.
      • Chao T.-C.
      • Wu M.-L.
      Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
      ,
      • Rua C.
      • Clarke W.T.
      • Driver I.D.
      • Mougin O.
      • Morgan A.T.
      • Clare S.
      • et al.
      Multi-centre, multi-vendor reproducibility of 7T QSM and R2* in the human brain: results from the UK7T study.
      ]. Good reproducibility was reported for repeated scans on the same system and for standardized protocols across scanners operating at the same field strength, while to obtain consistent results at different field strengths required careful optimization of echo times [
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ].
      The RIN – Neuroimaging Network (https://www.reteneuroscienze.it/en/progetti/neuroimaging/) aimed at creating a harmonized protocol, including a GRE sequence for QSM, and at setting it up on several clinical 3T MRI scanners from three vendors, in order to be able to aggregate data and boost our knowledge on brain structure and function in physiological and pathological conditions.
      In this study, a harmonized multi-echo GRE sequence was acquired at ten different sites across Italy on a population of healthy adults. In addition, four sites were involved in a “traveling brain” experiment in which six volunteers were scanned on 3T MRI systems from different vendors. We performed voxel-wise and ROI-based analyses on both cortical and subcortical structures to assess the similarity of the maps and the reproducibility of χ maps. We also provided an estimation of the susceptibility variability that can be expected over a healthy population for inter-site, inter-vendor or cross-vendor acquisitions.

      Methods

      Experimental design

      This study involved sites of the RIN - Neuroimaging Network equipped with 3T MRI scanners and data were acquired on scanner platforms from three vendors (Philips, Siemens and GE Healthcare).
      For the traveling brain experiment, four sites located in two different areas of Italy (two in area A1 and two in A2) with scanners from two different vendors (V1 and V2) were selected. Three healthy subjects (S1-3; 1 female and 2 males, 29, 30 and 33 years old) were scanned at both A1 sites and three other subjects (S4-6; 2 females and 1 male, 21, 21 and 43 years old) at both A2 sites, so that each subject was scanned once for vendor V1 and once for V2. This information is summarized in Table 1.
      Table 1Details on the hardware and the cohort of traveling brains for each site involved in the study.
      AreaSiteVendorCoil [ch]SubjectsAge [y]Sex
      A11V132329, 30, 331F/2M
      2V264
      A23V132321, 21, 432F/1M
      4V232
      To assess inter-site variability on a healthy population, fifty-one volunteers (33 females and 18 males, 29 ± 5 [
      • Wharton S.
      • Bowtell R.
      Effects of white matter microstructure on phase and susceptibility maps.
      ,
      • Lancione M.
      • Tosetti M.
      • Donatelli G.
      • Cosottini M.
      • Costagli M.
      The impact of white matter fiber orientation in single-acquisition quantitative susceptibility mapping.
      ,
      • Karsa A.
      • Punwani S.
      • Shmueli K.
      The effect of low resolution and coverage on the accuracy of susceptibility mapping.
      ,
      • Zhou D.
      • Cho J.
      • Zhang J.
      • Spincemaille P.
      • Wang Y.
      Susceptibility underestimation in a high-susceptibility phantom: dependence on imaging resolution, magnitude contrast, and other parameters.
      ,
      • Elkady A.M.
      • Sun H.
      • Wilman A.H.
      Importance of extended spatial coverage for quantitative susceptibility mapping of iron-rich deep gray matter.
      ,
      • Biondetti E.
      • Karsa A.
      • Thomas D.L.
      • Shmueli K.
      Investigating the accuracy and precision of TE-dependent versus multi-echo QSM using Laplacian-based methods at 3 T.
      ,
      • Cronin M.J.
      • Wang N.
      • Decker K.S.
      • Wei H.
      • Zhu W.-Z.
      • Liu C.
      Exploring the origins of echo-time-dependent quantitative susceptibility mapping (QSM) measurements in healthy tissue and cerebral microbleeds.
      ,
      • Lancione M.
      • Cencini M.
      • Costagli M.
      • Donatelli G.
      • Tosetti M.
      • Giannini G.
      • et al.
      Diagnostic accuracy of quantitative susceptibility mapping in multiple system atrophy: the impact of echo time and the potential of histogram analysis.
      ,
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ,
      • Sood S.
      • Urriola J.
      • Reutens D.
      • O’Brien K.
      • Bollmann S.
      • Barth M.
      • et al.
      Echo time-dependent quantitative susceptibility mapping contains information on tissue properties.
      ,
      • Spincemaille P.
      • Anderson J.
      • Wu G.
      • Yang B.
      • Fung M.
      • Li K.e.
      • et al.
      Quantitative susceptibility mapping: MRI at 7T versus 3T.
      ,
      • Emmerich J.
      • Bachert P.
      • Ladd M.E.
      • Straub S.
      On the separation of susceptibility sources in quantitative susceptibility mapping: theory and phantom validation with an in vivo application to multiple sclerosis lesions of different age.
      ,
      • Hinoda T.
      • Fushimi Y.
      • Okada T.
      • Fujimoto K.
      • Liu C.
      • Yamamoto A.
      • et al.
      Quantitative susceptibility mapping at 3 T and 1.5 T.
      ,
      • Voelker M.N.
      • Kraff O.
      • Goerke S.
      • Laun F.B.
      • Hanspach J.
      • Pine K.J.
      • et al.
      The traveling heads 2.0: multicenter reproducibility of quantitative imaging methods at 7 Tesla.
      ,
      • Wang R.
      • Xie G.
      • Zhai M.
      • Zhang Z.
      • Wu B.
      • Zheng D.
      • et al.
      Stability of R2* and quantitative susceptibility mapping of the brain tissue in a large scale multi-center study.
      ,
      • Deh K.
      • Nguyen T.D.
      • Eskreis-Winkler S.
      • Prince M.R.
      • Spincemaille P.
      • Gauthier S.
      • et al.
      Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
      ,
      • Lin P.-Y.
      • Chao T.-C.
      • Wu M.-L.
      Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
      ,
      • Rua C.
      • Clarke W.T.
      • Driver I.D.
      • Mougin O.
      • Morgan A.T.
      • Clare S.
      • et al.
      Multi-centre, multi-vendor reproducibility of 7T QSM and R2* in the human brain: results from the UK7T study.
      ,
      • Nigri A.
      • Ferraro S.
      • Gandini Wheeler-Kingshott C.A.
      • Tosetti M.
      • Redolfi A.
      • Forloni G.
      • et al.
      Quantitative MRI harmonization to maximize clinical impact: the RIN–Neuroimaging Network.
      ,
      • Schofield M.A.
      • Zhu Y.
      Fast phase unwrapping algorithm for interferometric applications.
      ] years old) underwent a 3T MRI scan in one of nine sites with MR systems from three different vendors (V1, V2, V3): five sites from V1, three sites from V2 and one from V3. Demographic and technical information on the subject cohort and the MRI system of each site are reported in Table 2.
      Table 2Details on the hardware and the cohort of healthy subjects for each site involved in the study.
      SiteVendorCoil [ch]SubjectsAge [y]Sex
      1V132532 ± 23F/2M
      2V132630 ± 41F/5M
      3V132532 ± 53F/2M
      4V132325 ± 21F/2M
      5V132532 ± 64F/1M
      6V264626 ± 62F/4M
      7V264725 ± 36F/1M
      8V212731 ± 53F/4M
      9V332729 ± 34F/3M
      The participants had no history of neurological diseases or psychiatric disorders and gave their written informed consent. The Protocol Study was performed under the Declaration of Helsinki (59th General Assembly of the World Medical Association, Seoul, October 2008) and the Medical Research Involving Human Subjects Act (WMO). All the procedures described were performed in compliance with security, integrity, and privacy. The studies involving human participants were reviewed and approved by the Ethics Committee of Lombardy Region - IRCCS Institute of Neurology Carlo Besta Foundation Section.

      MRI imaging protocol

      The MRI protocol was optimized and standardized across sites on 3T MRI platforms from three different vendors [
      • Nigri A.
      • Ferraro S.
      • Gandini Wheeler-Kingshott C.A.
      • Tosetti M.
      • Redolfi A.
      • Forloni G.
      • et al.
      Quantitative MRI harmonization to maximize clinical impact: the RIN–Neuroimaging Network.
      ], in order to minimize inter-vendor differences in acquisition parameters. The Standard Operating Procedures (SOPs) required that subjects be positioned inside the scanner so that the bicallosal alignment and the pure axial alignment corresponded, to minimize the effect of different head orientations across subjects. For susceptibility mapping, a flow-compensated 3D multi-echo GRE sequence with whole-brain coverage was prescribed axially for each subject. Details on the acquisition parameters for each vendor defined by the SOPs are reported in Table 3. Both magnitude and phase images were collected on Philips and Siemens MRI systems, while on GE scanners we acquired the magnitude and the real and imaginary parts of the data which were used to produce phase images. In addition, a sagittal whole-brain 3D T1-weighted sequence with 1-mm isotropic resolution was acquired for anatomical reference.
      Table 3Acquisition parameters for 3D multi-echo GRE sequence of each vendor.
      VendorPhilipsSiemensGE Healthcare
      Sequence type3D GR MULTIECHO

      FFE-3D
      3D GRE (swi3d8r)3D SWAN
      FOV [mm]224 × 224224 × 224224 × 224
      Resolution [mm3]1 × 1 × 11 × 1 × 11 × 1 × 1
      Matrix224 × 224224 × 224224 × 224
      Slice thickness [mm]111
      Slice gap [mm]000
      # slices140144140
      TR [ms]405151
      Average TE [ms]2125.227.4
      First TE [ms]5.45.65.6
      # echoes787
      Echo spacing [ms]5.25.65.6
      Flip angle (deg)181818
      k-space coverage6/80.85
      Slice k-space coverage6/8
      Acceleration factorSENSE = 2GRAPPA = 2ASSET = 2
      FilterCLEAR onPrescan Normalize on
      Bandwidth [Hz/px]271340279
      Duration [min]≈ 8:11≈ 8:45≈ 8:01

      Data processing

      Susceptibility maps were computed using the algorithms implemented in STI Suite (MATLAB toolbox, available at https://people.eecs.berkeley.edu/∼chunlei.liu/software.html from UC Berkeley, Berkeley, CA, USA) as follows. The raw phase images acquired at each echo were unwrapped using a Laplacian-based method [
      • Li W.
      • Wu B.
      • Avram A.V.
      • Liu C.
      Magnetic susceptibility anisotropy of human brain in vivo and its molecular underpinnings.
      ,
      • Schofield M.A.
      • Zhu Y.
      Fast phase unwrapping algorithm for interferometric applications.
      ]. The T2*-weighted image averaged across echoes was skull-stripped using Brain Extraction Toolbox (bet) [
      • Smith S.M.
      Fast robust automated brain extraction.
      ] in FSL 5.0.9 (FMRIB Software Library, Oxford Centre for Functional MRI of the Brain, Oxford, UK) and the brain mask was used to remove the background field via V-SHARP [
      • Schweser F.
      • Deistung A.
      • Lehr B.W.
      • Reichenbach J.R.
      Quantitative imaging of intrinsic magnetic tissue properties using MRI signal phase: an approach to in vivo brain iron metabolism?.
      ]. Finally, we applied the iLSQR algorithm [
      • Li W.
      • Wang N.
      • Yu F.
      • Han H.
      • Cao W.
      • Romero R.
      • et al.
      A method for estimating and removing streaking artifacts in quantitative susceptibility mapping.
      ,
      • Li W.
      • Wu B.
      • Liu C.
      Quantitative susceptibility mapping of human brain reflects spatial variation in tissue composition.
      ] to obtain QSM images for each echo that were then averaged to increase SNR [
      • Denk C.
      • Rauscher A.
      Susceptibility weighted imaging with multiple echoes.
      ].
      The T1-weighted images were processed in FreeSurfer 6.0 [
      • Desikan R.S.
      • Ségonne F.
      • Fischl B.
      • Quinn B.T.
      • Dickerson B.C.
      • Blacker D.
      • et al.
      An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest.
      ,
      • Fischl B.
      • Salat D.H.
      • Busa E.
      • Albert M.
      • Dieterich M.
      • Haselgrove C.
      • et al.
      Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain.
      ] for brain extraction and for the parcellation of cortical regions and the segmentation of subcortical nuclei. The skull-stripped TE-averaged T2*-weighted image of each subject was aligned to the corresponding T1-weighted image via an affine transformation using antsRegistrationSyN routine in ANTs [
      • Avants B.B.
      • Epstein C.L.
      • Grossman M.
      • Gee J.C.
      Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain.
      ]. The computed transformation was then inverted and applied to the segmentation from which eight Regions Of Interest (ROI), each divided into left (L) and right (R) ROIs, were selected. Specifically, ROI-based analysis was performed in four subcortical regions, i.e. Caudate Nucleus (CN), Putamen (Pu), Globus Pallidus (GP) and Hippocampus (HC), and in four cortical areas, i.e. Cuneus (Cu), Precuneus (PCu), Precentral Gyrus (PCG) and Superior Frontal Gyrus (SFG).
      For a voxel-wise analysis on traveling brain data, the TE-averaged T2*-weighted images of the same subject in different sites were co-registered via an affine transformation computed using antsRegistration in ANTs, which was then applied to susceptibility maps.

      Statistical analysis

      Statistical analysis was performed in MATLAB (MathWorks, Natick, MA, USA). Data from the traveling brain experiment were analyzed to assess inter-vendor reproducibility. A quantitative voxel-wise comparison was performed by computing structural similarity (SSIM) [
      • Wang Z.
      • Bovik A.C.
      • Sheikh H.R.
      • Simoncelli E.P.
      Image quality assessment: from error visibility to structural similarity.
      ] using the ssim built-in function in MATLAB, and root-mean-squared error (RMSE) [
      • Langkammer C.
      • Schweser F.
      • Shmueli K.
      • Kames C.
      • Li X.u.
      • Guo L.i.
      • et al.
      Quantitative susceptibility mapping: report from the 2016 reconstruction challenge.
      ] as follows:
      RMSE=i=1nχ1i-χ2i2n


      where χ1i and χ2i indicate measured susceptibility for the i-th voxel at the two sites and n is the number of voxels. Estimation of SSIM and RMSE were provided both globally, i.e. by considering all the voxels in the brain mask, and locally by computing them in a moving 3 mm-radius region. In addition, we computed the voxel-wise difference (DIFF) between the two repeated scans.
      Reproducibility evaluation was performed by computing an orthogonal linear fit and Pearson's correlation considering the susceptibility values in the voxels inside the brain mask for the scans performed at the two sites for each subject. An analogous ROI-based analysis was conducted after extracting average susceptibility and standard deviations from the selected ROIs. Via Bland-Altman plots, the mean difference and the 95 % limits of agreement were calculated for both voxel-wise and ROI-based analyses.
      Cross-site variability of susceptibility measurements on a healthy population was performed considering average and standard deviation of susceptibility values for each ROI independently. Specifically, cross-subject variability within-site, within-vendor and cross-vendors were calculated via the between-subjects standard deviation for each ROI.

      Results

      Susceptibility maps for two exemplary subjects, one from area A1 and one from A2, scanned at two sites on scanners from two different vendors are shown in Fig. 1, together with the maps of SSIM and RMSE computed using a searchlight analysis approach, and the voxel-wise DIFF map. A global estimation of SSIM considering all the voxels in the brain mask yielded an across-subject average of 0.58 ± 0.11 and 0.60 ± 0.04 for A1 and A2 respectively, while the global RMSE was respectively 0.028 ± 0.005 ppm and 0.029 ± 0.0023 ppm. However, the computation of local SSIM and RMSE values highlighted that very low SSIM and high RMSE values are limited to particular regions, mainly located at the borders of the brain, at the skull base and at the edges of the cortex. Overall, high SSIM values were reached for both A1 and A2, with a median of 0.66 ± 0.09 and 0.65 ± 0.05 on average across subjects and a mode of 0.85 ± 0.08 and 0.72 ± 0.12 for A1 and A2 respectively. The median of RMSE distribution was 0.018 ± 0.004 ppm and 0.017 ± 0.002 ppm while the mode was 0.009 ± 0.002 ppm and 0.010 ± 0.003 ppm on average across subjects for A1 and A2 respectively.
      Figure thumbnail gr1
      Fig. 1Left panel: susceptibility maps obtained for two representative subjects scanned on the MRI platforms from vendors V1 and V2 of areas A1 and A2. Right panel: local estimation of SSIM and RMSE computed using a searchlight analysis approach and voxelwise difference map (DIFF).
      The results of the voxelwise assessment of reproducibility are reported in Table 4. Good Pearson’s correlation was found and the orthogonal linear fit yielded angular coefficients close to 1, indicating high correspondence between χ values measured at the two sites, especially for A1, while the small intercepts indicate the absence of offset. No bias was reported in the Bland-Altman analysis and the 95 %-confidence intervals were C.I.A1 = 0.045 ± 0.009 ppm and C.I.A2 = 0.047 ± 0.002 ppm.
      Table 4Results for the reproducibility assessment in the traveling brain experiment obtained via voxelwise and ROI-based analysis. The angular coefficient and the intercept of the orthogonal linear fit, Pearson’s correlation coefficient and Bland-Altman analysis parameters are reported.
      Voxelwise analysisROI-based analysis
      A1A2A1A2
      FitAngular coefficient m1.01 ± 0.050.77 ± 0.030.970.82
      Intercept q [ppm](5 ± 3)·10-4(13 ± 4)·10-40.00130.0035
      Pearson’s r0.74 ± 0.080.69 ± 0.040.980.97
      Bland-AltmanMean difference [ppm](-5 ± 2)·10-4(-13 ± 6)·10-4−4·10-43·10-4
      95 % C.I. [ppm]0.045 ± 0.0090.047 ± 0.0020.020.02
      The results of the ROI-based analysis are reported in Table 4 and displayed in Fig. 2. High correlation between χ values measured on the scanner from V1 and V2 was observed in both A1 and A2 groups. The orthogonal linear fit yielded angular coefficients close to 1, indicating excellent reproducibility, though slightly lower for A2 than for A1. The subject-average difference between χ measures at the two sites ranged from 0.002 and 0.014 ppm with a median of 0.006 ppm for A1 depending on the ROI and from 0.001 to 0.017 ppm with a median of 0.007 ppm for A2. The Bland-Altman plots revealed the absence of biases and a 95 %-confidence interval C.I. = 0.02 ppm. However, for A2 a trend can be observed in susceptibility differences, being negative for negative χ and becoming positive for increasing χ.
      Figure thumbnail gr2
      Fig. 2Scatter plot (top row) and Bland-Altman plot (bottom row) of the ROI-based susceptibility measurement performed on scanners from vendors V1 and V2 in areas A1 (left column) and A2 (right column).
      Susceptibility values obtained for each ROI at each site are shown in Fig. 3. Cross-subject variability within-site was measured via the between-subjects standard deviation averaged across ROIs and ranged from 0.0064 ppm to 0.0083 ppm with a median of 0.0073 ppm. Cross-sites within-vendor variability was 0.0079 ppm, 0.0074 ppm and 0.0071 ppm for V1, V2 and V3 respectively on average across ROIs, while the average standard deviation across all sites and vendors was 0.0085 ppm. The average susceptibility for each ROI and the cross-vendor, within-vendor and within-site standard deviations are reported in Table 5. Fig. 4 shows the average susceptibility across the whole population and the susceptibility averaged across sites with MRI platforms from the same vendor with the corresponding standard deviations (See Table 5).
      Figure thumbnail gr3
      Fig. 3Susceptibility values obtained for each ROI at each site. The first five blue boxes refer to sites equipped with MRI systems from vendor V1, the following three pink boxes from vendor V2, and the last yellow box from vendor V3. The bottom and top edges of each box indicate the 25th and 75th percentile respectively of the susceptibility distribution at each site for a particular ROI and the central line indicates the median. The whiskers extend to the maximum and minimum values excluding outliers, which are plotted using the ‘+’ symbol.
      Table 5Susceptibility values and standard deviations measured on groups of healthy controls. As only one scanner from vendor V3 was available in this study, the within-vendor standard deviation for V3 coincides with the within-site estimation of QSM variability.
      Susceptibility values [ppm]
      ROICross-vendorWithin-vendorWithin-site
      MeanStdV1 stdV2 stdV3 stdMin stdMax stdMean std
      GP L0.1100.0220.0190.0170.0240.0130.0240.019
      GP R0.1110.0170.0160.0160.0170.0080.0220.016
      CN L0.0480.0110.0090.0080.0090.0060.0130.009
      CN R0.0500.0100.0090.0090.0080.0060.0140.009
      Pu L0.0390.0120.0120.0120.0130.0060.0180.012
      Pu R0.0380.0110.0120.0100.0100.0060.0170.011
      HC L−0.00650.00640.00750.00480.00440.00340.00900.0057
      HC R−0.00600.00560.00550.00630.00340.00340.00800.0052
      Cu L0.00500.00530.00520.00440.00540.00240.00660.0048
      Cu R0.00870.00670.00480.00570.00410.00340.00730.0050
      PCu L0.00220.00560.00440.00360.00220.00140.00580.0035
      PCu R0.00460.00580.00380.00330.00220.00160.00520.0035
      PCG L0.00750.00510.00470.00490.00400.00240.00600.0041
      PCG R0.00560.00470.00470.00330.00350.00220.00480.0036
      SFG L0.00280.00460.00370.00510.00160.00160.00670.0038
      SFG R0.00200.00330.00280.00390.00150.00150.00460.0031
      Figure thumbnail gr4
      Fig. 4Average susceptibility values and corresponding standard deviations. The gray dots represent the mean of the whole cohort of subjects across all sites, while the triangles and the square markers indicate average susceptibility values measured on average across sites from each of the three vendors.

      Discussion

      In this work, we assessed the reproducibility of susceptibility maps obtained via QSM in a multicentric study at 3T using an optimized and harmonized protocol across MRI scanner platforms from three different vendors, consisting of a multi-echo GRE sequence with 1 mm isotropic spatial resolution. A traveling brain experiment was run on four different scanners from two vendors while nine sites collected data on groups of healthy volunteers to provide an estimation of the variability that can be expected in a multicentric multi-vendor study across a normative population.
      The traveling brain experiment showed high SSIM (∼ 0.8) and low RMSE (∼ 0.01 ppm) of χ maps acquired on scanners from two different vendors across the whole brain, with some exceptions: areas at the border of the brain, at the skull base or near the interhemispheric or the transverse fissures suffer from the proximity to air-tissue interfaces or sinuses that may cause excessive phase wrap leading to incomplete dipole inversion, while χ comparisons at the edges of the cortex can be affected by registration errors. Excellent reproducibility was reported in both voxelwise and ROI-based analyses (m ∼ 0.8–1). Voxelwise analysis yielded a good correlation (r ∼ 0.7), partially affected by voxels in the previously mentioned areas showing higher RMSE, while excellent correlation (r ∼ 1) was found in ROI-based analysis. Depending on the ROI, we reported inter-site χ differences of 0.001–0.017 ppm on average across participants in the traveling brain experiment. This is consistent with previous multi-sites reproducibility studies at 3T, reporting inter-site standard deviation of 0.006–0.012 ppm [
      • Lin P.-Y.
      • Chao T.-C.
      • Wu M.-L.
      Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
      ]. In addition, the 95 % C.I. (1.96 times the standard deviation) of 0.04 ppm and 0.02 ppm for voxelwise and ROI-based analysis are compatible with previously reported inter-vendor 95 % C.I. [
      • Deh K.
      • Nguyen T.D.
      • Eskreis-Winkler S.
      • Prince M.R.
      • Spincemaille P.
      • Gauthier S.
      • et al.
      Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
      ].
      The within-site variability across subjects measured via the between-subjects standard deviation is slightly smaller on average than the inter-site within-vendor variability, which is in turn slightly smaller than the cross-vendor standard deviation. However, the variability estimations are very similar, being of the order of 0.01 ppm for deep gray matter nuclei except for HC, and of the order of 0.005 ppm for cortical regions and HC, even though the scanners involved in the study are equipped with different hardware and operate with different software versions, even when they refer to the same vendor. These calculations were performed on separated cohorts of healthy subjects which differed from one site to the other rather than a group of traveling brains, so taking into account the physiological variance observable for different subjects and reproducing more realistically the conditions of a multicentric study. Despite this additional contribution to the variance, both the within-site and cross-site variability are comparable to the standard deviation obtained by the Bland-Altman analysis on the traveling brain data and to what was reported in previous traveling brain studies [
      • Deh K.
      • Nguyen T.D.
      • Eskreis-Winkler S.
      • Prince M.R.
      • Spincemaille P.
      • Gauthier S.
      • et al.
      Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
      ,
      • Lin P.-Y.
      • Chao T.-C.
      • Wu M.-L.
      Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
      ].
      The QSM acquisition protocol implemented in this study was set up using product pulse sequences already installed on the MRI systems, with no additional programming required, and the reconstruction algorithms provided by the vendors in order to be able to easily export the protocol to other clinical hospitals or research centers and extend or create new multicenter networks. The drawback for portability is that the acquisition is not locally-optimized and may not represent the most performing option at a particular site. The sequence used in this study had an acceptable but slightly long acquisition time (∼ 8–9 min). However, some acceleration approaches such as compressed sensing were not available at all sites and are not widely spread in clinical settings, so they were not employed in this work. The spatial resolution obtained with this protocol may serve many clinical aims, but partial volume effect may limit studies targeting small subcortical structures such as nigrosome 1 or locus coeruleus. In this study we chose to use a well-established QSM pipeline [
      • Costagli M.
      • Donatelli G.
      • Cecchi P.
      • Bosco P.
      • Migaleddu G.
      • Siciliano G.
      • et al.
      Distribution indices of magnetic susceptibility values in the primary motor cortex enable to classify patients with amyotrophic lateral sclerosis.
      ,
      • Howard C.M.
      • Jain S.
      • Cook A.D.
      • Packard L.E.
      • Mullin H.A.
      • Chen N.
      • et al.
      Cortical iron mediates age-related decline in fluid cognition.
      ,
      • Monti S.
      • Pontillo G.
      • Russo C.
      • Cella L.
      • Cocozza S.
      • Palma G.
      RESUMEN: a flexible class of multi-parameter qMRI protocols.
      ,
      • Ward P.G.D.
      • Harding I.H.
      • Close T.G.
      • Corben L.A.
      • Delatycki M.B.
      • Storey E.
      • et al.
      Longitudinal evaluation of iron concentration and atrophy in the dentate nuclei in friedreich ataxia.
      ]. As different pipelines produce slightly different quantitative results [
      • Bilgic B.
      • Langkammer C.
      • Marques J.P.
      • Meineke J.
      • Milovic C.
      • Schweser F.
      QSM reconstruction challenge 2.0: Design and report of results.
      ], it is critical to standardize the QSM reconstruction pipeline when pooling data from different sites.
      QSM cannot quantify susceptibility in absolute terms, but only with respect to a reference value, due to an unknown region-independent offset [
      • Cheng Y.C.N.
      • Neelavalli J.
      • Haacke E.M.
      Limitations of calculating field distributions and magnetic susceptibilities in MRI using a Fourier based method.
      ,
      • Koch K.M.
      • Papademetris X.
      • Rothman D.L.
      • De Graaf R.A.
      Rapid calculations of susceptibility-induced magnetostatic field perturbations for in vivo magnetic resonance.
      ]. Susceptibility measures should be then referred to a reference region. Criteria for reference region selection include easy delineation and segmentation, susceptibility ideally not affected by age, head orientation or disease condition. Several candidate brain regions have been suggested [
      • Feng X.
      • Deistung A.
      • Reichenbach J.R.
      Quantitative susceptibility mapping (QSM) and R2* in the human brain at 3 T: evaluation of intra-scanner repeatability.
      ,
      • Straub S.
      • Schneider T.M.
      • Emmerich J.
      • Freitag M.T.
      • Ziener C.H.
      • Schlemmer H.-P.
      • et al.
      Suitable reference tissues for quantitative susceptibility mapping of the brain.
      ], such as frontal white matter, internal capsule, cerebrospinal fluids (CSF) in ventricles and the whole brain. However, measured susceptibility white matter tracts may be influenced by head orientation [
      • Li W.
      • Wu B.
      • Avram A.V.
      • Liu C.
      Magnetic susceptibility anisotropy of human brain in vivo and its molecular underpinnings.
      ,
      • Liu C.
      • Li W.
      • Johnson G.A.
      • Wu B.
      High-field (9.4T) MRI of brain dysmyelination by quantitative mapping of magnetic susceptibility.
      ,
      • Wharton S.
      • Bowtell R.
      Effects of white matter microstructure on phase and susceptibility maps.
      ,
      • Lancione M.
      • Tosetti M.
      • Donatelli G.
      • Cosottini M.
      • Costagli M.
      The impact of white matter fiber orientation in single-acquisition quantitative susceptibility mapping.
      ], while the ventricles do not only contain CSF and the presence of veins and choroid plexus can alter susceptibility. Moreover, ventricles are typically small in young healthy subjects and their susceptibility may be affected by partial volume effect. These factors make the delineation of an ROI in the ventricles complex and operator-dependent. Previous studies reported that whole-brain referencing reduces cross-site and within-site QSM variability with respect to CSF referencing [
      • Rua C.
      • Clarke W.T.
      • Driver I.D.
      • Mougin O.
      • Morgan A.T.
      • Clare S.
      • et al.
      Multi-centre, multi-vendor reproducibility of 7T QSM and R2* in the human brain: results from the UK7T study.
      ] and that the referencing operation represents a small adjustment compared to other factors such as age [
      • Acosta-Cabronero J.
      • Betts M.J.
      • Cardenas-Blanco A.
      • Yang S.
      • Nestor P.J.
      In Vivo MRI Mapping of brain iron deposition across the adult lifespan.
      ]. For this reason, as the QSM processing pipeline used in this study sets to zero the average susceptibility in the whole brain mask, we left measured χ values unreferenced. Similarly, the ROI-based analysis was performed using automatic segmentations of deep gray matter nuclei and automatic cortical parcellations performed on the T1-weighted images to eliminate operator-dependent bias.
      The harmonization across MRI systems from different vendors was not trivial and different sequence implementations led to different echo trains. As matching TEs is crucial to achieve reproducibility [
      • Lancione M.
      • Donatelli G.
      • Cecchi P.
      • Cosottini M.
      • Tosetti M.
      • Costagli M.
      Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
      ], we aimed at having the first TE, the average TE and the train length uniform at all sites. As shown in Table 3, the first TE and the average TE were similar, though not the same, across vendors, with the largest difference being 6.4 ms between average TE for Philips and GE. Moreover, eight echoes were acquired on Siemens scanners, while seven were acquired on Philips and GE. This can lead to differences in signal-to-noise ratio of the average susceptibility map.
      Limitations to the traveling brain experiment may relate to the small sample size and to the lack of test–retest repeatability assessment on the same scanner. Moreover, scanners from only two vendors were involved. However, the consistency of the results across subjects, across the two areas A1 and A2 and the correspondence with what has previously been reported in the literature [
      • Deh K.
      • Nguyen T.D.
      • Eskreis-Winkler S.
      • Prince M.R.
      • Spincemaille P.
      • Gauthier S.
      • et al.
      Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
      ,
      • Lin P.-Y.
      • Chao T.-C.
      • Wu M.-L.
      Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
      ] support the reliability of findings. Also the second part of the study concerning the estimation of QSM variability cross-sites in a healthy population is limited by the small population acquired at each site. Moreover, only one site with an MR system from vendor V3 was involved in the study. This should be considered when comparing the within-vendor standard deviations in Table 5, as for V3 it actually represents a within-site cross-subjects variability. It should be also taken into account that the standard deviation between different sites with systems from the same vendor can be affected by different scanner models, software releases and coils (see Table 1). Subject age might be another possible confounding factor, as some brain areas are known to accumulate iron due to aging [
      • Acosta-Cabronero J.
      • Betts M.J.
      • Cardenas-Blanco A.
      • Yang S.
      • Nestor P.J.
      In Vivo MRI Mapping of brain iron deposition across the adult lifespan.
      ,
      • Betts M.J.
      • Acosta-Cabronero J.
      • Cardenas-Blanco A.
      • Nestor P.J.
      • Düzel E.
      High-resolution characterisation of the aging brain using simultaneous quantitative susceptibility mapping (QSM) and R2* measurements at 7 T.
      ]. For this reason, we enrolled young and healthy subjects in a narrow age interval, equally distributed across sites.
      This study supports the effective feasibility of the potential use of QSM in multicentric studies evaluating within-site and cross-site reproducibility and repeatability. In future works, the RIN – Neuroimaging Network will enroll populations of patients with well-known patterns of susceptibility alteration, e.g. patients with dementia [
      • Acosta-Cabronero J.
      • Williams G.B.
      • Cardenas-Blanco A.
      • Arnold R.J.
      • Lupson V.
      • Nestor P.J.
      • et al.
      In Vivo quantitative susceptibility mapping (QSM) in Alzheimer’s disease.
      ,
      • Tiepolt S.
      • Schäfer A.
      • Rullmann M.
      • Roggenhofer E.
      • Gertz H.-J.
      • Schroeter M.L.
      • et al.
      Quantitative susceptibility mapping of amyloid-β aggregates in Alzheimer’s disease with 7T MR.
      ,
      • Tuzzi E.
      • Balla D.Z.
      • Loureiro J.R.A.
      • Neumann M.
      • Laske C.
      • Pohmann R.
      • et al.
      Ultra-high field MRI in Alzheimer’s disease: effective transverse relaxation rate and quantitative susceptibility mapping of human brain in vivo and ex vivo compared to histology.
      ], Parkinson’s disease [
      • Barbosa J.H.O.
      • Santos A.C.
      • Tumas V.
      • Liu M.
      • Zheng W.
      • Haacke E.M.
      • et al.
      Quantifying brain iron deposition in patients with Parkinson’s disease using quantitative susceptibility mapping, R2 and R2*.
      ,
      • Langkammer C.
      • Pirpamer L.
      • Seiler S.
      • Deistung A.
      • Schweser F.
      • Franthal S.
      • et al.
      Quantitative susceptibility mapping in Parkinson’s disease.
      ], multiple sclerosis and white matter disease [
      • Emmerich J.
      • Bachert P.
      • Ladd M.E.
      • Straub S.
      On the separation of susceptibility sources in quantitative susceptibility mapping: theory and phantom validation with an in vivo application to multiple sclerosis lesions of different age.
      ,
      • Deh K.
      • Ponath G.D.
      • Molvi Z.
      • Parel G.-C.
      • Gillen K.M.
      • Zhang S.
      • et al.
      Magnetic susceptibility increases as diamagnetic molecules breakdown: Myelin digestion during multiple sclerosis lesion formation contributes to increase on QSM.
      ,
      • Shin H.-G.
      • Lee J.
      • Yun Y.H.
      • Yoo S.H.
      • Jang J.
      • Oh S.-H.
      • et al.
      χ-separation: magnetic susceptibility source separation toward iron and myelin mapping in the brain.
      ,
      • Wisnieff C.
      • Ramanan S.
      • Olesik J.
      • Gauthier S.
      • Wang Y.
      • Pitt D.
      Quantitative susceptibility mapping (QSM) of white matter multiple sclerosis lesions: interpreting positive susceptibility and the presence of iron.
      ], and an age-matched cohort of healthy subjects in order to compare the within-site and the cross-site performance of QSM in discriminating the two groups, so evaluating the effective feasibility and sensitivity of multicentric studies.
      In conclusion, we implemented and optimized a harmonized protocol for QSM across 3T MRI scanner platforms from three different vendors. Importantly, we leveraged product sequences and signal reconstruction algorithms provided by the vendor, to set a protocol easy to implement on clinical scanners. Despite some differences in the acquisition parameters due to slightly different sequence implementations for each vendor, we obtained excellent reproducibility in the traveling brain experiment. We provided an ROI-dependent estimation of the variability of susceptibility measurements that could be expected in a healthy population in a multicentric study and reported a precision of the order of 0.005–0.025 ppm. Hence, we suggest that multicentric multi-vendor studies using QSM at 3T are feasible and reliable in quantifying magnetic susceptibility in both cortical and subcortical gray matter regions.

      The RIN–Neuroimaging Network

      Alberto Redolfi (IRCCS Istituto Centro San Giovanni di Dio Fatebenefratelli), Egidio D'Angelo (Fondazione IRCCS Istituto Neurologico Naz.le Mondino, University of Pavia), Gianluigi Forloni (Istituto di Ricerche Farmacologiche Mario Negri IRCCS), Raffaele Agati (IRCCS Istituto delle Scienze Neurologiche di Bologna), Marco Aiello (IRCCS SDN Istituto di Ricerca), Elisa Alberici (IRCCS Istituti Clinici Scientifici Maugeri), Carmelo Amato (Oasi Research Institute-IRCCS), Filippo Arrigoni (Istituto Scientifico, IRCCS E. Medea), Francesca Baglio (IRCCS Fondazione don Carlo Gnocchi onlus), Stefano Bastianello (Fondazione IRCCS Istituto Neurologico Naz.le Mondino), Lilla Bonanno (IRCCS Centro Neurolesi Bonino Pulejo), Francesca Bottino (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Marco Bozzali (Fondazione IRCCS Santa Lucia), Chiara Carducci (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Lorenzo Carnevale (IRCCS Neuromed), Antonella Castellano (IRCCS Ospedale San Raffaele), Carlo Cavaliere (IRCCS SDN Istituto di Ricerca), Mattia Colnaghi (Istituto Auxologico Italiano IRCCS), Giorgio Conte (Fondazione IRCCS Ca’ Granda Osp. Maggiore Policlinico), Silvia De Francesco (IRCCS Istituto Centro San Giovanni di Dio Fatebenefratelli), Greta Demichelis (Fondazione IRCCS Istituto Neurologico Carlo Besta), Valeria Elisa Contarino (Fondazione IRCCS Ca’ Granda Osp. Maggiore Policlinico), Andrea Falini (IRCCS Ospedale San Raffaele), Giulio Ferrazzi (IRCCS Ospedale San Camillo), Lorenzo Figà Talamanca (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Cira Fundarò (IRCCS Istituti Clinici Scientifici Maugeri), Simona Gaudino (IRCCS Fondazione Policlinico Universitario Agostino Gemelli), Francesco Ghielmetti (Fondazione IRCCS Istituto Neurologico Carlo Besta), Ruben Gianeri (Fondazione IRCCS Istituto Neurologico Carlo Besta), Marco Grimaldi (IRCCS Istituto Clinico Humanitas), Antonella Iadanza (IRCCS Ospedale San Raffaele), Fabrizio Levrero (IRCCS Ospedale Policlinico San Martino), Raffaele Lodi (IRCCS Istituto delle Scienze Neurologiche di Bologna), Daniela Longo (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Giulia Lucignani (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Martina Lucignani (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Maria Luisa Malosio (IRCCS Istituto Clinico Humanitas), Vittorio Manzo (Istituto Auxologico Italiano, IRCCS), M. Marcella Laganà (IRCCS Fondazione don Carlo Gnocchi onlus), Silvia Marino (IRCCS Centro Neurolesi Bonino Pulejo), Jean Paul Medina (Fondazione IRCCS Istituto Neurologico Carlo Besta), Edoardo Micotti (Istituto di Ricerche Farmacologiche Mario Negri IRCCS), Claudia Morelli (Istituto Auxologico Italiano IRCCS), Alessio Moscato (IRCCS Istituti Clinici Scientifici Maugeri), Francesco Padelli (Fondazione IRCCS Istituto Neurologico Carlo Besta), Sara Palermo (Fondazione IRCCS Istituto Neurologico Carlo Besta), Patrizia Pantano (RCCS Neuromed), Chiara Parrillo (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Denis Peruzzo (Istituto Scientifico, IRCCS E. Medea), Nikolaos Petsas (IRCCS Neuromed), Letterio S. Politi (IRCCS Istituto Clinico Humanitas), Luca Roccatagliata (IRCCS Ospedale Policlinico San Martino), Elisa Rognone (Fondazione IRCCS Istituto Neurologico Naz.le Mondino), Andrea Rossi (Ospedale Pediatrico Istituto Giannina Gaslini, Università di Genova), Maria Camilla Rossi-Espagnet (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Claudia Ruvolo (IRCCS Centro Neurolesi Bonino Pulejo), Marco Salvatore (IRCCS SDN Istituto di Ricerca), Emanuela Tagliente (IRCCS Istituto Ospedale Pediatrico Bambino Gesù), Claudia Testa (IRCCS Istituto delle Scienze Neurologiche di Bologna), Caterina Tonon (IRCCS Istituto delle Scienze Neurologiche di Bologna), Domenico Tortora (Ospedale Pediatrico Istituto Giannina Gaslini), Fabio Maria Triulzi (Fondazione IRCCS Ca’ Granda Osp. Maggiore Policlinico).

      Declaration of Competing Interest

      The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

      Acknowledgements

      This project was funded by the Italian Minister of Health (RRC-2016-2361095, RRC-2017-2364915, RRC-2018-2365796, and RCR-2019-23669119_001, RCR 2020-23670067, and RCR-2022-23682285) and the Ministry of Economy and Finance (CCR-2017-23669078). This study was also partially supported by grant RC and the 5x1000 voluntary contributions to IRCCS Fondazione Stella Maris, funded by the Italian Ministry of Health. The funding sources had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, and approval of the manuscript.

      References

        • Ndayisaba A.
        • Kaindlstorfer C.
        • Wenning G.K.
        Iron in Neurodegeneration – Cause or Consequence?.
        Front Neurosci. 2019; 13https://doi.org/10.3389/fnins.2019.00180
        • Ward R.J.
        • Zucca F.A.
        • Duyn J.H.
        • Crichton R.R.
        • Zecca L.
        The role of iron in brain ageing and neurodegenerative disorders.
        Lancet Neurol. 2014; 13: 1045-1060https://doi.org/10.1016/S1474-4422(14)70117-6
        • de Rochefort L.
        • Liu T.
        • Kressler B.
        • Liu J.
        • Spincemaille P.
        • Lebon V.
        • et al.
        Quantitative susceptibility map reconstruction from MR phase data using bayesian regularization: validation and application to brain imaging.
        Magn Reson Med. 2010; 63: 194-206https://doi.org/10.1002/mrm.22187
        • Li L.
        • Leigh J.S.
        Quantifying arbitrary magnetic susceptibility distributions with MR.
        Magn Reson Med. 2004; 51: 1077-1082https://doi.org/10.1002/mrm.20054
        • Wharton S.
        • Schäfer A.
        • Bowtell R.
        Susceptibility mapping in the human brain using threshold-based k-space division.
        Magn Reson Med. 2010; 63: 1292-1304https://doi.org/10.1002/mrm.22334
        • Ravanfar P.
        • Loi S.M.
        • Syeda W.T.
        • Van Rheenen T.E.
        • Bush A.I.
        • Desmond P.
        • et al.
        Systematic review: quantitative susceptibility mapping (QSM) of brain iron profile in neurodegenerative diseases.
        Front Neurosci. 2021; 15https://doi.org/10.3389/fnins.2021.618435
        • Vinayagamani S.
        • Sheelakumari R.
        • Sabarish S.
        • Senthilvelan S.
        • Ros R.
        • Thomas B.
        • et al.
        Quantitative susceptibility mapping: technical considerations and clinical applications in neuroimaging.
        J Magn Reson Imaging. 2021; 53: 23-37https://doi.org/10.1002/jmri.27058
        • Acosta-Cabronero J.
        • Cardenas-Blanco A.
        • Betts M.J.
        • Butryn M.
        • Valdes-Herrera J.P.
        • Galazky I.
        • et al.
        The whole-brain pattern of magnetic susceptibility perturbations in Parkinson’s disease.
        Brain. 2017; 140: 118-131https://doi.org/10.1093/brain/aww278
        • Biondetti E.
        • Santin M.D.
        • Valabrègue R.
        • Mangone G.
        • Gaurav R.
        • Pyatigorskaya N.
        • et al.
        The spatiotemporal changes in dopamine, neuromelanin and iron characterizing Parkinson’s disease.
        Brain. 2021; 144https://doi.org/10.1093/brain/awab191
        • Pyatigorskaya N.
        • Sanz-Morère C.B.
        • Gaurav R.
        • Biondetti E.
        • Valabregue R.
        • Santin M.
        • et al.
        Iron imaging as a diagnostic tool for parkinson’s disease: a systematic review and meta-analysis.
        Front Neurol. 2020; 11: 366https://doi.org/10.3389/fneur.2020.00366
        • Lancione M.
        • Donatelli G.
        • Del Prete E.
        • Campese N.
        • Frosini D.
        • Cencini M.
        • et al.
        Evaluation of iron overload in nigrosome 1 via quantitative susceptibility mapping as a progression biomarker in prodromal stages of synucleinopathies.
        NeuroImage. 2022; 260: 119454https://doi.org/10.1016/j.neuroimage.2022.119454
        • Acosta-Cabronero J.
        • Machts J.
        • Schreiber S.
        • Abdulla S.
        • Kollewe K.
        • Petri S.
        • et al.
        Quantitative susceptibility MRI to detect brain iron in amyotrophic lateral sclerosis.
        Radiology. 2018; 289: 195-203https://doi.org/10.1148/radiol.2018180112
        • Costagli M.
        • Symms M.R.
        • Angeli L.
        • Kelley D.A.C.
        • Biagi L.
        • Farnetani A.
        • et al.
        Assessment of silent T1-weighted head imaging at 7 T.
        Eur Radiol. 2016; 26: 1879-1888https://doi.org/10.1007/s00330-015-3954-2
        • Wang C.
        • Foxley S.
        • Ansorge O.
        • Bangerter-Christensen S.
        • Chiew M.
        • Leonte A.
        • et al.
        Methods for quantitative susceptibility and R2* mapping in whole post-mortem brains at 7T applied to amyotrophic lateral sclerosis.
        NeuroImage. 2020; 222: 117216https://doi.org/10.1016/j.neuroimage.2020.117216
        • Acosta-Cabronero J.
        • Williams G.B.
        • Cardenas-Blanco A.
        • Arnold R.J.
        • Lupson V.
        • Nestor P.J.
        • et al.
        In Vivo quantitative susceptibility mapping (QSM) in Alzheimer’s disease.
        PLoS ONE. 2013; 8: e81093https://doi.org/10.1371/journal.pone.0081093
        • Ayton S.
        • Fazlollahi A.
        • Bourgeat P.
        • Raniga P.
        • Ng A.
        • Lim Y.Y.
        • et al.
        Cerebral quantitative susceptibility mapping predicts amyloid-β-related cognitive decline.
        Brain. 2017; 140: 2112-2119https://doi.org/10.1093/brain/awx137
        • Cogswell P.M.
        • Wiste H.J.
        • Senjem M.L.
        • Gunter J.L.
        • Weigand S.D.
        • Schwarz C.G.
        • et al.
        Associations of quantitative susceptibility mapping with Alzheimer’s disease clinical and imaging markers.
        NeuroImage. 2021; 224: 117433https://doi.org/10.1016/j.neuroimage.2020.117433
        • Kim H.-G.
        • Park S.
        • Rhee H.Y.
        • Lee K.M.
        • Ryu C.-W.
        • Rhee S.J.
        • et al.
        Quantitative susceptibility mapping to evaluate the early stage of Alzheimer’s disease.
        NeuroImage Clin. 2017; 16: 429-438https://doi.org/10.1016/j.nicl.2017.08.019
        • Li W.
        • Wu B.
        • Avram A.V.
        • Liu C.
        Magnetic susceptibility anisotropy of human brain in vivo and its molecular underpinnings.
        NeuroImage. 2012; 59: 2088-2097https://doi.org/10.1016/j.neuroimage.2011.10.038
        • Liu C.
        • Li W.
        • Johnson G.A.
        • Wu B.
        High-field (9.4T) MRI of brain dysmyelination by quantitative mapping of magnetic susceptibility.
        NeuroImage. 2011; 56: 930-938https://doi.org/10.1016/j.neuroimage.2011.02.024
        • Wharton S.
        • Bowtell R.
        Effects of white matter microstructure on phase and susceptibility maps.
        Magn Reson Med. 2015; 73: 1258-1269https://doi.org/10.1002/mrm.25189
        • Lancione M.
        • Tosetti M.
        • Donatelli G.
        • Cosottini M.
        • Costagli M.
        The impact of white matter fiber orientation in single-acquisition quantitative susceptibility mapping.
        NMR Biomed. 2017; 30: e3798https://doi.org/10.1002/nbm.3798
        • Karsa A.
        • Punwani S.
        • Shmueli K.
        The effect of low resolution and coverage on the accuracy of susceptibility mapping.
        Magn Reson Med. 2018; 81: 1833-1848https://doi.org/10.1002/mrm.27542
        • Zhou D.
        • Cho J.
        • Zhang J.
        • Spincemaille P.
        • Wang Y.
        Susceptibility underestimation in a high-susceptibility phantom: dependence on imaging resolution, magnitude contrast, and other parameters.
        Magn Reson Med. 2017; 78: 1080-1086https://doi.org/10.1002/mrm.26475
        • Elkady A.M.
        • Sun H.
        • Wilman A.H.
        Importance of extended spatial coverage for quantitative susceptibility mapping of iron-rich deep gray matter.
        Magn Reson Imaging. 2016; 34: 574-578https://doi.org/10.1016/J.MRI.2015.12.032
        • Biondetti E.
        • Karsa A.
        • Thomas D.L.
        • Shmueli K.
        Investigating the accuracy and precision of TE-dependent versus multi-echo QSM using Laplacian-based methods at 3 T.
        Magn Reson Med. 2020; 84: 3040-3053https://doi.org/10.1002/mrm.28331
        • Cronin M.J.
        • Wang N.
        • Decker K.S.
        • Wei H.
        • Zhu W.-Z.
        • Liu C.
        Exploring the origins of echo-time-dependent quantitative susceptibility mapping (QSM) measurements in healthy tissue and cerebral microbleeds.
        NeuroImage. 2017; 149: 98-113https://doi.org/10.1016/j.neuroimage.2017.01.053
        • Lancione M.
        • Cencini M.
        • Costagli M.
        • Donatelli G.
        • Tosetti M.
        • Giannini G.
        • et al.
        Diagnostic accuracy of quantitative susceptibility mapping in multiple system atrophy: the impact of echo time and the potential of histogram analysis.
        NeuroImage Clin. 2022; 34: 102989https://doi.org/10.1016/j.nicl.2022.102989
        • Lancione M.
        • Donatelli G.
        • Cecchi P.
        • Cosottini M.
        • Tosetti M.
        • Costagli M.
        Echo-time dependency of quantitative susceptibility mapping reproducibility at different magnetic field strengths.
        NeuroImage. 2019; 197: 557-564https://doi.org/10.1016/j.neuroimage.2019.05.004
        • Sood S.
        • Urriola J.
        • Reutens D.
        • O’Brien K.
        • Bollmann S.
        • Barth M.
        • et al.
        Echo time-dependent quantitative susceptibility mapping contains information on tissue properties.
        Magn Reson Med. 2017; 77: 1946-1958https://doi.org/10.1002/mrm.26281
        • Spincemaille P.
        • Anderson J.
        • Wu G.
        • Yang B.
        • Fung M.
        • Li K.e.
        • et al.
        Quantitative susceptibility mapping: MRI at 7T versus 3T.
        J Neuroimaging. 2020; 30: 65-75https://doi.org/10.1111/jon.12669
        • Emmerich J.
        • Bachert P.
        • Ladd M.E.
        • Straub S.
        On the separation of susceptibility sources in quantitative susceptibility mapping: theory and phantom validation with an in vivo application to multiple sclerosis lesions of different age.
        J Magn Reson. 2021; 330: 107033https://doi.org/10.1016/J.JMR.2021.107033
        • Hinoda T.
        • Fushimi Y.
        • Okada T.
        • Fujimoto K.
        • Liu C.
        • Yamamoto A.
        • et al.
        Quantitative susceptibility mapping at 3 T and 1.5 T.
        Invest Radiol. 2015; 50: 522-530https://doi.org/10.1097/RLI.0000000000000159
        • Voelker M.N.
        • Kraff O.
        • Goerke S.
        • Laun F.B.
        • Hanspach J.
        • Pine K.J.
        • et al.
        The traveling heads 2.0: multicenter reproducibility of quantitative imaging methods at 7 Tesla.
        NeuroImage. 2021; 232: 117910https://doi.org/10.1016/j.neuroimage.2021.117910
        • Wang R.
        • Xie G.
        • Zhai M.
        • Zhang Z.
        • Wu B.
        • Zheng D.
        • et al.
        Stability of R2* and quantitative susceptibility mapping of the brain tissue in a large scale multi-center study.
        Sci Rep. 2017; 7https://doi.org/10.1038/srep45261
        • Deh K.
        • Nguyen T.D.
        • Eskreis-Winkler S.
        • Prince M.R.
        • Spincemaille P.
        • Gauthier S.
        • et al.
        Reproducibility of quantitative susceptibility mapping in the brain at two field strengths from two vendors.
        J Magn Reson Imaging. 2015; 42: 1592-1600https://doi.org/10.1002/jmri.24943
        • Lin P.-Y.
        • Chao T.-C.
        • Wu M.-L.
        Quantitative susceptibility mapping of human brain at 3T: a multisite reproducibility study.
        Am J Neuroradiol. 2015; 36: 467-474https://doi.org/10.3174/ajnr.A4137
        • Rua C.
        • Clarke W.T.
        • Driver I.D.
        • Mougin O.
        • Morgan A.T.
        • Clare S.
        • et al.
        Multi-centre, multi-vendor reproducibility of 7T QSM and R2* in the human brain: results from the UK7T study.
        NeuroImage. 2020; 223: 117358https://doi.org/10.1016/j.neuroimage.2020.117358
        • Nigri A.
        • Ferraro S.
        • Gandini Wheeler-Kingshott C.A.
        • Tosetti M.
        • Redolfi A.
        • Forloni G.
        • et al.
        Quantitative MRI harmonization to maximize clinical impact: the RIN–Neuroimaging Network.
        Front Neurol. 2022; 13855125https://doi.org/10.3389/fneur.2022.855125
        • Schofield M.A.
        • Zhu Y.
        Fast phase unwrapping algorithm for interferometric applications.
        Opt Lett. 2003; 28: 1194-1196https://doi.org/10.1364/OL.28.001194
        • Smith S.M.
        Fast robust automated brain extraction.
        Hum Brain Mapp. 2002; 17: 143-155https://doi.org/10.1002/hbm.10062
        • Schweser F.
        • Deistung A.
        • Lehr B.W.
        • Reichenbach J.R.
        Quantitative imaging of intrinsic magnetic tissue properties using MRI signal phase: an approach to in vivo brain iron metabolism?.
        NeuroImage. 2011; 54: 2789-2807https://doi.org/10.1016/j.neuroimage.2010.10.070
        • Li W.
        • Wang N.
        • Yu F.
        • Han H.
        • Cao W.
        • Romero R.
        • et al.
        A method for estimating and removing streaking artifacts in quantitative susceptibility mapping.
        NeuroImage. 2015; 108: 111-122https://doi.org/10.1016/j.neuroimage.2014.12.043
        • Li W.
        • Wu B.
        • Liu C.
        Quantitative susceptibility mapping of human brain reflects spatial variation in tissue composition.
        NeuroImage. 2011; 55: 1645-1656https://doi.org/10.1016/j.neuroimage.2010.11.088
        • Denk C.
        • Rauscher A.
        Susceptibility weighted imaging with multiple echoes.
        J Magn Reson Imaging. 2010; 31: 185-191https://doi.org/10.1002/jmri.21995
        • Desikan R.S.
        • Ségonne F.
        • Fischl B.
        • Quinn B.T.
        • Dickerson B.C.
        • Blacker D.
        • et al.
        An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest.
        NeuroImage. 2006; 31: 968-980https://doi.org/10.1016/j.neuroimage.2006.01.021
        • Fischl B.
        • Salat D.H.
        • Busa E.
        • Albert M.
        • Dieterich M.
        • Haselgrove C.
        • et al.
        Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain.
        Neuron. 2002; 33: 341-355https://doi.org/10.1016/S0896-6273(02)00569-X
        • Avants B.B.
        • Epstein C.L.
        • Grossman M.
        • Gee J.C.
        Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain.
        Med Image Anal. 2008; 12: 26-41https://doi.org/10.1016/j.media.2007.06.004
        • Wang Z.
        • Bovik A.C.
        • Sheikh H.R.
        • Simoncelli E.P.
        Image quality assessment: from error visibility to structural similarity.
        IEEE Trans Image Process. 2004; 13: 600-612https://doi.org/10.1109/TIP.2003.819861
        • Langkammer C.
        • Schweser F.
        • Shmueli K.
        • Kames C.
        • Li X.u.
        • Guo L.i.
        • et al.
        Quantitative susceptibility mapping: report from the 2016 reconstruction challenge.
        Magn Reson Med. 2018; 79: 1661-1673https://doi.org/10.1002/mrm.26830
        • Costagli M.
        • Donatelli G.
        • Cecchi P.
        • Bosco P.
        • Migaleddu G.
        • Siciliano G.
        • et al.
        Distribution indices of magnetic susceptibility values in the primary motor cortex enable to classify patients with amyotrophic lateral sclerosis.
        Brain Sci. 2022; 12: 942https://doi.org/10.3390/brainsci12070942
        • Howard C.M.
        • Jain S.
        • Cook A.D.
        • Packard L.E.
        • Mullin H.A.
        • Chen N.
        • et al.
        Cortical iron mediates age-related decline in fluid cognition.
        Hum Brain Mapp. 2022; 43: 1047-1060https://doi.org/10.1002/hbm.25706
        • Monti S.
        • Pontillo G.
        • Russo C.
        • Cella L.
        • Cocozza S.
        • Palma G.
        RESUMEN: a flexible class of multi-parameter qMRI protocols.
        Phys Medica Eur J Med Phys. 2021; 88: 23-36https://doi.org/10.1016/j.ejmp.2021.04.005
        • Ward P.G.D.
        • Harding I.H.
        • Close T.G.
        • Corben L.A.
        • Delatycki M.B.
        • Storey E.
        • et al.
        Longitudinal evaluation of iron concentration and atrophy in the dentate nuclei in friedreich ataxia.
        Mov Disord. 2019; 34: 335-343https://doi.org/10.1002/mds.27606
        • Bilgic B.
        • Langkammer C.
        • Marques J.P.
        • Meineke J.
        • Milovic C.
        • Schweser F.
        QSM reconstruction challenge 2.0: Design and report of results.
        Magn Reson Med. 2021; 86: 1241-1255https://doi.org/10.1002/mrm.28754
        • Cheng Y.C.N.
        • Neelavalli J.
        • Haacke E.M.
        Limitations of calculating field distributions and magnetic susceptibilities in MRI using a Fourier based method.
        Phys Med Biol. 2009; 54: 1169-1189https://doi.org/10.1088/0031-9155/54/5/005
        • Koch K.M.
        • Papademetris X.
        • Rothman D.L.
        • De Graaf R.A.
        Rapid calculations of susceptibility-induced magnetostatic field perturbations for in vivo magnetic resonance.
        Phys Med Biol. 2006; 51: 6381-6402https://doi.org/10.1088/0031-9155/51/24/007
        • Feng X.
        • Deistung A.
        • Reichenbach J.R.
        Quantitative susceptibility mapping (QSM) and R2* in the human brain at 3 T: evaluation of intra-scanner repeatability.
        Z Für Med Phys. 2018; 28: 36-48https://doi.org/10.1016/j.zemedi.2017.05.003
        • Straub S.
        • Schneider T.M.
        • Emmerich J.
        • Freitag M.T.
        • Ziener C.H.
        • Schlemmer H.-P.
        • et al.
        Suitable reference tissues for quantitative susceptibility mapping of the brain.
        Magn Reson Med. 2017; 78: 204-214https://doi.org/10.1002/mrm.26369
        • Acosta-Cabronero J.
        • Betts M.J.
        • Cardenas-Blanco A.
        • Yang S.
        • Nestor P.J.
        In Vivo MRI Mapping of brain iron deposition across the adult lifespan.
        Neurobiol Dis. 2016; 36: 364-374https://doi.org/10.1523/JNEUROSCI.1907-15.2016
        • Betts M.J.
        • Acosta-Cabronero J.
        • Cardenas-Blanco A.
        • Nestor P.J.
        • Düzel E.
        High-resolution characterisation of the aging brain using simultaneous quantitative susceptibility mapping (QSM) and R2* measurements at 7 T.
        NeuroImage. 2016; 138: 43-63https://doi.org/10.1016/j.neuroimage.2016.05.024
        • Tiepolt S.
        • Schäfer A.
        • Rullmann M.
        • Roggenhofer E.
        • Gertz H.-J.
        • Schroeter M.L.
        • et al.
        Quantitative susceptibility mapping of amyloid-β aggregates in Alzheimer’s disease with 7T MR.
        J Alzheimers Dis. 2018; 64: 393-404https://doi.org/10.3233/JAD-180118
        • Tuzzi E.
        • Balla D.Z.
        • Loureiro J.R.A.
        • Neumann M.
        • Laske C.
        • Pohmann R.
        • et al.
        Ultra-high field MRI in Alzheimer’s disease: effective transverse relaxation rate and quantitative susceptibility mapping of human brain in vivo and ex vivo compared to histology.
        J Alzheimers Dis JAD. 2020; 73: 1481-1499https://doi.org/10.3233/JAD-190424
        • Barbosa J.H.O.
        • Santos A.C.
        • Tumas V.
        • Liu M.
        • Zheng W.
        • Haacke E.M.
        • et al.
        Quantifying brain iron deposition in patients with Parkinson’s disease using quantitative susceptibility mapping, R2 and R2*.
        Magn Reson Imaging. 2015; 33: 559-565https://doi.org/10.1016/j.mri.2015.02.021
        • Langkammer C.
        • Pirpamer L.
        • Seiler S.
        • Deistung A.
        • Schweser F.
        • Franthal S.
        • et al.
        Quantitative susceptibility mapping in Parkinson’s disease.
        PLoS ONE. 2016; 11: e0162460https://doi.org/10.1371/journal.pone.0162460
        • Deh K.
        • Ponath G.D.
        • Molvi Z.
        • Parel G.-C.
        • Gillen K.M.
        • Zhang S.
        • et al.
        Magnetic susceptibility increases as diamagnetic molecules breakdown: Myelin digestion during multiple sclerosis lesion formation contributes to increase on QSM.
        J Magn Reson Imaging. 2018; 48: 1281-1287https://doi.org/10.1002/jmri.25997
        • Shin H.-G.
        • Lee J.
        • Yun Y.H.
        • Yoo S.H.
        • Jang J.
        • Oh S.-H.
        • et al.
        χ-separation: magnetic susceptibility source separation toward iron and myelin mapping in the brain.
        NeuroImage. 2021; 240: 118371https://doi.org/10.1016/j.neuroimage.2021.118371
        • Wisnieff C.
        • Ramanan S.
        • Olesik J.
        • Gauthier S.
        • Wang Y.
        • Pitt D.
        Quantitative susceptibility mapping (QSM) of white matter multiple sclerosis lesions: interpreting positive susceptibility and the presence of iron.
        Magn Reson Med. 2015; 74: 564-570https://doi.org/10.1002/mrm.25420