A comparison of phase imaging and quantitative susceptibility mapping in the imaging of multiple sclerosis lesions at ultrahigh field

Magn Reson Mater Phy (2016) 29:543–557 DOI 10.1007/s10334-016-0560-5 RESEARCH ARTICLE A comparison of phase imaging and quantitative susceptibility mapping in the imaging of multiple sclerosis lesions at ultrahigh field Matthew John Cronin1,2 · Samuel Wharton2 · Ali Al‑Radaideh2,4 · Cris Constantinescu3 · Nikos Evangelou3 · Richard Bowtell2 · Penny Anne Gowland2 Received: 2 December 2015 / Revised: 9 February 2016 / Accepted: 26 February 2016 / Published online: 25 April 2016 © The Author(s) 2016. This article is published with open access at Springerlink.com Abstract Objective The aim of this study was to compare the use of high-resolution phase and QSM images acquired at ultra-high field in the investigation of multiple sclerosis (MS) lesions with peripheral rings, and to discuss their usefulness for drawing inferences about underlying tissue composition. Materials and methods Thirty-nine Subjects were scanned at 7 T, using 3D T2*-weighted and T1-weighted sequences. Phase images were then unwrapped and filtered, and quantitative susceptibility maps were generated using a thresholded k-space division method. Lesions were compared visually and using a 1D profiling algorithm. Results Lesions displaying peripheral rings in the phase images were identified in 10 of the 39 subjects. Dipolar projections were apparent in the phase images outside of the extent of several of these lesions; however, QSM images showed peripheral rings without such projections. These projections appeared ring-like in a small number of phase images where no ring was observed in QSM. 1D profiles of six well-isolated example lesions showed that QSM * Penny Anne Gowland 1 Brain Imaging and Analysis Centre, Duke University, Durham, NC 27710, USA 2 Sir Peter Mansfield Imaging Centre, School of Physics and Astronomy, University of Nottingham, University Park, Nottingham NG7 2RD, UK 3 Sir Peter Mansfield Imaging Centre, Queens Medical Centre, University of Nottingham, Nottingham NG7 2RD, UK 4 Department of Medical Imaging, Faculty of Allied Health Sciences, Hashemite University, Zarqa, Jordan contrast corresponds more closely to the magnitude images than phase contrast. Conclusions Phase images contain dipolar projections, which confounds their use in the investigation of tissue composition in MS lesions. Quantitative susceptibility maps correct these projections, providing insight into the composition of MS lesions showing peripheral rings. Keywords Magnetic resonance imaging · Multiple sclerosis · White matter · Iron · Myelin Introduction T2*-weighted gradient echo sequences are often used to study multiple sclerosis (MS) due to the high signal to noise ratio (SNR) and good contrast [1–10] that they provide. With the rising availability of high (3 T) and ultrahigh field (7 T and higher) MRI systems, the phase data associated with the T2*-weighted magnitude images are increasingly being used both as an adjunct to conventional magnitude images [10–13], or in combination with them to produce susceptibility-weighted (SWI) images [7, 14, 15], as they provide a complimentary contrast mechanism. In recent years, phase and SWI images have been used to study the variation in iron levels in different anatomical brain regions, mainly deep brain nuclei, with age and gender [13, 16], as well as changes in iron levels in MS [3, 7, 10]. However, the use of phase contrast as a qualitative or quantitative measure of iron content assumes a direct relationship between signal phase and local iron levels. This assumption is flawed since a change in magnetic susceptibility, such as a local increase in iron concentration, produces a change in the magnetic field (and hence phase) that is not localized to the susceptibility perturbation, but 13 544 instead is dipolar in nature, causing both positive and negative field/phase perturbations in the surrounding region [14]. This non-local relationship can lead to incorrect inferences being drawn about local iron levels based on phase or SWI images. The non-local nature of the phase contrast can be overcome by quantitative susceptibility mapping (QSM) [17–19], which produces maps of the local variations in the tissue susceptibility that are responsible for the measured phase changes. The QSM technique has, amongst other applications, been used in vivo to measure changes in the magnetic susceptibility of the basal ganglia in MS patients [1, 2]. While this technique offers a local contrast directly linked to the physical property of magnetic susceptibility, it relies on accurate measurement of the local field perturbations originating only within the imaging volume. For this reason, care must still be taken in the choice of filtering algorithms and parameters for processing of the phase data. The white matter lesions occurring in MS are sometimes surrounded by rings on T2*-weighted magnitude and phase images. It has been suggested that these rings may be a marker of local changes in iron content [5, 7, 10]. However, this MR signature could also result from a local variation of the myelin density, or even from changes in tissue microstructure [20], and so the origin of peripheral rings remains a matter of some debate. The ability to detect iron changes around lesions would be useful in understanding the pathogenesis of MS lesions and in tracking disease progression. In previous work, phase and SWI images have been used to investigate the prevalence and nature of peripheral rings in MS lesions [3, 7, 10]. The aim of the work described here was to compare the depiction of white matter MS lesions in phase and QSM data, taking advantage of the high resolution (0.5 mm, isotropic) achievable in vivo at 7 T. The prevalence of peripheral rings was measured across a cohort of 39 MS patients in order to compare the effectiveness of phase and QSM images in such identification. More detailed analysis was applied to a subset of six lesions with peripheral rings in order to establish the sources of the contrast in each image type, including comparison of the effects of SHARP and high-pass (SWI) phase filtering algorithms on phase and QSM contrast. Simulated field maps were generated for models of a solid and shell-like susceptibility distribution based on one lesion in order to illustrate the field patterns that such structures produce. Magn Reson Mater Phy (2016) 29:543–557 system equipped with a 16 or 32 channel receiver head coil. Of the 39 subjects, 19 were diagnosed with clinically isolated syndrome (CIS) suggestive of MS (at 4-year follow-up, 14 were diagnosed with clinically definite RRMS), eight were diagnosed with relapsing remitting MS (RRMS), nine were diagnosed with primary progressive MS, and three were diagnosed with secondary progressive MS. Data for susceptibility mapping were acquired using a multi-stack, spoiled, interleaved, 3D T2*-weighted FLASH sequence. Each image was acquired with four stacks, overlapping by 10 voxels at each interface, and 0.5 mm isotrop (...truncated)