Synopsis

Keywords: White Matter, Quantitative Imaging, T1 mapping

Whole brain T1 mapping has been performed using several different sequences but remains mostly confined to research applications. A clinically reliable and fast method to evaluate the white matter (WM) is important for accurate diagnosis, evaluating the severity of disease and monitoring disease progression over time. In this study we assessed Multiplied, Added, Subtracted and/or Divided Inversion Recovery (MASDIR) technique in nine healthy subjects and found evidence that such technique greatly improves the visualisation of tissue of interest and allows calculation of T1 values in a reliable, fast and clinically usable manner.

Introduction

T1 is a basic tissue property that provides quantitative information about tissue biochemistry and microstructure, able to describe several pathological changes^[1]. White matter hyperintensities (WMHs) represent a variety of pathophysiologic processes such as demyelination, axonal loss or increase in interstitial fluid, visualised as areas of increased signal intensity on both T2 weighted (T2w) and Fluid Attenuated Inversion Recovery (T2-FLAIR) images. WMHs are important biomarkers of small vessel disease (SVD) but unfortunately, their heterogeneity and dynamic nature is not easily discernible on conventional MR images. Also, subtle lesions in the so-called Normal Appearing White Matter (NAWM) are not yet readily visible due to lack of sufficient contrast^[2]. Hence, a clinically reliable and fast method to evaluate the White Matter (WM) is important to make a more accurate diagnosis, evaluate the true extent of disease and monitor lesions over time. Our aim was to evaluate the feasibility and possibility of implementing Multiplied, Added, Subtracted and/or Divided Inversion Recovery (MASDIR) technique^[3] to calculate T1 map and to enhance T1 contrast in the WM of the whole brain by using Fast Spin Echo - Inversion Recovery (IR-FSE) sequences readily available on all scanners.

Methods

Data were acquired on 9 healthy subjects (from 26 to 59 yrs, 2M/7F) using a Philips 3T scanner and a 32 channel head coil. In addition to standard clinical acquisitions (3D T1w, 3D T2w, 3D FLAIR), six whole brain IR-FSE sequences (TR/TE=3500/10-200 ms; voxel size= 0.7x0.89 mm, slice thickness 5 mm) were acquired with increasing Inversion Times (TI= 350, 450, 550, 750, 950, 1200 ms). divided Substracted Inversion Recovery (dSIR: one of the MASDIR sequences) images were derived using the following equation: $$dSIR=\frac{IRTI_{short}-IRTI_{long}}{IRTI_{short}+IRTI_{long}}$$ where TI_short and TI_long are the two TIs that define the middle Domain (mD), i.e. the range of T1 values of interest ($$$TI_{short/long}=TI_{min/max}\cdot\ln(2)$$$) (Fig. 1). Since we focused on the WM tissue, we set the TI_short to suppress the WM and the TI_long to suppress the Grey Matter (GM). We also calculated a T1 map using dSIR (T1_dSIR) by linearising the relationship between the dSIR signal (S_dSIR) and T1 values within the mD: $$T1=\frac{TI_{long}-TI_{short}}{\ln(4)}\cdot S_{dSIR} + \frac{TI_{long}+TI_{short}}{\ln(4)}$$To validate T1 values obtained using dSIR, we compared them with T1 values obtained by signal regression from all IR-FSE images using Bloch equation (T1reg, gold standard). We further compared the T1 estimates in GM and WM regions derived using FreeSurfer and a JHU like atlas, respectively. We also show images of the normal brain and dSIR imaging in a patient with SVD.

Results

Table 1 presents average T1 values of GM, WM and cerebrospinal fluid (CSF, divided into ventricular, CSFvent, and subarachnoid space, CSFsas) with respective nulling TI values. Fig. 1 compares conventional MR images with three IR-FSE images (TIs=450, 750, 950 ms). A higher contrast and lesion load is observed on dSIR maps with respect to T2-w and FLAIR images in SVD (Fig.2). T1map obtained from dSIR method correlated very well with T1 values obtained using T1reg (m=1.04;p-value<0.001 two-tailed t test) (Fig. 3). However, some brain regions segmented using FreeSurfer, that lie outside the chosen mD cannot be evaluated using dSIR methodology (mD: 450-750 ms corresponds to T1: 649-1082 ms) (Fig. 4).

Discussion

The greatest strength of dSIR is to enhance contrast between structures characterised by small but different T1 values without complex post-processing. In addition, dSIR provided a highly reliable quantitative T1 estimate albeit restricted to brain areas that lie inside the mD. In our patient with SVD, dSIR greatly enhanced the qualitative visualisation of disease and detected more lesions with longer T1 in the NAWM. In addition, T1 quantification obtained is of significant value in monitoring the dynamic nature of such lesions. Such information can be key in characterising WMH. Quantification of T1 values is of added value when compared to several other T1 mapping methods described in literature, many of which are unavailable on most clinical scanners world-wide. The main limitation of this technique is that dSIR cannot evaluate tissues that lie outside the mD. However, the mD can be chosen to study brain areas of interest including the cortex as well as the CSF. Also, the nulling TIs used in adults may be different with respect to the developing brain and depend on the degree of myelin maturation.

Conclusion

dSIR is a reliable method for obtaining very high lesion T1 contrast and can be directly applied in routine clinical practice to provide qualitative and quantitative T1 maps. It can also be useful in the study of neurofluids to analyse changes in ISF and protein content. Future studies require improved spatial resolution to allow evaluation of lesions that are of small dimensions.

Acknowledgements

We thank John D. Port for interpretation of our images.