Reduction of susceptibility-induced signal losses in multi-gradient-echo images: Application to improved visualization of the subthalamic nucleus

Abstract

T2⁎-weighted gradient echo (GE) images yield good contrast of iron-rich structures like the subthalamic nuclei due to microscopic susceptibility induced field gradients, providing landmarks for the exact placement of deep brain stimulation electrodes in Parkinson's disease treatment. An additional advantage is the low radio frequency (RF) exposure of GE sequences. However, T2⁎-weighted images are also sensitive to macroscopic field inhomogeneities, resulting in signal losses, in particular in orbitofrontal and temporal brain areas, limiting anatomical information from these areas. In this work, an image correction method for multi-echo GE data based on evaluation of phase information for field gradient mapping is presented and tested in vivo on a 3 Tesla whole body MR scanner. In a first step, theoretical signal losses are calculated from the gradient maps and a pixelwise image intensity correction is performed. In a second step, intensity corrected images acquired at different echo times TE are combined using optimized weighting factors: in areas not affected by macroscopic field inhomogeneities, data acquired at long TE are weighted more strongly to achieve the contrast required. For large field gradients, data acquired at short TE are favored to avoid signal losses. When compared to the original data sets acquired at different TE and the respective intensity corrected data sets, the resulting combined data sets feature reduced signal losses in areas with major field gradients, while intensity profiles and a contrast-to-noise (CNR) analysis between subthalamic nucleus, red nucleus and the surrounding white matter demonstrate good contrast in deep brain areas.

Introduction

Deep brain stimulation plays an important role in the treatment of Parkinson's Disease (PD). Thus, there is a strong clinical interest in magnetic resonance (MR) imaging techniques that allow the localization of the subthalamic nuclei (STN) and the globus pallidus internus (GPi) as landmarks for the stereotactic implantation of stimulation electrodes (Dormont et al., 2004, Starr et al., 2002). In general, T2-weighted MR techniques are employed for this purpose since the resulting images are hypointense in iron-rich structures like the STN (Benabid et al., 2002, Kosta et al., 2006) and provide a good anatomical contrast. A common procedure is to fuse T2-weighted images with images of other contrast or imaging modalities (e.g. with T1-weighted data sets), which show relatively poor contrast in the STN but provide other important anatomical or pathological information and a higher spatial resolution. The contrast in T2-weighted images is even better for PD patients because T2 values are significantly reduced in the STN, the putamen, and the substantia nigra zona compacta due to increased tissue iron content (Kosta et al., 2006). In general, the contrast is improved with increasing field strength. As an example, a better visualisation of the STN at 3 T compared to 1.5 T has been demonstrated (Slavin et al., 2006).

The effective transverse relaxation time T2⁎ comprises the reversible (time constant T2′) and irreversible (time constant T2) components of transverse relaxation according to 1/T2⁎ = 1/T2 + 1/T2′. In general, the signal decay is accelerated in the presence of inhomogeneities of the static magnetic field. Field inhomogeneities can be classified into three categories (Fernandez-Seara and Wehrli, 2000) according to their spatial scale. Microscopic field inhomogeneities (smaller than voxel size and diffusion length) contribute to the irreversible component of transverse relaxation (T2). Mesoscopic field inhomogeneities (smaller than voxel size, but larger than diffusion length) are caused by susceptibility differences between tissues and contribute to the reversible component (T2′) and thus to T2⁎. They are of special interest, since they provide information about tissue microstructure. Thus, T2⁎-weighted images are more sensitive to tissue iron induced microscopic and mesoscopic susceptibility gradients (Bonny et al., 2001, Bourekas et al., 1999, Elolf et al., 2007), displaying pronounced image hypointensity in the nuclei when using gradient echo (GE) sequences due to local T2⁎ reduction. An additional advantage of T2⁎-weighted GE sequences over T2-weighted turbo spin echo (TSE) sequences is the lower specific absorption rate (SAR) of radiofrequency (RF) power due to the absence of refocusing pulses, which is an important issue especially at higher field strengths (Elolf et al., 2007). As in the case of T2-weighted images, the contrast is improved with increasing field strength, although the physical dimensions of the nuclei may be exaggerated because of susceptibility effects (Bourekas et al., 1999). The problem is that T2⁎-weighted images are also sensitive to macroscopic field inhomogeneities (larger than voxel size), in particular near air-filled cavities, e.g. in orbitofrontal and temporal brain areas (Abduljalil and Robitaille, 1999, Bourekas et al., 1999, Elolf et al., 2007). This results in signal losses in the affected brain areas which reduce the anatomical information content. To overcome this problem, a multi-echo GE readout has been proposed (Elolf et al., 2007), yielding several inherently co-localized T2⁎ weighted data sets with increasing TE values. Images acquired at the shortest TE do not suffer from major signal losses and can thus be used in stereotactic procedures, whereas images acquired at longer TE provide additional contrast of iron-rich structures like the STN.

In this paper we present an alternative technique based on a multi-echo FLASH (Frahm et al., 1986) sequence for acquisition of a series of GE images with different TE, similar to the method proposed by Elolf et al. However, images are directly combined to yield a single data set providing both a good STN contrast and reduced signal losses in critical brain areas. The post-processing algorithm is based on the simultaneous evaluation of modulus and phase information. In a first step, local macroscopic field inhomogeneities are calculated from the phase data and concomitant signal losses are corrected pixelwise. In a second step, images acquired with different TE are combined by introducing weighting factors that favor long TEs for improved T2⁎-contrast in brain areas free from major macroscopic field gradients (as for example deep brain areas), and short TEs in areas affected by field inhomogeneities to minimize signal losses. The resulting images do not suffer from major signal losses and provide excellent contrast in the midbrain: due to the pixelwise optimization of weighting factors the contrast-to-noise ratio (CNR) between the STN and surrounding tissue in the combined data set is increased by a factor of at least 2.4 as compared to the CNR in any of the original images. Furthermore, the algorithm is simple, allowing for its integration in the scanner's online image post-processing toolbox. As a consequence, the combined data set is exported immediately after acquisition and can be directly used to obtain landmarks for planning subsequent scans.