Study protocol: value of 7-T MRI with prospective motion correction and postprocessing for patients with nonlesional epilepsy

Detection of FCDs type 1 and 2 is generally more difficult than other types of epileptogenic lesions. The advantage of 7 T for the detection of FCDs can mainly be explained by improved gray–white delineation for the cortical convexity in 3D T1-weighted images. Colon et al. reported a comparison between 7‑T and 3‑T MRI for eight patients with suspected FCD, showing that 7‑T MRI scored significantly better for lesion conspicuity and demarcation. To differentiate between types of FCD, typical radiological characteristics of FCD were rated separately in this study. Significant characteristics were features such as gray–white matter blurring, abnormal internal structure, and transition to normal cortex [4].

In SWI sequences, visualization of intracortical signal changes (“black line sign”) can improve subtyping of FCD type 2 [2]. Three-dimensional T1-weighted (MPRAGE and MP2RAGE) sequences were shown to be most helpful for the detection of FCD when used with quantitative morphological analysis due to high image contrast at 7 T [7, 39, 44].

A study of ten patients with polymicrogyria previously diagnosed with 3‑T MRI demonstrated improved visualization with 7‑T MRI [6]. Special diagnostic landmarks were dilated superficial veins associated with the polymicrogyria revealed in SWI angiography. Furthermore, 3D T1-weighted sequences (MP2RAGE or MPRAGE) are important because they enable clear delineation of the lesion extent, which can guide surgical resection. Three-dimensional T1-weighted sequences can be used to screen the whole brain for polymicrogyria. In addition, 3D T2*-weighted images enable visualization of small pial vessels, seen as thin hypointense lines in the malformed cortex and sulci with an arboriform distribution as an additional identifying feature; the cortex itself appears extra hyperintense in these sequences.

It was shown that better SNR in 7‑T MRI and increased spatial resolution in T1-weigted (MPRAGE and MP2RAGE) and T2*-weighted sequences improve detection of cerebral lesions in tuberous sclerosis complex such as cortical and subependymal tubers. Moreover, a new finding first identified at 7 T is the presence of tortuous veins associated with subependymal tubers, frequently encountered but clearly visible in SWI sequences [27, 37].

Gangliogliomas and dysembryoplastic neuroepithelial tumors consist of a composition of mature neuronal cells and glial cells [35]. Radiological characteristics include a solid and/or cystic component and perifocal edema. With 7‑T MRI, 3D T1-weighted (MP2RAGE or MPRAGE) images better delineate the solid component because of increased image contrast. In addition, 7‑T 3D T2-weighted sequences meliorate the depiction of the walls between and around the solid/cystic components, and the extent of any associated edema is more precisely delineated. Both factors are important when planning the resection margin for surgical intervention.

Classic MRI features of hippocampal sclerosis are hippocampal atrophy, increased T2-weighted/FLAIR signal intensity, and loss of normal morphology. 7‑T MRI data show hippocampal morphology, including internal structure and surface features. Two-dimensional coronal TSE T2-weighted and 3D T1-weighted/FLAIR sequences are particularly suitable for this [33, 48]. Hippocampal subfields can be more precisely delineated with training based on landmarks and surface features at 7‑T MRI, including automated segmentation methods. On coronal 3‑T images, prominent infolding can cause the dark band to appear obscured because of partial volume effects, and high-resolution images at 7 T help to avoid this pitfall. The absence of digitations along the hippocampal head is another sensitive and specific finding for hippocampal sclerosis that is considerably more apparent on 7‑T images.

Furthermore, 7‑T MRI and automated subfield volumetry have enabled detection of hippocampal pathology also in subfield volumes of the CA1, CA2/3, CA4/DG, and the subiculum [11]. Specifically, among patients with unilateral mesial temporal lobe epilepsy (mTLE) with longer disease durations, volume loss was observed in the ipsilateral CA1 and CA2/3 subfields and contralateral CA1. There were no differences in subfield volumes in patients with neocortical epilepsy compared to controls [11].

In the task force consensus recommendations on the use of 7‑T MRI in clinical practice for epilepsy, the eight8 most useful sequences were identified in a survey from 19 7‑T MRI centers experienced in examining patients with epilepsy for research and/or diagnostic purposes. Nevertheless, there is no uniform acquisition protocol among clinical centers, especially concerning those studies focused on morphometric analysis [23].

A number of other indirect signs for epilepsy-associated changes were described in several 7‑T MRI pilot studies with different postprocessing methods (Table 2).

There are only few studies with relative homogeneous patient groups and concordant parallel analysis of 3‑T and 7‑T MRI data [38, 41, 44]. In the study by Wang et al. [44], the morphometric analysis of 3‑T MPRAGE sequences was compared with the same analysis on the 7‑T MP2RAGE sequence. The general diagnostic yield of 7‑T MRI with morphometric analysis was 43% (29 of 67 patients). The study was designed to best utilize 7‑T sequences for subtle lesion detection in a clinical setting, rather than to compare the two sequences at different field strengths. However, MP2RAGE sequences are of special importance in the visualization of MCD not only in 7‑T images but also in 3‑T images. In MP2RAGE sequences, the inhomogeneity effect can be largely canceled out by combining image data from the first and second readouts; T2* and B1 inhomogeneity effects can be largely canceled out, resulting in a strongly T1-weighted image with superior gray matter to white matter contrast than available with MPRAGE sequences. As noted by Demerath et al. [8], the study used only MP2RAGE imaging in their 7‑T MRI, so that the high detection rate was possibly due to this sequence and not due to the higher field strength.

To date, there have been no prospective clinical studies comparing MP2RAGE sequences in 3‑T and 7‑T MRI in parallel with the same acquisition protocol for morphometric analysis.

The main technical challenges for data acquisition under UHF in 7‑T MRI are to produce a strong, homogeneous transverse field and high-quality images. Different approaches are used to minimize the influence of nonuniformities caused by interferences at UHFs and to accelerate and increase the yield regarding the time spent for acquisition and the quality of the image achieved. Motion artifacts represent a substantial limiting factor for very high-resolution imaging, since they reduce the resolution and isotropy of voxel-based imaging and therefore influence the results of analyses and calculations, such as segmentation or gray matter volume and thickness estimates [13]. Motion correction techniques can be used to correct for motion either in real time (prospectively) or offline after the data have been collected (retrospectively). Spin history effects caused by through-plane motion are not corrected by retrospective correction. In contrast to retrospective techniques, prospective motion correction (PMC) ensures that the k‑space sampling density stays approximately homogeneous [13].

One of the PMC techniques with an external tracking device was suggested in 2015 by Stucht and colleagues [36]. This technology allows for high-precision tracking of out-of-plane rotations, by deriving pose information from changes in a moiré pattern visible on a 15-mm marker. Moreover, through an external optical tracking system and image processing, the image resolution in 7‑T MRI can be substantially increased and therefore is suitable for visualization of very fine structures such as cortical layers [13, 19].

Prospective, multi-center study of presurgically assessed patients with previous 3‑T MRI and a diagnosis of focal epilepsy regarded as nonlesional and with a clear seizure-onset hypothesis that meets the following two criteria:

Or:

Patients with temporal and extratemporal drug-resistant epilepsy will be included. According to electroclinical and semiological analysis, a selection of patients will be made oriented toward a search for “expected” malformations of cortical development.

The 3‑T MRI for the identification of the nonlesional patients should match basic standards of the use of structural MRI in the care of patients with epilepsy [3, 47].

The exclusion criteria for investigation of patients in a 7‑T MRI scanner are according to the 7‑T MRI information sheet (supplemental information), i.e., metal objects in their body, patients with claustrophobia or other psychiatric conditions, those who cannot stay in an appropriate position/keep still during the investigation, patients under the age of 14 years, and obese patients with body weight over 140 kg.

All patients will be examined in a 3‑T MRI scanner (Siemens Magnetom Prisma, Siemens Medical Solutions, Erlangen, Germany) and in a 7‑T MRI scanner (Siemens Magnetom Terra). The acquisition protocol for 3‑T MRI measurements (Table 3) will be designed according to Demerah et al. [7] and the 7‑T MRI acquisition will be as close as practicable to the 3‑T acquisition protocol.

For systematic reasons, implementing artifact reduction in 7‑T image acquisition asks for a direct comparison of this technique with the 3‑T imaging. However, the current literature supports the impact of movement artifacts on SNR during examination in 3‑T MRI to be insignificant. Additionally, its correction burdens the examination technically [43], and thus we decided for practical reasons to omit this point; especially since studies of 7‑T MRI at the same time have shown that it often plays a key role in the quality of the images obtained. The PMC techniques for the 7‑T study will be used according to Stucht and colleagues [36] and the morphometric analysis according to Huppertz [16, 17]. A dataset of a control group from the same 7‑T MRI scanner and with the same investigation will be used as a normal database for the morphometric analysis. The control group data for comparison of MP2RAGE measurements of 3‑T MR images were obtained with an MRI scanner of the same model (Siemens Magnetom Prisma) with an identical measurement protocol that will be used in the clinical group of probands. In each case, results will be reviewed via expert visual inspection by two neuroradiologist/epileptologists, with expertise in epilepsy imaging, who are blinded to the seizure-onset zone . The control dataset will be integrated and used for the SPM12-based MAP18 Toolbox analysis. For MAP analysis, z‑scores and probability maps will be used according to the well-known method of Huppertz et al. [17].

Our study protocol is designed to make a larger study after the current small pilot study possible. It places emphasis on the feasibility of the use of 7‑T imaging in a clinical study. Since all patients will undergo either subsequent invasive presurgical evaluation or a surgical intervention, if indicated, we will ask for pathology results and outcomes as they are of exceptional clinical importance. A retrospective analysis of the current small study with these patients and outcome parameters is planned.