Parallel transmit 7T MRI for adult epilepsy pre-surgical evaluation

Abstract

Objective: To implement parallel transmit (pTx) 7T magnetic resonance imaging (MRI) in the pre-surgical evaluation of 3T-negative patients with drug-resistant focal epilepsy, and to compare quality to conventional single transmit (specifically, circularly polarized [CP]) 7T MRI.

Methods: We implemented a comparative protocol comprising both pTx and CP 7T MRI in consecutive adult candidates for epilepsy surgery who had negative or equivocal 3T MRI imaging. Here we report the outcomes from the first 31 patients. We acquired pTx and CP T₁, T₂, fluid-attenuated inversion recovery (FLAIR) and edge-enhancing gradient echo (EDGE) images, all in the same three-dimensional (3D) 0.8 mm isotropic space. Two-dimensional (2D) high-resolution T₂ and T₂*-weighted sequences were acquired only in CP mode due to current technological limitations. Two neuroradiologists, a neurologist, and a neurosurgeon made independent, blinded quality and preference ratings of pTx vs CP images. Quantitative methods were used to assess signal dropout.

Results: 7T revealed previously-unseen structural lesions in nine patients (29%), confirmed 3T-equivocal lesions in four patients (13%), and disproved 3T-equivocal lesions in four patients (13%). Lesions were better visualized on pTx than CP in 57% of cases, and never better visualized on CP. Clinical management was altered by 7T in 18 cases (58%). Nine cases were offered surgical resection and one laser interstitial thermal therapy (LITT). Three cases were removed from the surgical pathway because of bilateral or extensive lesions. Five cases were offered stereo-electroencephalography (sEEG) with better targeting (in three because the 7T lesion was deemed equivocal by the multi-disciplinary team (MDT), and in two because the lesion was extensive). Blinded comparison confirmed significantly better overall quality of pTx FLAIR images (F(2, 184) = 13.7, p = 2.88 × 10^-6), whereas pTx MP2RAGE images were subjectively non-inferior and had improved temporal lobe coverage with quantitatively less signal drop-out.

Significance: pTx-7T is implementable in a clinical pathway, changed management in 58% of patients where 3T + FDG-PET had not enabled resection, and is superior to single transmit 7T MRI.

Keywords: 7T MRI; epilepsy surgery; focal epilepsy; parallel transmit.

FIGURE 1

New lesions detected in the first 31 patients, deemed concordant and of high confidence by the MDT. Columns show the sequences. The identified lesions in Patient 5 (FCD), Patient 9 (FCD), Patient 11 (amygdala enlargement), Patient 12 (FCD), Patient 14 (hippocampal sclerosis), Patient 23 (FCD), Patient 28 (dysplasia or low grade glial/glioneural lesion), Patient 29 (FCD), Patient 31 (right temporal encephalocele).

FIGURE 2

Putative lesions requiring further investigation before a surgical plan could be made. The putative lesions in: Patient 4 cortical–subcortical blurring of uncertain significance, stereo‐electroencephalography (sEEG) planned to evaluate significance; Patient 8 likely amygdala lesion, sEEG planned to evaluate significance; Patient 10 extensive polymicrogyria, sEEG planned to narrow down the epileptogenic zone; Patient 15 likely FCD, ictal SPECT requested for confirmation; Patient 17 rows a–d show four consecutive slices revealing that an equivocal lesion on 3T magnetic resonance imaging (MRI) is in fact a vessel, whereas row e shows a new possible FCD awaiting sEEG evaluation; Patient 24: Gray matter thinning likely representing perinatal infarction; sEEG planned to narrow down the epileptogenic zone.

FIGURE 3

Patient 11 amygdala enlargement, visible on pTx but not CP images, due to improved uniformity of FLAIR sequence giving better definition of margins, and better clarity of EDGE contrast showing distension of lateral border.

FIGURE 4

Further lesions better visualized with pTx than CP acquisitions. Patient 9: Focal cortical dysplasia much more prominently displayed and crisply defined, especially in the EDGE contrast. Patient 29: Focal cortical dysplasia not immediately evident on CP images, but clearly demonstrated with pTx. Patient 31: Inferior temporal encephalocele falling in an area of high signal dropout and low signal‐to‐noise ratio in CP FLAIR. Although the encephalocele can be seen with both acquisition methods it is impossible to assess the integrity of surrounding cortex in the CP images. MDT discussions often focus on whether an encephalocele may be associated with a more widespread cortical lesion and a more extensive resection may be warranted in addition to the skull‐base repair. This could be conducted with much greater confidence based on the pTx images compared to the CP images.

FIGURE 5

Left: Histograms of pixel intensity for the pTx and CP INV2 sequences. Right: Deviation of these distributions from a standard normal distribution, quantified by the Wasserstein distance—much higher for the CP than pTx sequences. CDF, cumulative distribution function.