High-Field Magnetic Resonance Imaging for Epilepsy




Epileptogenic lesions are often subtle, do not change during life, and are easily overlooked, if spatial resolution and signal to noise ratio are inappropriate. 2D or more recently 3D-FLAIR sequences are best suited to detect small cortical dysplasias which are often located at the bottom of a sulcus. 3D-T1-weighted gradient echo sequences are used for multiplanar, curved surface reformations, and voxel-based analyses. 3 T MR imaging is currently the state-of-the-art imaging modality for patients with suspected structural epilepsies in which an epileptogenic lesion has not yet been found.








  • High-field magnetic resonance (MR) imaging is currently the state-of-the-art technique for the detection and characterization of epileptogenic lesions.



  • An increased signal-to-noise ratio helps to increase the spatial resolution while maintaining a high contrast between normal cortex and epileptogenic lesion.



  • Subtle epileptogenic lesions are more frequently overlooked than invisible on high-quality MR imaging.



Key Points


Diagnostic Checklist
The primary goal of MR imaging in epilepsy patients is to detect an epileptogenic lesion, defined as a radiographic lesion that causes seizures. These lesions are often subtle, do not change during the life span, and may even be missed by careful visual inspection.




  • Do not start MR imaging without information about:



  • Seizure type



  • Seizure frequency (single vs multiple seizures, seizures in context of an acute disease)



  • Semiology of the seizure



  • Suspected etiology (genetic/idiopathic, structural/metabolic, unknown)






  • Check if MR images show:



  • Sufficient gray/white matter contrast



  • Appropriate spatial resolution



  • Symmetric visualization of unaffected anatomic structures for allowing side comparisons






  • Check if MR images show:



  • Sufficient gray/white matter contrast



  • Appropriate spatial resolution



  • Symmetric visualization of unaffected anatomic structures for allowing side comparisons






  • Interpret MR imaging



  • In a clinical context as described above



  • Incorporating electroencephalography findings if possible




Theoretical considerations


Signal-to-Noise Ratio


According to the increasing number of parallel spins at higher field strengths (as derived from the Boltzmann equation), the signal increases in proportion to the field strength B 0 (signal to noise ∼ B 0 ). The signal theoretically doubles from 1.5 to 3 T, and in practice it increases by a factor of around 1.8. However, this higher signal-to-noise ratio (SNR) can be used to increase the CNR and the spatial resolution, or to decrease the acquisition time ( Fig. 1 ).




Fig. 1


A circumscribed temporobasal cortical contusion ( arrow ) is better visible on a T2-weighted fast spin sequence with higher spatial resolution ( B , D : 0.47 × 0.64 × 2 mm vs A , C : 0.93 × 0.93 × 2 mm).


Relaxation Times


Relaxation times are a function of the applied magnetic field strength. With increasing field strength, spin-lattice or longitudinal relaxation time T1 increases by 20% to 40% for most tissue. At the same time, spin-spin or transversal relaxation time T2 decreases. T1 and T2 relaxation times of biological fluids such as water (ie, cerebrospinal fluid) and blood are constant between 1.5 T and 3 T; the strong prolongation of T1 of solid tissue (ie, brain) at 3 T leads to an overall lower tissue contrast on T1-weighted images.



Recommended basic MR protocol acquired at 3 T (based on Philips Intera)






















































































































































Sequence Type
3D T1 FFE FLAIR TSE T2 TSE FLAIR TSE T2 TSE FLAIR TSE
Orientation Sagittal Sagittal Axial Coronal Coronal Axial
FOV 256 240 230 230 240 256
RFOV 0.95 0.9 0.8 0.8 0.9 1
Matrix 256 256 512 256 512 256
Scan% 100 72.6 80 70.6 80 100
TI 833 2850 2850 2850
TR/TE 8.2 12000 3272 12000 5765 12000
TE 3.7 120 80 140 120 140
Flip angle 8 140 90 90 90 90
R-factor No No No 3 No
Slice thickness 1 3.5 5 3 2 2
Interslice gap 0 0 1 0 0 0
No. of slices 140 40 24 40 40 60
No. of excitations 1 1 1 1 6 1
Acquired voxel size (mm) 1 × 1 × 1 0.98 × 1.26 × 3.5 0.57 × 0.72 × 5 0.9 × 1.27 × 3 0.47 × 0.64 × 2 1 × 1 × 2
Reconstructed voxel size (mm) 1 × 1 × 1 0.49 × 0.49 × 3.5 0.45 × 0.45 × 5 0.45 × 0.45 × 3 0.23 × 0.23 × 2 1 × 1 × 2
Acquisition time 3 min 11 s 4 min 48 s 1 min 58 s 4 min 00 s 4 min 53 s 5 min 24 s

Abbreviations: 3D, 3-dimensional; FFE, fast field echo; FLAIR, fluid-attenuated inversion recovery; FOV, field of view; R-factor, reduction factor; RFOV, rectangular field of view; T1, T1-weighted; T2, T2-weighted; TE, echo time; TI, inversion time; TR, repetition time; TSE, turbo spin echo.

Special sequences are added to the protocol based on imaging findings or clinical hints: if there is an epileptogenic lesion other than hippocampal sclerosis, nonenhanced and contrast-enhanced spin-echo sequences are added. Contrast-medium injections are needed to characterize a lesion but not to find it.

If lesions contain areas with signal loss on T2-weighted images suggesting hemosiderin deposits or calcifications or if patients present with a trauma anamnesis, T2-weighted gradient echo sequences or, as of more recently, susceptibility-weighted sequences (SWI) are added.

3 T scanners allowing the generation of 3-dimensional (3D) fluid-attenuated inversion recovery (FLAIR) sequences with isotropic voxels depict the whole brain with high spatial resolution and are, in addition, usable for multiplanar reformations and voxel-based analyses. An elegant reformation is the planar surface (pancake) view, which facilitates anatomic orientation and is helpful in determining the boundaries of epileptogenic lesions. This view is generated by defining a path along the brain surface on coronal reformations, and constructing a planar curved surface view enabled by parallel shifting in an anterior and posterior direction (see Figs. 7 and 8 later in this article).

Diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI) reversible splenium lesions are helpful in characterizing so-called reversible splenium lesions, which occur more often if antiepileptic drugs are reduced or withdrawn in order to provoke epileptic seizures during presurgical workup.


Larmor Frequency


According to the relationship between the Larmor precession frequency and the magnetic field strength (ω = γB 0 ), the Larmor frequency increases from 63.9 MHz at 1.5 T to 127.8 MHz at 3 T.


Chemical Shift, Susceptibility, and B 1 Homogeneity


Chemical shift increases proportionally with the field strength, that is, from 220 Hz at 1.5 T to 440 Hz at 3 T. The shorter echo time (TE) at 3 T potentially decreases the sensitivity to pulsation and motion artifacts. Susceptibility, defined as the extent to which a material becomes magnetized when placed within a magnetic field, scales in proportion to the field strength, which can be both an advantage ( Fig. 2 ) and disadvantage. Susceptibility can affect local tissue resonance frequencies (spin-spin relaxation) and lead to dephasing and signal loss typically in gradient echo pulse sequences. The relaxation time T2∗ that describes the susceptibility-induced signal loss is shortened at 3 T compared with 1.5 T.



Additional MR sequences for 3 T (based on Philips Achieva)






















































































































































Sequence
T1 TSE T2 FFE DWI DTI SWI 3D FLAIR
Orientation Coronal Axial Axial Axial Axial Sagittal
FOV 230 230 256 256 220 250
RFOV 0.8 0.8 1 1 0.8 100
Matrix 256 256 128 128 256 228
Scan% 79.9 79.9 97.8 98.4 100 100
TI 1600
TR 550 601 3151 11374 16 4800
TE 13 18 69 63 23 309
Flip angle 90 18 90 90 10 90
R-factor No No 3 AP 2.2 AP 1.5 RL 2.5 AP, 2 RL
Slice thickness 5 5 5 2 1 1.1
Interslice gap 1 1 1 0 0 0
No. of slices 24 24 24 60 200 327
No. of excitations 1 1 2 1 1 2
Acquired voxel size (mm) 0.9 × 1.12 × 5 0.9 × 1.12 × 5 2 × 2.4 × 5 2 × 2.03 × 2 1 × 1 × 1 1.1 × 1.1 × 1.1
Reconstructed voxel size (mm) 0.45 × 0.45 × 5 0.45 × 0.45 × 5 1 × 1 × 5 2 × 2 × 2 0.43 × 0.43 × 0.5 0.43 × 0.43 × 0.55
Acquisition time 4 min 33 s 1 min 41 s 1 min 09 s 6 min 26 s 3 min 17 s 4 min 43 s

Abbreviations: AP, anterior-posterior; RL, right-left.



Fig. 2


A cavernoma in the left hippocampal head was overlooked on a T2-weighted fast spin echo sequence at 1.5 T ( A , C : arrow ). Because of higher susceptibility, it is clearly visible on the same sequence at 3 T ( B , D : arrow ).


Radiofrequency Power Deposition


Radiofrequency (RF) energy deposition is monitored by measuring the specific absorption rate (SAR), which must not exceed 4 W/kg over a 15-minute period. For comparison, RF energy deposition of most mobile phones is in the range of 0.5 to 0.75 W/kg. Because RF energy deposition scales with the square of B 0 in the range from 1.5 T to 3 T, SAR limits will be reached much earlier, limiting especially fast pulse sequences with high RF energy deposition, such as fast spin-echo sequences.


Adaption of Imaging Protocols


SAR limitations, different relaxation times, and increased susceptibility are the most important parameters to be adapted at 3 T. Parallel imaging reduces RF energy deposition by reducing the number of phase-encoding readouts (determined by the reduction factor R) at a given TE, or it allows for the reduction of echo-train length in, for example, echo planar imaging (EPI), yielding a shorter effective TE. This approach allows a substantial reduction in image distortion and improves image quality. In addition, the shorter TE is more motion resistant and reduces blurring in the image. Moreover, as the application of high R-factors are limited by the parallel imaging inherent signal loss (S ∼ 1/√R), the abundant SNR at 3 T can be used to implement R-factors of greater than 2.


Another way to reduce SAR is to acquire 3D sequences and sequences with variable refocusing flip angle. However, 2-dimensional (2D) FLAIR sequences with in-plane resolutions of less than 1 mm and slice thicknesses between 2 and 5 mm still have a higher CNR than 3D FLAIR sequences with isotropic, typically 1-mm 3 voxels, which are useful for the generation of multiplanar reformations and voxel-based analyses.


With respect to T1-weighted spin echo images, gray matter/white matter contrast is reduced at 3 T for several reasons. T1 relaxation times of gray and white matter lengthen and converge. Shielding effects induced by eddy currents prevent central parts of the image from being properly excited, which results in reduced signal intensity of the basal ganglia region. Stronger magnetization transfer effects reduce signal intensity and contrast. T1-weighted gradient echo images provide a good contrast between gray and white matter; however, contrast enhancement is different and may remain undetected on gradient echo images.


A recent survey about the distribution of epileptogenic lesions operated on in a large epilepsy surgery center shows hippocampal sclerosis (HS) as the most common and also almost constant lesion over time. The number of long-term epilepsy-associated tumors (LEATs) and of patients negative on MR imaging has decreased, whereas the number of focal cortical dysplasias (FCDs) has steadily increased. These main groups of epileptogenic lesions are now described in detail:




Hippocampal sclerosis


Background


HS is by far the most common cause of temporal lobe epilepsy and is found in approximately half of patients undergoing resective surgery. Resective surgery, either by anterior two-thirds temporal lobe resection or selective amgydalohippocampectomy, results in complete seizure relief in at least 58% of patients. Conversely, with medical therapy only 8% of patients achieve freedom from seizures.


Imaging


On MR imaging, the sclerotic hippocampus is atrophic and has an increased signal intensity on FLAIR and T2-weighted fast spin-echo images. This pattern is best visualized on thin (2–3 mm thick) coronal images perpendicular to the longitudinal axis of the hippocampus. CNR is higher on FLAIR than on T2-weighted images ( Figs. 3 and 4 ); however, already normal limbic structures have an increased FLAIR signal intensity compared with the remaining cortex. High-resolution T2-weighted fast spin-echo images at 3 T visualize hippocampal substructures precisely (see Figs. 3 and 4 ) and are therefore suited to detect mild forms of HS, which, however, account for only around 5% of epilepsy surgery series. Associated lesions such as amygdala atrophy, entorhinal cortex atrophy, atrophy of ipsilateral corpus mamillare, atrophy of ipsilateral fornix, and gray/white matter demarcation loss of the anterior temporal lobe are also better visualized on 3 T than on 1.5 T MR imaging.


Mar 20, 2017 | Posted by in NEUROLOGICAL IMAGING | Comments Off on High-Field Magnetic Resonance Imaging for Epilepsy

Full access? Get Clinical Tree

Get Clinical Tree app for offline access