Epilepsy affects approximately 150,000 people in the U.S. (Epilepsy Foundation). 15%-30% of epilepsy patients are refractory or incompletely responsive to pharmacotherapy [1,2]. Ultrahigh field MRI scanners, such as those operating at 7 Tesla (7T), facilitate the visualization of the brain with unprecedented resolution and contrast. This has enabled the identification and characterization of lesions not detectable at lower field strengths. Epilepsy-related brain activity may affect the size and structure of these atypical findings, which suggests that even MRI findings that are not directly related to the epileptogenic focus may be non-invasive biomarkers which can help localize the seizure onset zone (SOZ) or elucidate the etiology of the disease. Perivascular spaces (PVS) are cerebrospinal fluid-filled spaces surrounding the vasculature in the white matter. These spaces have been reported in previous work but with uncertain significance [3,4]. Due to the increased resolution enabled at 7T, PVSs are detected with increasing frequency, both in healthy volunteers and in epilepsy patients. We investigated the symmetry in the distribution of PVSs in the brains of non-lesional epilepsy patients as compared to healthy controls. Additionally, we compared the laterality of the asymmetrical PVS clusters to that of the SOZ in patients where the SOZ was suspected to be unilateral.

We measured PVSs in 7T images obtained on 10 epilepsy patients (age: 32.4± 6.4 years) with a previously normal clinical MRI exam and 10 normal healthy subjects (age: 34.7±6.1 years). Axial T₂ TSE images (TR 6000 ms, TE 69 ms, voxel 0.4x0.4x2.0mm³) were obtained using a Siemens 7T whole body MRI scanner (Siemens, Erlangen) on all patients and controls (Figure 1). Although all abnormalities detected at 7T were recorded, we are focusing on the detected PVSs for this study. Perivascular spaces were noted and measured manually on Osirix (Pixmeo, Geneva) and the location of prominent PVSs of diameter (d) > 0.5 mm were manually marked and recorded. 16 common anatomical landmarks were identified in each brain and used to divide both the right and left hemisphere of the brain into 7 regions (Figure 2). The asymmetry index (AI), weighted by the area of perivascular spaces in the right and left of each region, was calculated using Eq. 1.

$$$AI= \frac{|S_r-S_l|}{\frac{1}{2}(S_r+S_l)}(1)$$$

where S_l and S_r are the sum of the perivascular space area on the left and right hemisphere of each region (respectively) calculated using Eq 2.

$$$S_j = \sum\limits_{i=1}^{N_{region}} d_i^2(2)$$$

N_region is total number of PVSs in each region and d_i is the diameter of each individual PVS. AI_max, the largest AI between each of the 7 right-left pairs of regions, was calculated for each subject. A Student’s t-test was performed to compare the AI_max in epilepsy subjects to healthy controls. 9 out of the 10 epilepsy patients with a suspected unilateral SOZ (as determined by semiology and EEG) for further analysis were selected for further analysis. Finally, the hemisphere with the larger area of total PVSs for the brain region with greatest asymmetry was compared to the hemisphere of the suspected SOZ in the epilepsy subjects.

Prominent perivascular spaces appeared frequently in both epilepsy patients (n = 677-3884) and healthy controls (n = 832 - 3310) with no significant difference in the total number of spaces between the population groups. However, there was a significant difference (p = 0.012) between the AI_max in epilepsy patients (mean AI_max ± stderr = 1.0±0.15), and the AI_max in controls (mean AI_max ± stderr = 0.67±0.10). Figure 3 indicates that the AI was elevated with respect to healthy controls in at least 1 of the 7 regions analyzed. Table 1 compares the hemisphere in which the largest total area of PVSs exist in the region containing AImax to the hemisphere of the suspected SOZ, as determined by EEG and semiology. In 7 out of these 9 subjects there was a prominence of PVSs contralateral to the suspected SOZ with a high AI_max (mean AI_max = 1.07 ± 0.44). In 2 out of 9 subjects, the most prominent PVSs were ipsilateral to the suspected SOZ. In these two cases AI_max was low (AI_max = 0.48 and 0.54), similar to the AI_max values seen in controls.

These findings suggest that epilepsy may result in an asymmetrical distribution of PVS in the brains of patients, with more or larger PVSs clustered contralateral to the hemisphere of the suspected SOZ. Future work includes performing this analysis in a larger group of patients and controls and using automated methods of detection for PVSs.