Outcome of temporal lobe epilepsy surgery predicted by statistical parametric PET imaging.

UNLABELLED

PET is useful in the presurgical evaluation of temporal lobe epilepsy. The purpose of this retrospective study is to assess the clinical use of statistical parametric imaging in predicting surgical outcome.

METHODS

Interictal 18FDG-PET scans in 17 patients with surgically-treated temporal lobe epilepsy (Group A-13 seizure-free, group B = 4 not seizure-free at 6 mo) were transformed into statistical parametric imaging, with each pixel representing a z-score value by using the mean and s.d. of count distribution in each individual patient, for both visual and quantitative analysis.

RESULTS

Mean z-scores were significantly more negative in anterolateral (AL) and mesial (M) regions on the operated side than the nonoperated side in group A (AL: p < 0.00005, M: p = 0.0097), but not in group B (AL: p = 0.46, M: p = 0.08). Statistical parametric imaging correctly lateralized 16 out of 17 patients. Only the AL region, however, was significant in predicting surgical outcome (F = 29.03, p < 0.00005). Using a cut-off z-score value of -1.5, statistical parametric imaging correctly classified 92% of temporal lobes from group A and 88% of those from Group B.

CONCLUSION

The preliminary results indicate that statistical parametric imaging provides both clinically useful information for lateralization in temporal lobe epilepsy and a reliable predictive indicator of clinical outcome following surgical treatment.