Purpose: Detection of hypometabolic areas on interictal FDG-PET images for assessing the epileptogenic zone is hampered by partial volume effects. We evaluated the performance of an anatomy-based maximum a-posteriori (A-MAP) reconstruction algorithm which combined noise suppression with correction for the partial volume effect in the detection of hypometabolic areas in patients with focal cortical dysplasia (FCD).
Methods: FDG-PET images from 14 patients with refractory partial epilepsy were reconstructed using A-MAP and maximum likelihood (ML) reconstruction. In all patients, presurgical evaluation showed that FCD represented the epileptic lesion. Correspondence between the FCD location and regional metabolism on a predefined atlas was evaluated. An asymmetry index of FCD to normal cortex was calculated.
Results: Hypometabolism at the FCD location was detected in 9/14 patients (64%) using ML and in 10/14 patients (71%) using A-MAP reconstruction. Hypometabolic areas outside the FCD location were detected in 12/14 patients (86%) using ML and in 11/14 patients (79%) using A-MAP reconstruction. The asymmetry index was higher using A-MAP reconstruction (0.61, ML 0.49, p=0.03).
Conclusion: The A-MAP reconstruction algorithm improved visual detection of epileptic FCD on brain FDG-PET images compared to ML reconstruction, due to higher contrast and better delineation of the lesion. This improvement failed to reach significance in our small sample. Hypometabolism outside the lesion is often present, consistent with the observation that the functional deficit zone tends to be larger than the epileptogenic zone.