Seizure-onset zone localization by statistical parametric mapping in visually normal (18) F-FDG PET studies

Area of Science:

• Neuroimaging
• Epilepsy Research
• Medical Diagnostics

Background:

• Neuroimaging is vital for presurgical evaluation in medically refractory epilepsy.
• Standard (18)F-FDG-PET has moderate sensitivity for localizing the seizure-onset zone (SOZ).
• Visual assessment of (18)F-FDG-PET can be normal despite underlying pathology.

Purpose of the Study:

• To assess the utility of statistical parametric mapping (SPM) for SOZ localization.
• To improve the diagnostic yield of (18)F-FDG-PET in epilepsy patients with visually normal scans.
• To evaluate SPM's effectiveness in identifying the seizure focus when conventional methods are insufficient.

Main Methods:

• Retrospective analysis of 55 epilepsy patients with visually normal (18)F-FDG-PET scans.
• Application of SPM8 with various statistical thresholds (p < 0.05 corrected; p < 0.0001, p < 0.001, p < 0.005 uncorrected) and cluster sizes (k > 0, k > 20, k > 100, k > 200).
• Comparison of SPM-identified seizure zones (SZ) with the clinically determined SOZ.

Main Results:

• The most liberal SPM threshold (p < 0.005/k > 200) correctly localized the SOZ in 40% (22/55) of cases.
• Substantial concordance (κ = 0.607) was observed in patients who underwent surgery, comparable to MRI findings (κ = 0.783).
• SPM analysis significantly enhanced SOZ localization in visually negative (18)F-FDG-PET studies.

Conclusions:

• SPM analysis offers a valuable method to improve SOZ localization in (18)F-FDG-PET studies that are visually normal.
• Thresholding statistical parametric maps with liberal p-values and restrictive cluster sizes is recommended for optimal results.
• This approach can aid in identifying the epileptogenic zone, improving presurgical planning for refractory epilepsy.