When a standard MRI does not show a clear lesion, presurgical evaluation becomes a problem of integrating several imperfect signals. Modern epilepsy programmes increasingly combine structural imaging, metabolism, perfusion, electrophysiology and functional mapping rather than looking for one decisive test.
MRI-negative does not mean lesion-free
Some epileptogenic abnormalities are below routine MRI resolution, subtle enough to be missed or visible only with specialist sequences and post-processing.
Repeat epilepsy-protocol MRI and expert neuroradiology review can reveal lesions that were previously overlooked.
PET adds metabolic information
Interictal FDG-PET can identify regional hypometabolism associated with focal epilepsy.
A 2024 systematic review and meta-analysis found that localising PET abnormalities are associated with postsurgical outcome, although PET cannot define the epileptogenic zone by itself.
SPECT adds perfusion information
Ictal and interictal SPECT can show seizure-related perfusion change, especially when injection occurs early in the seizure.
Subtraction and MRI coregistration can improve localisation.
MEG and EEG source imaging add electrophysiological localisation
MEG and high-density EEG source imaging estimate the cortical generators of interictal discharges.
Concordance with MRI, PET or semiology can strengthen a presurgical hypothesis.
Functional MRI and tractography protect function
fMRI can help map language and motor networks, while tractography estimates critical white-matter pathways.
These techniques answer the safety question—what should be preserved—as well as the localisation question.
SEEG tests the integrated hypothesis
When non-invasive data remain discordant or incomplete, SEEG can sample a targeted set of regions based on the combined hypothesis.
SEEG is not a fishing expedition; its value depends on the quality of the pre-implantation reasoning.
Image integration platforms
Modern surgical planning can coregister MRI, PET, SPECT, vascular imaging, source localisation, fMRI and electrode locations into a common patient-specific space.
This helps teams see where multiple independent data streams converge—or disagree.
Artificial intelligence is emerging, not definitive
AI-assisted lesion detection and outcome prediction are active research areas.
They may improve sensitivity and reduce interpretation burden, but current tools require careful external validation and should not replace multidisciplinary judgement.
Related reading
Sources and further reading
- Multimodal imaging in epilepsy surgery for personalised planning — 2025 review
- Localising FDG-PET hypometabolism and epilepsy surgery outcome — 2024 meta-analysis
- Epilepsy surgery in the last 10 years — 2025 review
Evidence checked: September 2026.