The objective of the IIIrd International MagneticResonance and Epilepsy symposium was to discuss ourpresent knowledge of magnetic resonance (MR) tech-niques and their application to the study of human epi-lepsy. This symposium focused on three essential topics:the enhancement of imaging methods to improve lesiondetection in patients with epilepsy, the application ofmetabolic imaging using MR spectroscopy, and finally,investigation of functional techniques such as functionalMRI (fMRI) combined with EEG and MR diffusion-weighted imaging (DWI) and perfusion-weighted imag-ing techniques to examine functional changes for thedetection of abnormalities in patients with epilepsy.As in any other medical field, knowledge of epilepsy,its diagnosis, and its treatment has developed in succes-sive stages based on clinical observation and intuitioncoupled with technologic advances (1). The preeminenceof EEG as the essential laboratory investigation in thestudy of epilepsy has now been diminished by the intro-duction of modern neuroimaging techniques, particularlyMRI (2–4). This has become even more apparent in thelast few years with the addition of further noninvasiveneuroimaging techniques, such as positron emission to-mography (PET), magnetic resonance spectroscopy(MRS), and magnetic source imaging (MSI), which havepermitted the detection of previously unsuspected pathol-ogy in a large number of patients with epilepsy (2,5–9).Furthermore, these modern neuroimaging techniques havefundamentally influenced the surgical treatment of epilepsybecause the need for invasive electrophysiologic techniqueshas decreased. The role of the lesion in epileptogenesis hasnow been reevaluated and is now well recognized.The importance of MRI is well known to all thosetreating patients with epilepsy. In 1997, the InternationalLeague Against Epilepsy Neuroimaging Commissionpublished the recommendations for neuroimaging of pa-tients with epilepsy (10). Among the recommendationsof the Commission is the one that all patients in whomepilepsy develops should have, in the ideal situation, ahigh-quality MRI. Application of this principle to allpatients with epilepsy, except for those with clear-cutidiopathic generalized epileptic syndromes, should never becompromised. It is clear that the rationale for neuroimagingin epilepsy is identification of pathology that can then dic-tate specific treatment strategies and also identification ofspecific etiologic substrates that can aid in the identificationand classification of specific epileptic syndromes (11,12).MRI can also predict the nature of structural pathology andmay be able to detect the presence of abnormalities distantfrom the putative epileptogenic region (13). This has led toelaboration of the concept of pseudolocalization, and therealization that seizures may appear to originate in oneregion while a structural abnormality may be present inanother region. Finally, there is sufficient evidence thatMRI is the most important predictor for successful surgicaltreatment of temporal lobe epilepsy (14,15).Although current MR techniques are able to detectpathology in many more patients with epilepsy than waspossible a few years ago, a large number of patients withepilepsy have normal MR studies. How can we improvethe detection capability of MR in epilepsy? Enhancedimaging can be achieved with techniques for improvingspatial resolution, image contrast, or both. These goalscan be achieved using phased-array surface coils, imageaveraging, or high-field-strength MRI (16). For example,surface coil imaging can add value to the examination inl50% of patients with normal MRIs in selected patientswith well-localized electroclinical syndromes. Similarly,image averaging can be used to improve the signal-to-noise ratio (SNR) of an MRI by performing several MRIstudies and spatially co-registering and then combiningthe images into a single study (17). Finally, high-fieldMR (3–8 T) can provide high SNR with the potential forimproved imaging. Experimental results show an ap-proximately linear increase in SNR with increasing fieldstrength. The expected increase in SNR with small voxelsize can provide improved imaging details. These tech-niques and examples of their potential applications aredescribed under high-resolution imaging.More than a decade ago, Jack et al. (18,19) introducedto the epilepsy community the use of hippocampal volu-metry for the quantitative determination of hippocampalatrophy in temporal lobe epilepsy. It is now clear thatalthough quantitative volumetric analysis of medial tem-
Read full abstract