Multimodality Image Post-processing in Detection of Extratemporal MRI-Negative Cortical Dysplasia

Abstract

Purpose: To determine the diagnostic value of individual image post-processing techniques in a series of patients who underwent extratemporal operations for histologically proven, MRI-negative focal cortical dysplasia (FCD). Methods: The morphometric analysis program (MAP), PET/MRI co-registration and statistical parametric mapping (SPM) analysis of PET (SPM-PET) techniques were analyzed in 33 consecutive patients. The epileptogenic zone (EZ) assumed by MAP, PET/MRI, and SPM-PET was compared with the location of the FCD lesions determined by stereoelectroencephalography (SEEG) and histopathological study. The detection rate of each modality was statistically compared. Results: Three lesions were simultaneously detected by the three post-processing methods, while two lesions were only MAP positive, and 8 were only PET/MRI positive. The detection rate of MAP, PET/MRI, SPM-PET and the combination of the three modalities was 24.2, 90.9, 57.6, and 97.0%, respectively. Taking the pathological subtype into account, no type I lesions were detected by MAP, and PET/MRI was the most sensitive method for detecting FCD types II and IIA. During a mean follow-up period of 22.94 months, seizure freedom was attained in 26/33 patients (78.8%) after focal corticectomy. Conclusions: MAP, PET/MRI, and SPM-PET provide complementary information for FCD detection, intracranial electrode design, and lesion resection. PET/MRI was particularly useful, with the highest detection rate of extratemporal MRI-negative FCD.

Keywords: MRI negative; PET/MRI co-registration; SPM-PET; focal cortical dysplasia; morphometric analysis program.

Figure 1

Flow diagram of patient selection. RFC, radiofrequency coagulation.

Figure 2

Flow diagram of detection process. The second row refers to the primary analysis results by post-processing methods. The third row refers to the presumed EZ when clinicians read the results with other evaluation information and the last row are the results validated by SEEG or histopathology.

Figure 3

Statistical data of MAP, PET/MRI, and SPM-PET in detecting FCD. (A) Number of patients with FCD lesions detected by PET visual analysis, MAP, PET/MRI, and SPM-PET and their concordance. (B) Detection rates of MAP, PET/MRI, SPM-PET, and the combination of those methods for all FCD types. (C) Detection rates of MAP, PET/MRI, and SPM-PET for FCD types I and II. (D) Detection rates of MAP, PET/MRI, and SPM-PET for FCD types IIA and IIB. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 4

Image post-processing examples. Patients with FCD lesions detected by all methods (A), only MAP (B), PET/MRI, and SPM-PET (C).

Figure 5

Patient 21 with FCD tended to be mislocalized by conventional visual analysis of PET images. The semiology of the 13-year-old girl was nocturnal dystonic posturing of the left limbs, numbness in the left arm with ambiguous localization before rare diurnal seizures. Scalp EEG showed periodic sharp-slow wave complexes on channel C3-P3 and T3-T5 before seizure onset (A). MRI scans were negative. Focal hypometabolism in the left parietal lobe was identified by conventional visual analysis (B), which was nonconcordant with the ambiguous localization of the numbness and epileptic waves on channel T3-T5. PET/MRI and SPM-PET demonstrated mild hypometabolism in the central sulcus of the left insula (C). Intracranial electrodes were implanted for coverage of the left insular and central areas (D). Interictal SEEG recording showed that continuous, repetitive spikes arose from the left PSG and ALG and spread to the central area (E). Ictal data showing seizures (10 onsets during SEEG monitoring) originating from the left PSG (F). The left PSG, ALG, and PLG were removed (G); the pathological finding was FCD IIA, and the patient was seizure-free through the last follow-up at 21 months. ALG, anterior long gyrus; PLG, posterior long gyrus; PostCG, postcentral gyrus; PreCG, precentral gyrus; PSG, posterior short gyrus.

Figure 6

Subtle hypometabolic change tended to be overlooked by conventional visual analysis. A 12-year-old boy (patient 32) with nocturnal seizures manifesting as aura (fear/fluster), tachycardia, fearful expression, dystonic posturing of the right limbs, and hypermotor activity. Ictal scalp EEG demonstrated sharp waves in the left frontal lobe (A). Visual analysis identified focal PET hypometabolism in the left mesial frontal region (white arrow). PET/MRI detected a very subtle hypometabolic change (black arrow) in the ipsilateral superior frontal sulcus (B). Five electrodes were implanted in the left frontal area including the two hypometabolic sulci (C). Asynchronous spikes arose from the two sulci during the interictal period (D), and seizures (14 onsets during SEEG monitoring) originated from the left superior frontal sulcus (E). The two sulci were removed, and both were FCD IIB (F). The patient remained seizure-free through the last follow-up at 18 months. ACC, anterior cingulate cortex; OFC, orbitofrontal cortex; PreCS, precentral sulcus.