SUMMARYPurpose: We assessed lateralization of interictal epileptiform discharges (IEDs) in children with intractable epilepsy secondary to tuberous sclerosis complexes (TSCs) during rapid eye movement sleep (R), compared with non-rapid eye movement sleep (NR) and wakefulness (W), to determine epileptogenicity of R-IEDs. Methods: We retrospectively studied 23 children with TSC, who underwent prolonged scalp video-electroencephalography (EEG) and magnetic resonance imaging (MRI). We determined the lateralization of ictal EEG, clinical semiology, and the largest tuber on MRI. We analyzed a minimum of 100 IEDs during R, NR, and W to classify right/left/generalized spikes to compare the lateralization with ictal EEG, clinical semiology, and MRI.

We read with great interest the paper by Liu et al. entitled, "Disrupted intrinsic connectivity of the periaqueductal gray in patients with functional dyspepsia: a resting-state fMRI study. 1 " The authors reported altered functional connectivity of the periaqueductal gray (PAG) in patients with functional dyspepsia (FD), which is proposed to be the basis for aberrant nociception and emotion processing in patients with FD.However, we are concerned about the authors' approach to the processing of imaging data. Global signal regression (GSR) was not performed before assessing functional connectivity (FC). Correction of global signals is a widely adopted preprocessing step of fMRI data to mitigate noise perturbation and prevent a positive bias of the voxel-based FC analysis, such as yielding higher connectivity along the interhemispheric fissure. 2 However, GSR use is highly debatable, 3-5 particularly because of the production of artifactual negative correlation.The authors regarded the global signals as nuisance signals and eliminated these in previous functional connectivity studies. 6,7 However, in the referred study, they applied a different procedure for nuisance signal regression without explaining the reason for the change. Thus, we were led to question the existence of established guidelines for identifying GSR in resting state fMRI data analysis in the study. Recently, a new method to adjust for GSR, which used a temporal down-weighting process in data analysis, revealed a good approximation of the inherently existent anticorrelations and not simply the usual artifacts of GSR. 8 This may provide an alternative method to weigh the benefits of GSR with less artifactual information.In addition, we are concerned about the authors' interpretations of their findings. The authors indicated that PAG activity with the insula may strengthen the discomfit in patients with FD.Nevertheless, the figures supporting the authors' statement were not consistent with their descriptions. The slope of the correlation between symptom severity and PAG functional connectivity with the insula ( Figure 3B) 1 seems much positive than that attributable to a value r=.035.We acknowledge the importance of the authors' work on the evaluation of the functional role of PAG in patients with FD by examining changes in the endophenotype with techniques ranging from fluorodeoxyglucose-PET to resting state fMRI. 9,10 However, we would like credibility to be conferred to the results and conclusions after careful data analysis and interpretation. ACKNOWLEDGMENTWe thank our colleagues from Integrated Brain Research Unit who provided insight and expertise that greatly assisted the research. COMPETING INTERESTS