Regional cerebral blood flow changes in patients with internet addiction.

Dear Editor, Internet addiction (IA) has become a severe challenge of our modern world today, though little is known about its pathology. In this context, the interesting study by Liu et al. in the May-August 2016 issue of HJNM using Tc-labelled ethylene biyldicysteinate dimer single photon emission tomography (SPET) at rest and after pharmaceutical (adenosine) stress is more than welcomed. As this seems to be the first perfusion SPET study in this indication, the obtained data may be discussed carefully. There are mainly the following questions: a) Regional cerebral blood flow (rCBF): There is no description on how the rCBF was calculated: Was it scaled relatively to the whole brain mean value or to the cerebellar mean value? b) P value threshold and clusters: There is no indication of whether the authors are performing any kind of correction for multiple comparisons in the statistical parametric mapping (SPM) t-test. This, combined with the use of a really "liberal" voxel P value of only 0.01 could be subject to providing many false positive results. Generally a P value threshold of 0.001 should be used. In addition, there is no information related to the clusters. For the question of the validity of parametric statistical methods used for the analysis of functional neuroimaging data, we would like to mention the important recent paper by Eklund et al. 2016. c) Data analysis: The authors state (p. 97): "As some abnormal rCBF in adenosine-stressed state might relate with normal responses to adenosine compared to resting state, we excluded those regions that showed abnormal rCBF in stressed state in healthy controls (Table 4) from those in IA group (Table 5). The rest abnormal regions were compared between the IA group and the control group". For this, with SPM a flexible factorial design with all the data rather than only t-tests would have been interesting to find out whether the difference between the groups at stress is the same difference observed between the groups at rest. In "traditional" region-of-interest statistics, a repeated 2-way ANOVA to account for individual variance would have been done. The authors are, however, just "manually" removing the regions that pop-up in the previous analysis. In conclusion, it would be of great scientific interest if the authors of this first SPET study on an important indication, IA, could give us some more details on their data.