Stereo-electroencephalography (SEEG) is widely used to record the electrical activity of patients' brain in clinical. The SEEG-based epileptogenic network can better describe the origin and the spreading of seizures, which makes it an important measure to localize epileptogenic zone (EZ). SEEG data from six patients with refractory epilepsy are used in this study. Five of them are with temporal lobe epilepsy, and the other is with extratemporal lobe epilepsy. The node outflow (out-degree) and inflow (in-degree) of information are calculated in each node of epileptic network, and the overlay between selected nodes and resected nodes is analyzed. In this study, SEEG data is transformed to bipolar montage, and then the epileptic network is established by using independent effective coherence (iCoh) method. The SEEG segments at onset, middle and termination of seizures in Delta, Theta, Alpha, Beta, and Gamma rhythms are used respectively. Finally, the K-means clustering algorithm is applied on the node values of out-degree and in-degree respectively. The nodes in the cluster with high value are compared with the resected regions. The final results show that the accuracy of selected nodes in resected region in the Delta, Alpha and Beta rhythm are 0.90, 0.88 and 0.89 based on out-degree values in temporal lobe epilepsy patients respectively, while the in-degree values cannot differentiate them. In contrast, the out-degree values are higher outside the temporal lobe in the patient with extratemporal lobe epilepsy. Based on the out-degree feature in low-frequency epileptic network, this study provides a potential quantitative measure for identifying patients with temporal lobe epilepsy in clinical.