Epilepsy is a neurological disorder characterized by recurring, unprovoked seizures. For refractory epilepsy patients, resective surgery does not always result in seizure freedom. The purpose of this study was the localization of seizure activity from sEEG data using dynamic mode decomposition (DMD) to aid in resective surgery planning. Analysis was performed on retrospective sEEG data. A band pass filter was applied between 1 and 40 Hz. Individual Component Analysis was performed to reduce noise. A common average reference filter was applied. The data were down sampled to 128 Hz. A two-minute window of data centered on clinically documented seizure onset was selected. The data were then quantified using our DMD method to provide a low-dimensional representation of the complex high-dimensional seizure data (blind source separation). The real part of the dominant mode was extracted for each DMD update. A matrix was created for each seizure with a row per recording channel and a column per each DMD update. Each matrix was visualized as a heatmap. Channels in which the ten highest values occurred were compared to clinically documented seizure onset locations. Using the binomial cumulative distribution, we assessed whether the percentage agreement was better than random chance. Statistically significant agreement (p < 0.05) was found in three sEEG patients with additional analysis ongoing. We concluded that DMD can identify the “seizure mode” of activity and may therefore aid in the analysis of seizure source localization. Further analysis will show the extent to which DMD can contribute to improved patient outcomes.