Pharmacologically refractory epilepsy afflicts more than a million Americans. In a substantial proportion (25%) of these cases, epilepsy originates in the mesial temporal lobes (mesial temporal lobe epilepsy; MTLE), engaging and perturbing the neural mechanisms underlying episodic memory. While many patients with MTLE are candidates for resective surgery, no reasonable resective or ablative options are feasible in locations where viable memory and epilepsy co-exist – a condition we term Epilepsy in a Precious Hippocampus (EPH). Current neuromodulation techniques and devices, inspired by stimulation parameters for movement disorders, deliver unifocal high frequency stimulation. They have limited impact on seizure burden and rarely result in durable seizure freedom. For the EPH conundrum, there is an urgent requirement for more efficacious neuromodulatory solutions, since the threat of major morbidity and mortality is otherwise substantial. We propose a novel network-based neuromodulation approach to disrupt seizures at levels above what has previously been accomplished. Eight patients with EPH who have been scheduled for stereo-electroencephalography (SEEG) will be recruited at two busy epilepsy centers - UTHealth Houston and the Mayo Clinic. They will undergo recordings from the anterior nucleus of the thalamus (ANT) in addition to traditional SEEG targets in MTLE, while seizure foci are localized. Patients will then undergo chronic depth electrode implants targeting pyriform cortex (area tempestus), ento-rhinal cortex, fornix and ANT, and implantation of the Medtronic Percept PC Neurostimulator DBS System with BrainSense Technology. Intracranial data from both implants will be used to build a comprehensive individual and generalizable network model. We will demonstrate safety and efficacy of the Percept PC device for detecting seizures. We will then utilize low frequency stimulation at each of these targets and patterned network stimulation to determine their impact on seizures. The most effective approach will be implemented for longer durations, and comparisons on seizure control will be made relative to baseline and to traditional duty cycle high frequency ANT stimulation, that will be used as the control, wash-out stimulation parameter between the trials of each stimulation type. This work promises to dramatically improve our understanding of the MTLE network and develop a new approach to network based neuromodulation.