Power Spectral Frequency and Coherence Abnormalities in Patients with Intractable Epilepsy and Their Usefulness in Long-Term Remediation of Seizures Using Neurofeedback

Medically intractable seizures appear to be highly correlated with focal slow activity (delta or theta). They also correlate highly with decreases in the coherence of theta. Normalization of focal slowing and of decreased theta coherence will probably be the neurofeedback approaches most likely to decrease or eliminate seizures in future cases. Neurofeedback has been used for over 35 years to reduce the incidence and severity of seizures. With power training to decrease theta and increase the sensorimotor rhythm (12–15 Hz), an average of 82% of patients experienced a significant reduction in seizure frequency, and occasional remissions were seen. Recent improvements using QEEG to guide neurofeedback training have made it possible to eliminate seizures in most patients, even those with intractable seizures. Following our previous study in 2005, we report an additional 25 patients so treated. We also report an analysis of the frequency of QEEG abnormalities in this patient group. All of the intractable epileptic patients had one or more slow foci (excessive theta or delta compared with the normal database). One third had a relative deficiency of beta power. One fourth had a deficiency of absolute delta. Eighteen percent had excessive absolute alpha power, 18% had deficient absolute alpha power, 18% percent had excessive absolute beta power, and 18% percent had deficient absolute beta power. Hypocoherence of theta was found in 75%, and decreases in alpha coherence were noted in 42%. Hypocoherence of beta was found in 50%, and hypocoherence of delta was found in 25%. Increases in alpha coherence were noted in 33%. Seventeen percent had no coherence abnormalities. When most of the power and coherence abnormalities were normalized with neurofeedback training, all the patients became seizure-free; 76% no longer required an anticonvulsant for seizure control.