The comparison of long-term decoding performance between low and high frequency local field potentials in macaque monkeys

The long-term stability is one of the most important issues to be considered when designing practical Brain Machine Interfaces(BMIs). Because Local Field Potentials (LFPs) could provide durable recording, they were supposed to be an alternative promising signal source. Previous studies have shown that much kinematics information could be extracted from LFPs in motor cortex, especially from the low (0.3-5 Hz) and high frequency band (200-400 Hz). However few studies had compared their long term stability. Here, we assessed the decoding performance by offline analysing LFPs recorded from motor cortex in three well-trained monkeys performing 2-D center out task. The signal recording covered nearly one year for two monkeys and two years for the other monkey. The results showed that in most sessions, high frequency LFPs showed better decoding performance than low frequency band in both discrete and continuous decoders and high frequency band was less sensitive to the number of channels with significant tuning than that of low frequency band. Therefore, all the results indicate that the decoding performance of high frequency band is able to keep higher decoding power and it is more suitable for long-term BMIs applications.