Separation of signal and noise from in vivo optical recording in Guinea pigs using independent component analysis.

Optical recording in vivo severely suffers from the interference of heartbeat noise. So far, heartbeat noise has been minimized by subtracting from each experimental trial an average of interlaced control recordings. This method, however, is time-consuming and increases tissue damage due to phototoxicity. Here we applied independent component analysis (ICA) to in vivo optical recordings, for separation of auditory signals and noises. Our results show that ICA can be successfully used to separate sound-evoked signals and heartbeat noises. Compared with the previous method, ICA has a comparable power of separation and does not require background recordings.