BLIND SEPARATION OF NEURONAL ACTIVITIES FROM OPTICAL IMAGING DATA
Susumu Takahashi, Yuichiro Anzai, Yuki Kobayashi, Takeshi Aihara, Minoru Tsukada
ftsusumu,anzaig@ayu.ics.keio.ac.jp, fkyuki,aihara,tsukadag@eng.tamagawa.ac.jp
We applied Independent Component Analysis(ICA) to
data of neuronal activities in the brain acquired by op
tical imaging method. We constructed a model of noise
and membrane potential mixed sources, and using suit
able parameters detected from its simulation based on
ICA, succeeded in estimating activities of neurons em
bedded in optical imaging data from hippocampal s
lices. Comparing our results with those obtained by
the averaging set of data popular in the application
of optical imaging method, it was empirically that our
method could extract neuronal activities in the brain
by oneshot data without averaging, which may pro
vides a more precise method for the analysis of the
complex mechanisms of the brain.