The use of least squares lattice algorithm in the parameterization and sorting of action potentials signals

摘要

The understanding of neuronal function under the action of a certain stimulus can be facilitated using techniques to distinguish the potential action from different neurons. Thus, from simultaneous recording of multiple neurons one can determine the firing patterns of each of them. Usually these techniques are implemented in three stages. From raw electrical potentials recorded using an intracranial electrode, spikes are detected, then parameterized and finally sorted, attributing every single spike observed to a particular neuron. Recently, it was proposed an on-line sorting method based on the noise level. Nevertheless, sorting is done directly based on the raw samples. In this paper we introduce an alternative way using the modified Least Squares algorithm based on the priori error with error feedback to parameterize the raw signals before classification. Preliminary simulations results show that using parameters provides performance near to results where the sorting is done directly based on the raw samples.