A multiresolution approach to the extraction of the pyloric rhythm

Abstract

Abstract-This paper describes our work toward the development of a computationally robust methodology to identify the pyloric neurons in the stomatogastric ganglion of Cancer pagurus using voltage-sensitive dye imaging. The multi-resolution signal decomposition procedure constructed using the sequential Singular Spectrum Analysis approach to isolate the pyloric rhythm from optical recordings of dyed live cells is presented. Early results suggest that the developed procedure offers a demonstrably reliable way to extract the rhythm from the recording data of these cells.

Towards an Accurate Identification of Pyloric Neuron Activity with VSDi

“…Several key challenges had to be overcome. (1) The relatively low yield 1 of the potential-dependent fluorescence change (2-10%/100mV) achieved by the Di-4-ANEPPS dye used in our study make the PR extraction from the recorded dye bath/images a complex signal processing task [5]. (2) The inherently tri-phasic PR monitored on the lvn by the activity of three distinct types of neurons; namely, lateral pyloric (LP, one cell) neuron, the pyloric constrictor (PY, 5 cells) neurons and the pyloric dilator (PD, two cells) neurons, is recorded as extracellular neuronal potentials (i.e.…”

“…The HRI was used to identify neurons and select the pixels corresponding to the neurons as distinct regions of interest (ROIs). Signals corresponding to these ROIs were extracted from the 3-D stack in the form of 1-D time series data [5].…”

“…This provided the discerning references with which the detected pyloric rhythm from the individual ROIs can be compared and analysed [5].…”

“…However, while a detailed discussion or review on this approach is outside the scope of this paper, it should be noted that it is most difficult to identify the best algorithm with sufficient generality and also to define which spike sorter is the most appropriate given the task at hand [13]. In particular, the features-based approach that is primarily based on PCA and ICA to group spikes into clusters both rely on the orthogonality of components in a mixture and are unlikely to perform well, as the tri-phasic pyloric waveforms generated by the participating neurons are necessarily non-orthogonal [5]. Likewise, the template-based approach is unlikely to work well, as both the spatial and temporal variations displayed in the lvn would require the availability of an impractically large number of samples to be compared with the demonstrably noisy data obtained using the VSDi [5].…”

“…This problem is discussed in [5]. In this paper, we present our work toward addressing the outstanding challenges.…”