Aims: 

Calcium transients in cells of the dorsal cochlear nucleus (DCN) are studied using a multi-neuronal calcium imaging technique. In order to decide whether a specific event is linked to an action potential, simultaneous patch clamp recording would be done. Our aim was to establish a method that could differentiate between action potential-linked and other events on calcium images.

Methods: 

A fluorescent imaging system was used to capture image series of DCN slices loaded by Oregon Green 488-AM calcium binding dye. 172 by 130 pixel frames were captured every 100 ms. Simultaneous extracellular loose patch recordings were done in voltage clamp mode at a sampling rate of 10 kHz. We developed and implemented algorithms for automatic region of interest (ROI) selection and automatic analysis of data in each ROI.

Results: 

In order to select ROIs, time series data of individual pixels are analysed using stationary wavelet transform (SWT, Szab'o et al. 2010). Frequency components not corresponding to calcium transients are removed and neighboring pixels with high intensities are formed into ROIs. Transients from each ROI are also analysed using SWT. In addition to traditional parameters (amplitude, slope etc), a wavelet-based parameter is used to differentiate fast, action-potential linked events from others. In case of recordings where both calcium imaging and patch clamp recording were done, the classification algorithm reached o ver 90% accuracy.

Conclusion: 

The developed algorithm enables us to detect action potential linked (neuronal) calcium transients on fluorescent calcium images, eliminating the need for parallel patch clamp measurement.

Reference: 

Szab'o L.Z., Vincze J., Csernoch L. & Szentesi P. J Theor Biol 264, 1279–92