Abstract
Functional multineuron calcium imaging (fMCI) provides a useful experimental platform to simultaneously capture the spatiotemporal patterns of neuronal activity from a large cell population in situ. However, fMCI often suffers from low signal-to-noise ratios (S/N). The main factor that causes the low S/N is shot noise that arises from photon detectors. Here, we propose a new denoising procedure, termed the Okada filter, which is designed to reduce shot noise under low S/N conditions, such as fMCI. The core idea of the Okada filter is to replace the fluorescence intensity value of a given frame time with the average of two values at the preceding and following frames unless the focused value is the median among these three values. This process is iterated serially throughout a time-series vector. In fMCI data of hippocampal neurons, the Okada filter rapidly reduces background noise and significantly improves the S/N. The Okada filter is also applicable for reducing shot noise in electrophysiological data and photographs. Finally, the Okada filter can be described using a single continuous differentiable equation based on the logistic function and is thus mathematically tractable.
Highlights
Functional multineuron calcium imaging is a promising optical technique that simultaneously records suprathreshold activity from a large number of neurons [1,2,3]
This conditional process is important to preserve the waveform of signal; note that wildly used filters, such binomial and Savitzky-Golay filters, inevitably smooth the waveforms of sharp signal and may lead to erroneous detections of the exact timings of signal onsets (Fig 2)
We originally developed the Okada filter to denoise vector sequences with low signal-to-noise ratios (S/N) and with low temporal resolution, such as Functional multineuron calcium imaging (fMCI) data, we tried to apply it to other forms of data
Summary
Functional multineuron calcium imaging (fMCI) is a promising optical technique that simultaneously records suprathreshold activity from a large number of neurons [1,2,3]. In fMCI time-series data, shot noise is usually reduced through post hoc filtering using a moving window. A single outlier due to shot noise affects nearby values within the window This undesirable effect is relevant in fMCI data that are acquired at sampling rates as low as tens of hertz. A median filter replaces the value at a focused time with the median value within a moving window This procedure tends to preserve the structure of a signal; unlike linear filters, a median filter selects the replaced values from a pool of values that already existed in the original data, thereby resulting in less efficient denoising. The filtered data contain "more centered" values that did not exist in the original values (as in linear filter) and have a greater likelihood of preserving the structure of the putative signal because any values during a consecutive change are unlikely replaced (as occurs when using a median filter)
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
