New math tools reveal brain activity more accurately by correcting noisy calcium signals
When a firework bursts in the sky, what often follows is a cloud of smoke in the shape of the firework—a visual echo that lingers long after the spectacular explosion itself is over. A similar phenomenon happens in the brain when a neuron fires: The explosion of activity causes a change in calcium levels, resulting in fluorescence that neuroscientists use as a proxy for brain activity.…
When neurons fire, they generate a burst of calcium activity that produces a fluorescence signal, which scientists use as a proxy for brain activity. However, this proxy is imperfect, as the fluorescence only approximates the true shape, size, and origin of the neural activity. Neuroscientists have attempted to correct for these inaccuracies using a mathematical algorithm called Lucy–Richardson, which was originally developed for astrophysics.
However, this algorithm relies on the assumption that noise in the data follows a specific statistical distribution (Poisson), which may not hold true for brain activity data.
Merav Stern, a physics fellow at Rockefeller University, has developed three entirely new computational algorithms that more accurately recover the underlying neural dynamics from wide-field calcium imaging data. These algorithms—dynamical binning, first differences, and continuously varying—outperformed the traditional Lucy–Richardson deconvolution method in correcting the statistical noise present in wide-field calcium imaging data.
The key insight behind these new algorithms was Stern's discovery that brain noise follows a Gaussian distribution rather than a Poisson distribution.
By modeling the noise as Gaussian, Stern's algorithms can more accurately recover the neural activity itself, rather than using calcium fluorescence as a proxy. This approach allows neuroscientists to precisely observe how behavior relates to brain activity and vice versa, using light-based techniques like wide-field calcium imaging. The new algorithms can be adapted to various types of data that involve aggregated neuron activity, such as neuropil recordings, fiber photometry, and nonspiking neurons.
Stern's algorithms are now available for download on GitHub, enabling other researchers to analyze and publish improved brain imaging data. These tools have already been used by some scientists to enhance their findings. By providing neuroscientists with more reliable estimates of neural activity, Stern's algorithms have the potential to significantly improve the conclusions drawn from brain activity recordings.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.