One of the challenges in neuroscience is the difficulty in knowing precisely which neurons are firing at any given time. This requires tracking the activity of individual neurons of which there are billions in a human brain. Brain imaging software creates such large volumes of data that researchers have to spend months to manually analyse the activity of thousands of neurons.
Andrea Giovannucci et al. at the Flatiron Institute, Simons Foundation, United States developed an open source software called CalmAn that provides a much quicker method to analyse large sets of brain imaging data. It is already used by over one hundred laboratories around the globe. This freely available software can be customised and developed further by its users. There is no need for specialised equipment, it can simply run on a laptop.
CalmAn analyses data in real time while scientists run their experiments. It replaces the manual counting process with a combination of standard computational methods and machine learning techniques. The software is able to achieve near-human accuracy as shown in the image below. The advantage of the software is its much greater speed.
One of the main problems for human researchers is speed. “People spent more time analyzing their data to extract activity traces than actually collecting it,” says Dmitri Chklovskii, leader of the neuroscience group at the Center for Computational Biology (CCB) at the Flatiron Institute in New York City.
The software also deals with motion correction, neural activity identification as well as registration across different sessions of data collection, all common problems to pre-processing. In addition, CalmAn is suitable for both, two-photon and one-photon imaging.
The detection of active neurons is based on calcium imaging data. A special dye is added to brain tissue which binds to the calcium ions responsible for activating neurons. The dye lights up under ultraviolet light but only when the dye binds to a calcium ion. This allows the neuron’s activity to be tracked.
Another problem is the noise within the data as neurons often overlap and brain tissue doesn’t stay perfectly still. This can lead to the same neuron being mapped multiple times.
The speed of the software now enables researchers to adapt their experiments during the mapping process showing how different behaviour is influenced by different neurons.
This new technique will hopefully accelerate neuroscience research and bring a greater variety of therapies and cures to waiting patients in less time.