Turn on cookies to sign in
Signing in saves your progress to your Curie account. We can only do that with cookies on — turn them on to continue.

Seeing the brain more clearly — without better hardware. A Rockefeller University physicist developed three new computational algorithms that recover actual neural firing patterns from calcium imaging data, outperforming the decades-old astrophysics-borrowed method currently in use. The key insight? Brain imaging noise follows a bell curve, not the Poisson distribution that existing tools assume, making those tools a poor fit for neuroscience data.
Seeing the brain more clearly — without better hardware
Neuroscientists studying brain activity often rely on calcium imaging, a technique that uses fluorescent light signals as a stand-in for actual neuron firing. The problem: those signals are blurry, delayed, and contaminated by noise from blood flow, light scattering, and camera artifacts — making it hard to pinpoint exactly when and where neurons fired.
Merav Stern, a physics fellow at Rockefeller University, tackled this not by building better imaging hardware, but by rethinking the math. She developed three new computational algorithms — dynamical binning, first differences, and continuously varying — plus an adapted Wiener Filter approach, all published in Cell Reports Methods. Her critical discovery: noise in brain imaging data follows a Gaussian (bell curve) distribution, not the Poisson distribution assumed by the widely used Lucy-Richardson algorithm borrowed from astrophysics.
Key Takeaways:
Why it matters: More accurate brain activity data means more reliable conclusions about how the brain drives behavior — a foundational step for advancing neuroscience research and, ultimately, our understanding of neurological conditions.