Astronomy Teaches Us to See the Invisible in the Brain ⚡ экспресс
Miniature synaptic calcium transients (mSCTs) in fluorescence microscopy videos cause only subtle signal changes, making automatic detection difficult. The challenge is analogous to finding astronomical transients across wide fields of view with variable noise. The proposed Astro-BEATS algorithm borrows image estimation and source-finding methods from astronomy to segment mSCTs in Ca²⁺ imaging. Astro-BEATS outperforms existing threshold-based methods in detection and segmentation quality. The resulting segmentation masks are suitable as training data for deep neural networks, and its high speed and ability to work on new datasets without retuning make Astro-BEATS an efficient tool for generating ground truth in neuroimaging.
To spot a dim star above a brightly lit city, astronomers subtract one image from another—removing the glare of streetlights. They do the same with living neurons: their faint flashes drown in noise, but if you subtract everything unchanging, only they remain. Measuring brightness and frame subtraction have been honed for centuries to find supernovae in galaxies. Now the Astro-BEATS program does the same with cell recordings, catching the tiniest changes in glow.
This allows rapid processing of hours of video and training of artificial intelligence. It brings us closer to understanding the language of neurons and, perhaps, to building brain-computer interfaces.
🎯 Tiny flashes of neural activity happen all by themselves, without any external trigger. Scientists still debate what they're for.
🎬 Science fiction writers have long dreamed of telepathy and controlling machines with thought. The ability to catch the brain's hidden signals brings those dreams closer to reality.