Necoda-compressed Allen Brain Observatory calcium imaging dataset

We present Necoda, a transformative deep learning framework that compresses functional neural imaging data by over 1,000-fold while faithfully preserving signal fidelity. This work addresses the critical "data gravity" bottleneck that currently impedes research reproducibility, data sharing, and collaborative discovery in neuroscience.To demonstrate real-world utility, we applied Necoda to the Allen Brain Observatory visual coding dataset. We compressed 4.82 TB of data into a portable 4.81 GB package.

我们提出了Necoda——一款革命性的深度学习框架，可将功能性神经成像数据压缩千倍以上，同时精准保留信号保真度。本研究解决了当前阻碍神经科学领域研究可重复性、数据共享与协同发现的关键「数据引力（data gravity）」瓶颈问题。为验证其实际应用价值，我们将Necoda应用于艾伦脑观测数据库（Allen Brain Observatory）视觉编码数据集。我们将4.82 TB的原始数据压缩为仅4.81 GB的便携数据包。