Low computational-cost cell detection method for calcium imaging data

The rapid progress of calcium imaging has reached a point where the activity of tens of thousands of cells can be recorded simultaneously. However, the huge amount of data in such records makes manual analysis difficult. Consequently, there is a pressing need for automatic analysis for large-scale image data. Some automatic cell detection methods use machine learning, but their scalability to the data size remains a fundamental problem; they cannot be completed within a practical period of time on large-scale data acquired with recently developed ultra-large field-of-view microscopies. Here, we propose a low computational-cost cell detection (LCCD) method, which can process huge amounts of data within a practical period. We compared it with two previously proposed methods, constrained non-negative matrix factorization and Suite2P. The detection accuracy of LCCD was close to those of the other methods, whereas its calculation time was about ten times shorter.