This code is essentially forked from the following work:

Usiigaci: Label-free instance-aware cell tracking in phase contrast microscopy enabled by machine learning Hsieh-Fu Tsai^1,2, Joanna Gajda³, Tyler F.W. Sloan⁴, Andrei Rares⁵, and Amy Q. Shen¹

¹Micro/Bio/Nanofluidics Unit, Okinawa Institute of Science and Technology Graduate University, Okinawa Japan

²Research Fellow of Japan Society of Promotion for Science

³AGH University of Science and Technology, Krakow, Poland

⁴Quorumetrix Solutions, Canada

⁵ImagineA, The Netherlands

And adapted for 2p calcium imaging. The main changes are

• Added processing of 2p movies into single frames for initial segmentation
• Added code to extract signals from movies following segmentation by the network

For the trained network weights, contact: noah.dolev@weizmann.ac.il