A Systematically Optimized Miniaturized Mesoscope (SOMM) for large-scale calcium imaging in freely moving mice
Interrogating neural circuits in freely behaving mammals is poised to shed a light on the neuronal systems dynamics underlying complex naturalistic behaviors. However, optical recording of neuronal activity in freely behaving animals has remained limited to a small scale and is vulnerable to motion-induced focus drifting. Here, we present a systematically optimized miniaturized mesoscope (SOMM), a widefield, head-mounted fluorescent mesoscope that overcomes these obstacles and allows imaging during free behavior at mesoscopic field-of-view, single-cell resolution, with uniform illumination, and robust axial accessibility. Powered by compact diffractive optics and associated computational algorithms, SOMM can capture neuronal network activity within a field-of-view of 3.6 à 3.6 mm¬2 at 4 ¡m resolution and at up to 16 Hz in the cortex of freely moving mice, with great defocus tolerance across 300 ¡m and a weight of less than 2.5 g. Using SOMM, we recorded large-scale population activity during social interactions, cross-region neuronal activity evoked by visual and electrical stimuli, and neurovascular coupling in dual-color, all at single-cell spatial resolution and physiologically relevant temporal resolution
In this part we introduce the DOE optimization in SOMM.
- Ubuntu 16.04
- Python 3.6
- tnesorflow = 1.14
- NVIDIA GPU (24 GB Memory) + CUDA
- run DOE_optimization\gen_zernike_polynomial.m to generate Zernike basis for optimization
- Code for generating NAOMi samples built for one-photon (or widefield) imaging modality can be found in https://github.com/yuanlong-o/Deep_widefield_cal_inferece
- Run main_LFOV_DOE_train.py to train a DOE and corresponding decovnolution algorithm for large FOV capability and depth robustness. Optical parameters should be corresondingly modified for different systems.
- The output phase would
- The mechanical design for single-color SOMM can be found in Mechanical\SOMM folder. Require Solidworks >= v.2022.
- The mechanical design for dual-color SOMM can be found in Mechanical\Dual_Color_SOMM folder. Require Solidworks >= v.2022.
- The soft PCB design for dual-color SOMM can be found in TODO folder. Require Solidworks >= v.2022.
- We use L2 regularized deconvolution for each patches. A demo data and demo script can be found in Processing\Deconvolution.
- We use DeepWonder for extracting neuronal spatial and temporal profiles from deconvolved videos. The full processing code can be found in Processing\DeepWonder. A standalone readme containing environment setup and running instruments can be found in Processing\DeepWonder\readme.md
Yuanlong Zhang*, Lekang Yuan*, Jiamin Wu, Tobias NΓΆbauer, Rujin Zhang, Guihua Xiao, Mingrui Wang, Hao Xie, Qionghai Daiβ‘, and Alipasha Vaziriβ‘, "A Systematically Optimized Miniaturized Mesoscope (SOMM) for large-scale calcium imaging in freely moving mice", bioRxiv 2022.
We are pleased to address any questions regarding the above tools through emails (yuanlongzhang94@gmail.com).