Problem with using DeepCAD in conjunction with suite2p

[JessicaYoujiaLi]

Hi,

I have trained DeepCAD with my calcium imagining dataset and the denoised results looks good, however when I am trying to use the denoised data in another image processing pipeline(suite2p) for cell detection (data have been re-formatted to what is required for suite2p input), the denoised image ended up looking more blurry than the raw image after processed by suite2p. I am wondering if this is something have occurred to anyone before, and if so then is there any good ways to fix the problem?

The suite2p pipeline also included a simple denoise module using pca decomposition, I am wondering if it could be something that interfere with result? Or overall, is there anything to look out for when using the DeepCAD denoised result in conjunction with other analysis pipeline?

Thank you so much! Greatly appreciate for any tips and advices.

[cabooster]

Sorry, I have never used suite2p and we used CaImAn​ for neuron and signal extraction. In some cases, especially the input SNR is very ​low, DeepCAD-RT will output blurry results because of the low-pass nature of convolutional neural networks.

[JessicaYoujiaLi]

Thank you so much for getting back to me fast and explaining the process! In this case, I was wondering if you have an existing pipeline that takes the output from DeepCAD-RT and feeds it into CaImAn for signal extraction?