Registration--vEMDiffuse-a

[ThatIsMyUsername]

This is excellent work. In the “Denoise data pre-processing” section of the Methods in your paper, you mentioned that “properly aligned data is important,” and the code you provided also includes a registration operation. I would like to ask whether all datasets used in vEMDiffuse-a were registered. In other words, which datasets in vEMDiffuse-a did you apply registration to?
Your reply would be very helpful to me. I wish you all the best.

[Luchixiang]

Thank you for your interest. The vemdiffuse-a is trained on anisotropic volume data, and yes, it should be aligned along the z-axis. But we didn't do it for ourselves since training data is all public data and aligned with data curation. If you are collecting your own vEM data, you should do alignment along the z axis. You can use something like https://imagej.net/plugins/linear-stack-alignment-with-sift.

[ThatIsMyUsername]

Thank you for your interest. The vemdiffuse-a is trained on anisotropic volume data, and yes, it should be aligned along the z-axis. But we didn't do it for ourselves since training data is all public data and aligned with data curation. If you are collecting your own vEM data, you should do alignment along the z axis. You can use something like https://imagej.net/plugins/linear-stack-alignment-with-sift.

Thank you for your reply. Since the public datasets shown in the vEMDiffuse-a paper appear relatively clear, I would like to ask whether any preprocessing was applied to these public datasets, such as image enhancement operations or other related processing steps.

[Luchixiang]

Hardly not, only adjusting the contrast.

[ThatIsMyUsername]

Hardly not, only adjusting the contrast.

Which method was used for contrast adjustment? Would it be possible for you to describe it in more detail?
Also, was the contrast adjusted before training, or was it adjusted after training only for visualization?
Thank you for your reply. This information would be very helpful to me.

[Luchixiang]

Contrast adjustment is applied after the training, only for visulization. FIJI autocontrast is used.

[ThatIsMyUsername]

Contrast adjustment is applied after the training, only for visulization. FIJI autocontrast is used.

Thank you very much for your reply. You are as outstanding as your work. My very best wishes to you.