So I actually do two things. One is to use an annotation layer to just drop points to see where I’ve been. It’s the closest thing to using Custom Highlight or Scythe Complete. But also, yes, I do actually split things into a few chunks that I can hide until I’m ready to get to them. I join them back together after.
3 Likes
Interesting but i tried and it does not work for me. As i’m dyslecsic it’s pretty difficult to mentaly manage with.
I choose the split option that allows me to hide sections or just let as it is with different color, and it’s sooooo handy !
1 Like
Yeah, I do a lot of intentional color changes to help me keep track too! And I will say that if you want, feel free to put an annotation layer where you drop points along an entire branch, not just the ends. I use multiple annotation layers for different branch ends and also a layer for just tracking along the branch itself. It does create a lot of little dots when you zoom out but you can just hide that layer as well when needed, too.
2 Likes
Cell looks good by the way! And yes as Nseraf mentioned, these types of cells usually have their “soma tract” (branch to the cell body) near the dendrite ends.
2 Likes
By the way i just complete my first and certainly biggest one ! There’s still some cleaning to do, but as i understood it’s not the purpose/needed ?
1 Like
Yes, the banc-bot “todo” lists consists of cells that are in the “neck”. Most of them have been proofread so what remains in the list are probably some very tough/hard to trace neurons 
And yup, just double click on something in the 2D once you’ve navigated your axis somewhere inside the red box (looks like you got it already!)
And no problem! Even the GMs all have different learning styles, so we’re here to help! 
There are some demo videos in the blog post, but we’ll look at putting together a quick proofread process video (if folks are good with it, we’ll record the Zoom Intro video on Thurs too that will demo the process).
2 Likes
i think i would have had a second look at the connection too that long branch straight branch, from the picture it looks like it could be a second cell, but could ofc be wrong
On flywire i think some of the scientist did say that we should not split cells for easier proofreading because this made some sort of a problem, but can´t remember what. is it ok to do it here.
i have mostly been taking a fast look at the cell after any obvious mistakes and making sure the backbone is correct, will ofc miss things like this but it is a tradeof betwheen speed and accuracy.
i have mostly marked my cells backbone proofread and central neuron ( think that is covering everything not going down trough the neck even if that is a very broad category)
4 Likes
In FlyWire we weren’t allowed to split cells, that were already marked as completed. The reason was, that the scientists had already have enough data to do some science. In BANC case, they, so far, have around 10 000 cells completed, mostly in the neck area, so, I think, it’s to early to do some research.
As for the speed vs accuracy - I agree. I only look for the cell body, for dendrites and for axons. If all these parts are visible, then I search for any thicker branches, that are broken or for any obvious missing parts (e.g. gaps in dendritic or axonic tree).
Otherwise, if there are, lets say, 100 people working on the dataset, and they would do 1 cell per day, and the dataset has more than 140 000 neurons, it would took them over 4 years, assuming each person would work everyday and each would complete their cell. And that seems to be pretty optimistic assumption.
3 Likes
I do not understand how annotations work. I try to add a layer and chose annotation but there’s no possibility :
I only can add annotations on the “2 segmentation proofreading” layer
That’s weird. Maybe try doing hard refresh (Ctrl + R or Ctrl + F5).
The annotation system works for me without any problem.
1 Like
yeah but let’s face it, none of us is doing only 1 cell in a day…more like…10? 100? lol i give us 6 months and the thing’s done. 
1 Like
If we were trying to add every single missing trace, we wouldn’t do much more. Hence my answer, that we should only look for obvious things, without delving into details.
And I very much doubt about that 6 months 
FAFB (FlyWire) was a, in a sense, subset of BANC and it took 3 (?) years to complete it. And I have a feeling, that BANC is a bit more difficult to trace. For example, the mitochondria, or thicker slides (40 vs 45nm). As always, we’ll see
2 Likes
did it take that long? Are we talking tracing neurons alone or including annotating and classifying each neuron type/cell as well? Didn’t feel that long lol.
When we started (may 2022?) most of it was already traced, especially the central brain. It started somewhere around 2020, if I’n not mistaken.
2 Likes
Yep i checked twice this long branch and found nothing wrong.
1 Like
If you could have a look at this cell, just to tell me if it’s complete to you ?
I dont 3 other today, but would like a check for this one.
1 Like
Looks good to me! If you want to get a little more detailed, I added a few ends on the dendrites as I was checking (but honestly, don’t worry about getting these on most cells): Neuroglancer
Update for getting cells to proofread in the left side of the GNG! 
Simply message banc-bot in Slack
todo gng
and it will give you a list of 5 cells that haven’t been proofread (warning it may take up to 30 seconds to respond).
The todo list works from the automated nucleus segmentation query that was run in the BANC. This query detected all the “nuclei” (looks like the circle area/round bean in the somas/cell bodies) so be aware that the nuclei may not be attached to their surrounding cell body structure when you start. You’ll need to add in the surrounding membrane for the nucleus seed.
This also means that the list only pulls the somas in the left GNG red box. So if you prefer to just click and select random segments to proofread in that zone, feel free!
1 Like