Post here about Eyewire II
if a starting segment has 2 or more cells (sac and/or other) complete all of them?
Hmm, that’s a good question. Could you share with me the segment ID or link? Let me have a look and I’ll see if the researchers have any preferences. Thanks!
1 Like
i dont have one yet, just thinking/asking ahead of time, as I’m guessing it’s only a matter of time before it comes along someone’s path.
Ah yes, well since our current goal is proofreading SACs, then just proofread the SAC if there’s more than one cell merged together for the segment (a good example of this situation we’re discussing is line 77 with starting segID 720575940561395458).
I’ll get back to about how we should proceed with potential other cases.
1 Like
gotcha, thanks!
I found an example neuroglancer
line 67 in gsheet, and I think it’s an on and an off SAC? (I’ve split them but the 2 segments are 1 in starting segment 720575940558800732
Do I trace both to completion? (I mean I will but do I place both in gsheet?)
The soma annotation - can it be placed on the surface of the soma or should be inside it (and confirmed in 2D, that is in the actual segment, not in the nucleus or a mitochondrium)? I’m asking, because from my programming experience, when we click on the surface of a segment, take the coords for such point and then try to get the segment by coords, if often turns out, that the point was actually in one of the neighbouring segments.
3 Likes
if one of the gsheet entries is a wrong type cell (ganglion, bp, etc), and we complete it should we still do the whole true end/dataset end etc annotations?
1 Like
is there a keyboard shortcut that allows a back and forth movement between these tabs w/o clicking w/ the mouse? (ie: alt + 1/2/3 etc)
1 Like
You were close - it’s Ctrl + 1/2/3, etc. 
3 Likes
wut? For real? xD well thank you. lol
1 Like
Probably inside soma - but I will confirm.
2 Likes
Yes, please confirm you are inside the soma, and inside the selected segment (i.e. not in the nucleus) when placing the soma annotation. Place in 2D, that should fix any oopsie the neighbor got labeled by accident issues.
2 Likes
What level of proofreading are we looking for in these tasks? Are we only concerned about obvious mergers & checking the ends, or should we be aiming for complete proofing along the branches?
2 Likes
Yes, obvious mergers and checking the branch ends. Similar to the level of proofreading we did for FlyWire/BANC cells. Since we’re familiar with retinal cells, if you see a gap where you generally know a branch belongs (cough SACs cough) then we’ll need to go and find it. However, for the small “twig” length branches we don’t need to do complete proofing to find and attach all of those.
In general, the segmentation version we chose for Eyewire II tends to lean towards having more mergers. This is because it’s easier for us to see what’s wrong than to find something that is missing. Hence, the dataset has a higher threshold of connecting segments together (sort of like the MSTY slider in the Mystic).
3 Likes
Looks like I managed to find a double SAC combo on the sheet - for now I’m annotating both on the single line in the sheet, but if it’s preferred I split them into two tasks on the sheet just let me know!
(This is from starting seg 720575940546970297 in line 18, after splitting the merger).
3 Likes
in Pyr/ CO3 dataset gsheet we have “too small” for this kind of a thing:
end w/2D is end of dataset, there’s nothing else of this cell inside (that I can find).
What/how do I/we declare these here in ew2? Wrong type?
3 Likes
Good idea, I’ve added this into the drop menu for “Status”
2 Likes
a clarification: in the special task, you want us to keep (and hide if/where needed) any mergers removed:
- from the initial/origin Segment
or - from initial segment and any from whatever missing branches (find a missing branch it is mergered with a bunch of other stuff, split that, add another segment to that it also has mergers etc etc) we might find?
(I’m asking b/c in 2 in some cases it’ll get quite ‘crowded’ quite fast, lol).
2 Likes