Eyewire II Question Box

[sorek_m]

Unfortunately, the supervoxels you see are the smallest segments available in the segmentation. So we sort of just have to go with leaving the supervoxel with what branch makes the most sense and leave gaps like we had to in Eyewire.

The MSTY threshold slider was to control the association of the supervoxels to each other. So the higher affinity threshold meant more segments/supervoxels were connected together by the system and the lowest affinity threshold was the max supervoxels/segments we could break down into individual pieces.

For EW2, we decided to go with a slightly higher affinity threshold for the dataset overall because it’s easier to spot mergers than missing segments. However, ng allows us to be able to cut/view the supervoxels (smallest segmentation) at the same time in the 2D. So we don’t have to worry about sliding the threshold slider up and down like we had to in MSTY.

[Nseraf]

the start link is missing the soma annotations tab, is that intended?

vs (from an older link)

[celiad]

Good catch, added back in now!

[AzureJay]

Hey all - I have been really preoccupied with some other stuff in my life and haven’t been able to focus on working the SAC list. If anyone wants to take over responsibility for line 37 (Index B33, ‘stuck’) and 143 (B139, WIP) - please feel free. I hate to see them just hang in limbo.

@celiad - Just a little response about feedback. I actually really enjoy this SAC project - the one thing that holds me back is that the newer version of neuroglancer has been so demanding on my system, that I have to pretty much close everything down on my computer to avoid things crashing while I’m editing. It’s frustrating - I can run high end games (16 GB RAM and 8 GB GPU) and other things without stressing the system, but I still can’t seem to find the magic settings for neuroglancer. As a result I tend to check in and work on neurons less often - because I have to fully prepare my workspace for it.

[Nseraf]

it may or may not help but try and delete all of -neurog- browser related history, neurog can create up to tens if not hundreds of GB’s worth of Browser history which can slow things down (and in my case constantly crash chrome). But yeahhhhhh in gen. neurog is one high maintenance gf to have lol it inhales RAM like a <inappropriate joke/metaphor>. xD

[KrzysztofKruk]

For the history problem, I’m using an extension from the Chrome webstore. My history used to grow to tens of GBs in a week or so.
I’ve started with deleting the whole history file (via the file system, the history inside the browser wasn’t able to load). Of course, it also deleted my whole history (but no cookies and remembered passwords).
Here are the locations of the history file for each OS: Google Chrome History Location | Chrome History Viewer . If you’re using Brave (as I do), replace “Google” with “BraveSoftware” and “Chrome” with “Brave-Browser” in the paths.
After deleting the file, I’ve installed HistoryGuard and configured it to block all the entries from the appropriate domains (keep in mind - whole domains, e.g. https://spelunker.cave-explorer.org/, without the rest of the URL).
It’s not my extension, so I can’t guarantee, that there won’t be any issues with it, but I’ve been using it for over a month and my history file is mere 42MB (mega, not giga! ).

[celiad]

Thanks for feedback on the slowness issues. One thing that may help is changing your settings for GPU and System memory. To access these click the gear icon in the upper right side of your screen. Once the menu is open you can change the settings to what is shown in this image. You can change them to whatever you want, but this is what’s recommended.

I’m going to add a link to the SAC spreadsheet that contains these settings as well for ease of access.

[celiad]

Thanks for explaining this fix for excess browser history data! Good to have going forward.

[Nseraf]

Got a Q:

In a situation/case like this:

https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/5362139964375040
where the/a branch doesn’t end in ‘true’ EOD (black slides, no more 2D etc) but the super white ‘tissue’ that surrounds EOD and neurons end abruptly b/c the cell membrane has been damaged, do I annotate these with true end or hits edge?

[Nseraf]

also for some reason all the annotation tabs in start link (havent checked the new low specs link) are all default yellow vs green/red/orange etc.

I fixed a link but it won’t let me paste it.
https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/5670276695064576

[Nseraf]

what kind/type of cell is the red one?
https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/5848989070000128

[Nseraf]

What even is this cell? Is it glia or something else?
https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/5473684432093184

[KrzysztofKruk]

I think, both of them are glias, because I don’t see any synapses, but what kind of glia, I have no idea
If I have to guess, I’d say, the first one is a microglia and the other one - an oligodendrocyte, but it’s just based on what I think those types should look like xD.

[Nseraf]

Well I knew glia are/were cool, but daaayyyyuuummn xD I could proofread those bad boys all day every day lol, I kinda thought of -mouse retina- glia as muller cells (only) lolxD

[KrzysztofKruk]

If I’m not mistaken, Muller cells are just glia characteristic for retina, but all the other members of the gang also hang up in there (as well as in all the other parts of the nervous system). After all, something has to take care of all the neurotransmitters in the synapses (astrocytes), produce myelin (oligodendrocytes), remove dying neurons and foreign proteins (microglia), etc.
To bo precise, all those things are also done by the Muller cells, as well as giving a stable structure for the retina, but the other glia are probably just more specialized in their tasks.

[Nseraf]

Yeah ik, it was just a ‘black swan’ moment (you know the old, Europeans pre-colonial age used to think that all swans were white only and then they went to Australia and saw black swans for the first time ever). Kinda same I thought that in retinal samples all glia were muller(s). It makes sense that (all) the other kind(s) may be there and we never saw them in EW1 dataset b/c they may not have been in the dataset and/or we didn’t trace glia etc, just like we’re now seeing bp cells w/ CBs and axons coming from cells outside of the dataset/retina b/c we also didnt trace those in EW1 and so on. It’s way cool to be able to see all that stuff we wouldnt/couldnt in EW1.

[Nseraf]

Finished the newest batch of SB cells, did AJ’s WIPs as requested by AJ and did the WIPs that have been in “WIP” status since Feb/March 2025, I figured if someone took a cell and hasn’t worked at all on it for 4-5 months they’re most likely not coming back to finish it, apologies if I was wrong. There is still 1 WIP left, but that row/cell was claimed mid-May so I didn’t complete it, if it’s still in WIP status by mid-end August I’ll do it then.

[Nseraf]

what is this in the 2D? It’s causing the worst misalignment I’ve ever seen, EW, FW, BANC and/or EW2 combined.

It’s several slides in depth.

https://spelunker.cave-explorer.org/#!middleauth+https://global.daf-apis.com/nglstate/api/v1/6612930748481536

(I’m undoing mega mergers and I came upon it).

[celiad]

I would annotate these as “Can’t Fix” since they aren’t true ends but also don’t hit the dataset edge.