-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
issue with lesion dectected in cerebellum #37
Comments
Hi Simon,
Not sure what is causing it, but LINDA relies heavily on asymmetries. Also,
it runs multiple iterative registrations to the template, and things can go
wrong there, too. It was not trained with any cerebellar lesion, so it is
just something during processing. So I would not spend much time on it
unless you are interested on exactly what causes it for methodological
reasons. In that case you could look at the features produced,
registrations, etc.
I would just ignore them and remove that cluster from the mask, it should
be easy on itksnap, and it's good that this is happening on just few
subjects with old data.
Dorian
…On Fri, Dec 22, 2023, 5:14 PM Simon Thibault ***@***.***> wrote:
Hello,
I compared old hand lesion drawings and automated lesion drawing. LINDA
seems to draw lesions in the cerebellum that does not exist for a couple of
my subjects. I am not sure what it is causing this issue.
The automated lesion drawings is in red, while the hand lesion drawing is
in blue on one relevant example.
The scans are quite old for most of them (2004 - 2006) so I don't know if
something in the scan could cause the issue (I am sharing the T1w as well).
image.png (view on web)
<https://github.com/dorianps/LINDA/assets/67144860/8974eec3-9ac3-4f59-bd2b-acaf619c3f4f>
image.png (view on web)
<https://github.com/dorianps/LINDA/assets/67144860/c60459a1-303a-4e1d-8497-acdcc0dc89c1>
—
Reply to this email directly, view it on GitHub
<#37>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACFJU7ISYYPOU5WX6RXGYBLYKYA33AVCNFSM6AAAAABBAJ3GSSVHI2DSMVQWIX3LMV43ASLTON2WKOZSGA2TINJUGQZTEOI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
Thanks for the recommandation. I will do that. What do you think also about small cortical lesions (less than 10 voxels) that LINDA may detect sometimes and that are apart of the bigger lesion. Do you think this could be an image artefact as well in some cases? Is it worth asking a neurologist checking LINDA lesion drawings and manually correct the lesions if needed? |
No, I would never use LINDA drawings as a basis to suspect lesions. It may
only be useful to make you look somewhere and if you suspect a lesion from
what you see, then you can ask a radiologist. By itself LINDA is a dumb
tool with no clinical value. It was created only to help drawing lesions
more quickly. It also fails more often on small lesions than larger ones.
So LINDA should be the elf that does some work, but you should always check
carefully if it draws the lesion the way you would have drawn it yourself.
All segmentations should be manually QCed at minimum and, if necessary,
edited.
…On Sat, Dec 23, 2023, 11:11 AM Simon Thibault ***@***.***> wrote:
Thanks for the recommandation. I will do that. What do you think also
about small cortical lesions (less than 10 voxels) that LINDA may detect
sometimes and that are apart of the bigger lesion. Do you think this could
be an image artefact as well in some cases? Is it worth asking a
neurologist checking LINDA lesion drawings and manually correct the lesions
if needed?
—
Reply to this email directly, view it on GitHub
<#37 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACFJU7MHLV5K44XXYKX42ZLYK37C7AVCNFSM6AAAAABBAJ3GSSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNRYGMZDGMBVG4>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hello,
I compared old hand lesion drawings and automated lesion drawing. LINDA seems to draw lesions in the cerebellum that does not exist for a couple of my subjects. I am not sure what it is causing this issue.
The automated lesion drawings is in red, while the hand lesion drawing is in blue on one relevant example.
The scans are quite old for most of them (2004 - 2006) so I don't know if something in the scan could cause the issue (I am sharing the T1w as well).
The text was updated successfully, but these errors were encountered: