Computation of advanced lesion-wise metrics

[rcorredorj]

Hi guys,

https://github.com/Inria-Visages/Anima-Public/blob/c5c5acb0c97360905229028528f5da8453fc900f/Anima/segmentation/validation_tools/segmentation_performance_analyzer/animaSegPerfCAnalyzer.cxx#L73

Is this condition ok? I think it should be:

if(bAdvancedEvaluation && m_uiNbLabels > 1)

All the best!

RaC

[ocommowi]

Well that is a good question. We are going to verify but I think the line you are refering to is right in its current version. If I remeber well, this switch activates advanced evaluation i.e. looking at labels independently. Because the background is counted as one label as well, this makes a binary image have two labels. Thus the > 2

[rcorredorj]

Hi @ocommowi, Thanks for your reply. Actually, that's exactly the point. If there are counted as two labels, the condition should be > 1 or >= 2. Leaving > 2 get's into an error. I did the change and it works fine ;-). Btw, I may contact you via email to ask you a few other questions. Cheers Ricardo RaC

…

On Thu, Feb 25, 2021 at 5:57 PM Olivier Commowick ***@***.***> wrote: Well that is a good question. We are going to verify but I think the line you are refering to is right in its current version. If I remeber well, this switch activates advanced evaluation i.e. looking at labels independently. Because the background is counted as one label as well, this makes a binary image have two labels. Thus the > 2 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#75 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AATQVBQ3QE3AXTW5Q7QV55LTAZ6P7ANCNFSM4XWF4QZA> .

[ocommowi]

After looking into it, I agree it can be changed to > 1 . I am pushing this to master tonight. Thanks !