BreastPathQ: After trianing the model by WSIs, the result which evaluated by Task-specific supervised fine-tuning is bad
xiaozhu0816 opened this issue · 1 comments
xiaozhu0816 commented
-
For BreastPathQ, when I used the pretrian model author released and did the finetune, I could get a similar result to your paper.
-
But I tried to pretrain the model myself, following the instructions (1. Self-supervised pretext task: Resolution sequence prediction (RSP) in WSIs). I used the best pretrain model and did the same things as before. After finetuning, the result was not good enough like paper.
-
paper:
-
my reslut:
-
During my training, 4 WSIs which are bad wsis cannot be used. But I don’t think it is an essential problem for me, because just lose hundreds of data.
srinidhiPY commented
I am really sorry for the late reply.
In our experiments for BreastPathQ, we have used *69 WSI’s* of the training
set for pretraining the self-supervised model. I think the issue in your
case may be because of 4 WSIs, which were not used.
As the number of patches that can be extracted from a given WSI
mainly depends on the tissue content, there might be a high probability
that you have lost some significant amount of data for SSL training.
…On Sun, 25 Sept 2022 at 10:34, xiaozhu0816 ***@***.***> wrote:
-
For BreastPathQ, when I used the pretrian model author released and
did the finetune, I could get a similar result to your paper.
-
But I tried to pretrain the model myself, following the instructions
(1. Self-supervised pretext task: Resolution sequence prediction (RSP) in
WSIs). I used the best pretrain model and did the same things as before.
After finetuning, the result was not good enough like paper.
-
paper:[image: image]
<https://user-images.githubusercontent.com/72501744/192129127-48496486-edf8-4a6e-a180-975e4aea5e1e.png>
-
my reslut: [image: image]
<https://user-images.githubusercontent.com/72501744/192129132-9028df2d-758f-42dd-855a-180e17b24c02.png>
-
During my training, 4 WSIs which are bad wsis cannot be used. But I
don’t think it is an essential problem for me, because just lose hundreds
of data.
—
Reply to this email directly, view it on GitHub
<#7>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AD4AXU3CCSL4736WLTUZFVLV77MMZANCNFSM6AAAAAAQU5YVKI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>