search

med-air / DLTTA

[IEEE TMI'22] DLTTA: Dynamic Learning Rate for Test-time Adaptation on Cross-domain Medical Images
16 stars 2 forks source link

On OCT dataset #6

Open zenglingqi647 opened 1 year ago

zenglingqi647 commented 1 year ago

Hi!

In test_oct.sh, it has

--vimg_path="$DATAROOT"/cirrus/image/ \ --vlabel_path="$DATAROOT"/cirrus/label/ \ --sub_name="$DATAROOT"/cirrus/cirrus_name.txt \

However, I was wondering how to obtain the cirrus data. The data I obtained from the website has names like ms21_spectralis_macula_v1_s1_R.mat, and hc01_spectralis_macula_v1_s1_R.vol.

What's more, the data downloaded are in .mat format in delineation. Do we need to process them or can we directly use them for the model?

HongzhengYang commented 1 year ago

Hi, following the ATTA paper, the cirrus data should be processed as PNG/JPG images. The label should be also transformed into the grayscale images.