mikevoets / jama16-retina-replication

JAMA 2016; 316(22) Replication Study
https://doi.org/10.1371/journal.pone.0217541
MIT License
110 stars 35 forks source link

question about eyepacs.sh #20

Closed Irish-kw closed 1 year ago

Irish-kw commented 1 year ago

HI:

I'm still not understand after using eyepacs.sh. I got bin2/pool, two folder, there is grade folder in pool (0/1/2/3/4), but only 0/1 in bin2.

which one can use for training and test ? thank you so much

mikevoets commented 1 year ago

Hi @Irish-kw,

The EyePACS data set was graded according to the International Clinical Diabetic Retinopathy scale, with no, mild, moderate, severe or proliferative severity. Our neural network however outputs only one binary prediction: moderate or worse diabetic retinopathy (referable diabetic retinopathy).

The eyepacs.sh script moves the EyePACS images labeled with 0,1 into the bin2/train/0 and bin2/test/0 folder, and images labeled with 2,3,4 are moved into the bin2/train/1 and bin2/test/1 folder, for training and testing, respectively.

Hope that clears things up!