jackyko1991/vnet-tensorflow

question about data

Opened this issue · 2 comments

Hi,
I am using this code for brain tumor segmention(BRATS dataset),and the result is not satisfying.In another issue you mentioned the data you uesd,i'm wondering if you can share the data so i can have a try.
Thanks

You may need to adjust the config.json for image patch size (for head data, i recommend to be around 25625632 with a 11GB GPU)

Also preprocessing pipeline need to be adjusted with different normalization methods, for LITS dataset you may try manual normalization with HU value [0,300], this is call soft tissue window and better visualizing specific tissues for both human and the machine

Several factors like decay rate, initial learning rate would also affect final result