search

Rubics-Xuan / TransBTS

This repo provides the official code for : 1) TransBTS: Multimodal Brain Tumor Segmentation Using Transformer (https://arxiv.org/abs/2103.04430) , accepted by MICCAI2021. 2) TransBTSV2: Towards Better and More Efficient Volumetric Segmentation of Medical Images(https://arxiv.org/abs/2201.12785).
Apache License 2.0
390 stars 81 forks source link

About 3D transformer #7

Closed vv123-star closed 3 years ago

vv123-star commented 3 years ago

Hello, I run the tranBTS code for 3D medical images, the shape of the generated attention weight is (Batch_size,num_heads, n_patches,n_patches ), is this not applicable to 3D images? Thank you in advance!

Rubics-Xuan commented 3 years ago

Thanks for your questions. In fact, I am sure what is your question. After downsampling, we reshape each volume into a feature vector instead of patch. I think you might be wondering is the way of patch-style applicable to 3D images. You can check for more details from our paper.