Transformer-based 3D U-Net with attention blocks for Kidney Tumor Segmentation

Bohdan Petryshak, Dmytro Fishman, Danylo Kolinko, Igor Krashenyi

Keywords: segmentation, attention block, transformer

Abstract: Here we describe our envisioned solution to The 2021 Kidney and Kidney Tumor Segmentation Challenge which constitutes training the original U-Net architecture with a wide range of modifications, including transformers, residual connections, and attention blocks. Once quantitative results are obtained we are happy to modify the present paper.