Kidney tumor segmentation from computed tomography images using DeepLabv3+ 2.5D model

作者：

Highlights：

• A new computational method for kidney tumor segmentation in CT images is proposed.

• Proposed method uses the KiTS19 dataset consisting of 210 CTs.

• The method uses DeepLabv3+ 2.5D with DPN-131 encoder and digital image processing.

• The method achieved a Dice of 85.17%, gaining a prominent place in the literature.

摘要

•A new computational method for kidney tumor segmentation in CT images is proposed.•Proposed method uses the KiTS19 dataset consisting of 210 CTs.•The method uses DeepLabv3+ 2.5D with DPN-131 encoder and digital image processing.•The method achieved a Dice of 85.17%, gaining a prominent place in the literature.

论文关键词：Computed tomography,Deep learning,Kidney cancer,Kidney tumor segmentation,Medical images

论文评审过程：Received 24 March 2021, Revised 21 October 2021, Accepted 20 November 2021, Available online 20 December 2021, Version of Record 24 December 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116270