Multi-planar 3D breast segmentation in MRI via deep convolutional neural networks
作者：
Highlights：
• A novel approach for 3D fully automatic and accurate breast tissue segmentation from MRI data is proposed.
• Our approach avoids parameters explosion by using a suitably modified 2D deep U Net CNN model on 3D data.
• The method relies on a multi planar combination of deep CNNs by a suitable projection fusing approach.
• The proposal was evaluated by using two different datasets on a total of 109 MRI/DCE MRI studies with histopathologically proven lesions.
• The proposal demonstrated to be able to cope with a multi protocol requirement.
摘要
•A novel approach for 3D fully automatic and accurate breast tissue segmentation from MRI data is proposed.•Our approach avoids parameters explosion by using a suitably modified 2D deep U Net CNN model on 3D data.•The method relies on a multi planar combination of deep CNNs by a suitable projection fusing approach.•The proposal was evaluated by using two different datasets on a total of 109 MRI/DCE MRI studies with histopathologically proven lesions.•The proposal demonstrated to be able to cope with a multi protocol requirement.
论文关键词：MRI,Breast,Segmentation,Convolutional neural networks,U-Net
论文评审过程：Received 22 November 2018, Revised 18 November 2019, Accepted 19 December 2019, Available online 23 December 2019, Version of Record 7 January 2020.
论文官网地址：https://doi.org/10.1016/j.artmed.2019.101781
