首页> 外文期刊>Journal of visual communication & image representation >Towards better semantic consistency of 2D medical image segmentation
【24h】

Towards better semantic consistency of 2D medical image segmentation

机译:Towards better semantic consistency of 2D medical image segmentation

获取原文
获取原文并翻译 | 示例
           

摘要

The latest deep neural networks for medical segmentation typically utilize transposed convolutional filtersand atrous convolutional filters for spatial restoration and larger receptive fields, leading to dilution andinconsistency of visual semantics. To address such issues, we propose a novel attentional up-concatenationstructure to build an auxiliary path for direct access to multi-level features. In addition, we employ a newstructural loss to bring better morphological awareness and reduce the segmentation flaws caused by thesemantic inconsistencies. Thorough experiments on the challenging optic cup/disc segmentation, cellularsegmentation and lung segmentation tasks were performed to evaluate the proposed methods. Further ablationanalysis demonstrated the effectiveness of the different components of the model and illustrated its efficiency.The proposed methods achieved the best performance and speed compared to the state-of-the-art models inthree tasks on seven public datasets, including DRISHTI-GS, RIM-r3, REFUGE, MESSIDOR, TNBC, GlaS andLUNA.

著录项

获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号