Multi-scale segmentation using deep graph cuts: Robust lung tumor delineation in MVCBCT
Multi-scale segmentation using deep graph cuts: Robust lung tumor delineation in MVCBCT
复制标题
DOI：
10.1109/isbi.2018.8363628
复制
发表时间：
2018-04
期刊：
影响因子：
--
通讯作者：
Xiaodong Wu;Zisha Zhong;J. Buatti;Junjie Bai
中科院分区：
文献类型：
--
作者：
Xiaodong Wu;Zisha Zhong;J. Buatti;Junjie Bai
随着深度学习的显著进步，深度网络在医学图像分析中的应用越来越多。在本文中，我们将多尺度分割表示为深度网络（图）中的马尔可夫随机场（MRF）能量最小化问题，可以通过在适当构造的图中计算最小s-t切割来有效准确地解决。所提出的方法的性能进行评估的肺肿瘤分割在38兆电压锥束计算机断层扫描数据集的应用。
Deep networks have been used in a growing trend in medical image analysis with the remarkable progress in deep learning. In this paper, we formulate the multi-scale segmentation as a Markov Random Field (MRF) energy minimization problem in a deep network (graph), which can be efficiently and exactly solved by computing a minimum s-t cut in an appropriately constructed graph. The performance of the proposed method is assessed on the application of lung tumor segmentation in 38 mega-voltage cone-beam computed tomography datasets.