A unified uncertainty network for tumor segmentation using uncertainty cross entropy loss and prototype similarity

A unified uncertainty network for tumor segmentation using uncertainty cross entropy loss and prototype similarity
复制标题

使用不确定性交叉熵损失和原型相似性进行肿瘤分割的统一不确定性网络

DOI:
10.1016/j.knosys.2022.108739
复制
发表时间:
2022-04
影响因子:
8.8
通讯作者:
Tianyu Shi
Tianyu Shi
中科院分区:
计算机科学1区
文献类型:
--
作者:
Zhaoshuo Diao;Huiyan Jiang;Tianyu Shi

文献摘要

参考文献

相似文献

不确定性估计和分布外(OOD)检测是基于深度卷积神经网络的肿瘤分割任务中具有实际意义的主题。我们提出了一个统一的不确定性分割网络,用一个网络来处理这两个任务。提出了不确定性交叉熵损失,引导网络直接输出每个像素的预测不确定性,而不是在预测阶段执行多次。我们扩展的不确定性的情况下,解决OOD样本检测问题。基于原型相似性估计实例级不确定性。我们在四个数据集上进行像素级不确定性实验和OOD检测实验。实验结果表明,我们提出的方法是更适合在肿瘤分割的不确定性估计比现有的方法。我们提出的方法只需要修改网络输出和损失函数,不需要执行网络多次估计不确定性。此外,我们的方法在肿瘤分割方面表现出更好的性能。
Uncertainty estimation and out-of-distribution (OOD) detection are topics of practical significance in deep convolutional neural network-based tumor segmentation tasks. We propose a unified uncertainty segmentation network to handle these two tasks with a single network. The uncertainty cross entropy loss is proposed to guide the network to directly output the prediction uncertainty of each pixel instead of executing several times in the prediction phase. We extend the uncertainty to the case level to address OOD sample detection problems. Case-level uncertainty is estimated based on prototype similarity. We perform pixel-level uncertainty experiments and OOD detection experiments on four datasets. The experimental results show that our proposed method is more suitable for uncertainty estimation in tumor segmentation than existing methods. Our proposed method only needs to modify the network output and loss function and does not need to execute the network multiple times when estimating uncertainty. Moreover, our method shows improved performance in tumor segmentation.
DOI: 10.1109/tmi.2020.3048055
发表时间: 2021-10-01
影响因子: 10.6
作者:
Nath, Vishwesh;Yang, Dong;Roth, Holger R.
通讯作者: Roth, Holger R.
DOI: 10.1109/tmi.2018.2845918
发表时间: 2018-12-01
影响因子: 10.6
作者:
Li, Xiaomeng;Chen, Hao;Heng, Pheng-Ann
通讯作者: Heng, Pheng-Ann
DOI: 10.1016/j.compbiomed.2021.104418
发表时间: 2021-08-21
影响因子: 7.7
作者:
Abdar, Moloud;Samami, Maryam;Nahavandi, Saeid
通讯作者: Nahavandi, Saeid
DOI: --
发表时间: 2020-07
期刊: --
影响因子: --
作者:
Lingkai Kong;Jimeng Sun;Chao Zhang
通讯作者: Lingkai Kong;Jimeng Sun;Chao Zhang
DOI: --
发表时间: 2016-12
期刊: --
影响因子: --
作者:
Balaji Lakshminarayanan;A. Pritzel;C. Blundell
通讯作者: Balaji Lakshminarayanan;A. Pritzel;C. Blundell