A Multi-Task Convolutional Neural Network for Lesion Region Segmentation and Classification of Non-Small Cell Lung Carcinoma.
A Multi-Task Convolutional Neural Network for Lesion Region Segmentation and Classification of Non-Small Cell Lung Carcinoma.
复制标题
基于多任务卷积神经网络的非小细胞肺癌病变区域分割与分类
DOI：
10.3390/diagnostics12081849
复制
发表时间：
2022-07-31
期刊：
影响因子：
3.6
通讯作者：
Zhang, Sasa
中科院分区：
文献类型：
--
作者：
Wang, Zhao;Xu, Yuxin;Tian, Linbo;Chi, Qingjin;Zhao, Fengrong;Xu, Rongqi;Jin, Guilei;Liu, Yansong;Zhen, Junhui;Zhang, Sasa
关键词：
靶向治疗是治疗非小细胞肺癌的有效方法。在治疗之前，病理学家需要确认肿瘤的形态和类型，这是一项耗时和高度重复性的工作。在这项研究中，我们提出了一种基于卷积神经网络的多任务深度学习模型，用于关节癌病变区域分割和组织亚型分类，使用放大的病理组织图像。首先，构建共享的特征提取通道，提取视觉空间的抽象信息，用于联合分割和分类学习。然后，调整分割和分类任务的加权损失，以平衡多任务模型的计算偏差。我们在山东大学齐鲁医院收集的私人内部病理组织图像数据集上对我们的模型进行了评估。该方法对鳞癌(SCC)和腺癌(AD)样本的Dice相似系数分别达到93.5%和89.0%。此外，该方法对鳞癌和正常组织的分类准确率为97.8%，对AD和正常组织的分类准确率为100%。实验结果表明，我们的方法优于其他最先进的方法，在病变区域分割和亚型分类方面都表现出了良好的性能。
Targeted therapy is an effective treatment for non-small cell lung cancer. Before treatment, pathologists need to confirm tumor morphology and type, which is time-consuming and highly repetitive. In this study, we propose a multi-task deep learning model based on a convolutional neural network for joint cancer lesion region segmentation and histological subtype classification, using magnified pathological tissue images. Firstly, we constructed a shared feature extraction channel to extract abstract information of visual space for joint segmentation and classification learning. Then, the weighted losses of segmentation and classification tasks were tuned to balance the computing bias of the multi-task model. We evaluated our model on a private in-house dataset of pathological tissue images collected from Qilu Hospital of Shandong University. The proposed approach achieved Dice similarity coefficients of 93.5% and 89.0% for segmenting squamous cell carcinoma (SCC) and adenocarcinoma (AD) specimens, respectively. In addition, the proposed method achieved an accuracy of 97.8% in classifying SCC vs. normal tissue and an accuracy of 100% in classifying AD vs. normal tissue. The experimental results demonstrated that our method outperforms other state-of-the-art methods and shows promising performance for both lesion region segmentation and subtype classification.
登录
查看更多内容
影响因子：
3.5
作者：
Freitas, Nuno R.;Vieira, Pedro M.;Lima, Carlos S.
通讯作者：
Lima, Carlos S.
DOI：
10.3390/diagnostics11030528
发表时间：
2021-03-16
期刊：
Diagnostics (Basel, Switzerland)
影响因子：
--
作者：
Boumaraf S;Liu X;Wan Y;Zheng Z;Ferkous C;Ma X;Li Z;Bardou D
通讯作者：
Bardou D
影响因子：
3.8
作者：
ElTanboly, Ahmed;Ismail, Marwa;El-Azab, Magdi
通讯作者：
El-Azab, Magdi
影响因子：
3.4
作者：
Shah, Abhay;Zhou, Leixin;Wu, Xiaodong
通讯作者：
Wu, Xiaodong
影响因子：
3.4
作者：
Amini, Mehdi;Hajianfar, Ghasem;Zaidi, Habib
通讯作者：
Zaidi, Habib