The contrast-enhanced MRI can be substituted by unenhanced MRI in identifying and automatically segmenting primary nasopharyngeal carcinoma with the aid of deep learning models: An exploratory study in large-scale population of endemic area
The contrast-enhanced MRI can be substituted by unenhanced MRI in identifying and automatically segmenting primary nasopharyngeal carcinoma with the aid of deep learning models: An exploratory study in large-scale population of endemic area
复制标题
在原发性鼻咽部的识别和自动分割方面，对比增强MRI可以被非增强MRI替代
DOI：
10.1016/j.cmpb.2022.106702
复制
发表时间：
2022-02-25
影响因子：
6.1
通讯作者：
Ke, Liangru
中科院分区：
文献类型：
--
作者：
Deng, Yishu;Li, Chaofeng;Ke, Liangru
背景和目标：在临床环境中，并非所有病例都需要使用造影剂，并且在深度学习模型开发的序列选择方面尚未达成共识，因此我们的目标是在大规模队列中探索在深度学习模型的帮助下，对比增强磁共振成像（ceMRI）是否可以在鼻咽癌（NPC）的识别和分割中取代。方法：共有4478名合格的个体被随机分为训练集、验证集和测试集，分别使用轴向T1加权成像（T1 WI）、T2 WI或增强T1 WI（T1 WIC）图像开发了自约束3D DenseNet和V-Net模型。使用卡方检验比较模型之间NPC和良性增生的鉴别诊断性能。分割评价指标，包括骰子相似系数（DSC）和平均表面距离（ASD），使用配对学生t检验在T1 WIC和T1 WI或T2 WI模型或M_T1/T2（来自T1 WI和T2 WI模型的恶性区域的合并输出）之间进行比较。所有模型在区分NPC与良性增生方面表现出相似的令人满意的诊断性能，在NPC的所有T阶段中均达到99.00%以上的总体准确度。T_1WIC模型的DSC和ASD与M_T_1/T_2模型相似（DSC，0.768 +/- 0.070 vs 0.764 +/- 0.070; ASD，1.573 ± 10.954 mm vs 1.626 ± 10.975 mm 1.626 ± 0.975 mm vs 1.573 ± 0.954 mm，所有p > 0.0167），但与T1 WI和T2 WI模型相比，DSC显著升高，ASD显著降低（DSC，0.759 +/- 0.065或0.755 +/- 0.071; ASD，分别为1.661 +/- 0.898 mm或1.722 +/- 1.133 mm，所有p < 0.01）。T1 WIC模型组与M_T1/T2组的平均DSCs和ASD差异均无统计学意义。(c)2022爱思唯尔有限公司版权所有。
Background and objectives: Administration of contrast is not desirable for all cases in clinical setting, and no consensus in sequence selection for deep learning model development has been achieved, thus we aim to explore whether contrast-enhanced magnetic resonance imaging (ceMRI) can be substituted in the identification and segmentation of nasopharyngeal carcinoma (NPC) with the aid of deep learning models in a large-scale cohort.Methods: A total of 4478 eligible individuals were randomly split into training, validation and test sets, and self-constrained 3D DenseNet and V-Net models were developed using axial T1-weighted imaging (T1WI), T2WI or enhanced T1WI (T1WIC) images separately. The differential diagnostic performance between NPC and benign hyperplasia were compared among models using chi-square test. Segmentation evaluation metrics, including dice similarity coefficient (DSC) and average surface distance (ASD), were compared using paired student's t-test between T1WIC and T1WI or T2WI models or M_T1/T2, a merged output of malignant region derived from T1WI and T2WI models.Results: All models exhibited similar satisfactory diagnostic performance in discriminating NPC from benign hyperplasia, all attaining overall accuracy over 99.00% in all T stages of NPC. And T1WIC model exhibited similar average DSC and ASD with those of M_T1/T2 (DSC, 0.768 +/- 0.070 vs 0.764 +/- 0.070; ASD, 1.573 +/- 10.954 mm vs 1.626 +/- 10.975 mm 1.626 +/- 0.975 mm vs 1.573 +/- 0.954 mm , all p > 0.0167) in primary NPC using DenseNet, but yielded a significantly higher DSC and lower ASD than either T1WI model or T2WI model (DSC, 0.759 +/- 0.065 or 0.755 +/- 0.071; ASD, 1.661 +/- 0.898 mm or 1.722 +/- 1.133 mm, respectively, all p < 0.01) in the entire test set of NPC cohort. Moreover, the average DSCs and ASDs were not statistically significant between T1WIC model and M_T1/T2 in both. (c) 2022 Elsevier B.V. All rights reserved.