深度卷积神经网络应用于肝脏影像报告和数据系统(LI-RADS)2014版类别分类:一项初步研究
本研究利用314个肝脏病灶(163个CT、151个MRI)的多期图像开发深度卷积神经网络模型,按LI-RADS 2014版进行LR-1至LR-5分类。迁移学习模型表现优于自制模型,内部测试集总体准确率60.4%,外部验证准确率41.2%(EXT-CT)和47.7%(EXT-MR)。结论提示CNN从较小数据集分配LI-RADS类别具有可行性,但模型开发与验证存在挑战。
要点
- 迁移学习模型内部测试总体准确率60.4%,LR-1/2、LR-3、LR-4、LR-5的AUROC分别为0.85、0.90、0.63、0.82。
- 外部验证EXT-CT总体准确率41.2%,各AUROC为0.70、0.66、0.60、0.76;EXT-MR总体准确率47.7%,各AUROC为0.88、0.74、0.69、0.79。
- 研究提示CNN从较小数据集分配LI-RADS类别具有可行性,但模型开发与验证存在挑战。
核心结论
- 研究显示CNN从相对较小的数据集分配LI-RADS类别具有可行性,但突出模型开发与验证的挑战。
研究设计与样本
- 使用预存在数据集共314个肝脏病灶(163个CT、151个MRI),由两名LI-RADS指导委员会成员共识分配直径和LI-RADS类别(LR-1至LR-5)。
- 开发两种CNN:预训练网络输入三期图像(迁移学习训练)和自制网络输入四期图像(从头训练)。
- 数据集随机分为训练集、验证集和内部测试集(70:15:15)。
- 内部测试集表现更好的模型用两个外部数据集验证:EXT-CT 68个病灶和EXT-MR 44个病灶。
诊断标准与关键数值
- 迁移学习模型内部测试总体准确率60.4%,LR-1/2、LR-3、LR-4、LR-5的AUROC分别为0.85、0.90、0.63、0.82。
- EXT-CT总体准确率41.2%,LR-1/2、LR-3、LR-4、LR-5的AUROC分别为0.70、0.66、0.60、0.76。
- EXT-MR总体准确率47.7%,LR-1/2、LR-3、LR-4、LR-5的AUROC分别为0.88、0.74、0.69、0.79。
临床应用与阅片要点
- 原文未据此给出推荐。
局限性
- 原文指出模型开发与验证存在挑战。
- 原文称数据集相对较小。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
PURPOSE: To develop a deep convolutional neural network (CNN) model to categorize multiphase CT and MRI liver observations using the liver imaging reporting and data system (LI-RADS) (version 2014).
METHODS: A pre-existing dataset comprising 314 hepatic observations (163 CT, 151 MRI) with corresponding diameters and LI-RADS categories (LR-1-5) assigned in consensus by two LI-RADS steering committee members was used to develop two CNNs: pre-trained network with an input of triple-phase images (training with transfer learning) and custom-made network with an input of quadruple-phase images (training from scratch). The dataset was randomly split into training, validation, and internal test sets (70:15:15 split). The overall accuracy and area under receiver operating characteristic curve (AUROC) were assessed for categorizing LR-1/2, LR-3, LR-4, and LR-5. External validation was performed for the model with the better performance on the internal test set using two external datasets (EXT-CT and EXT-MR: 68 and 44 observations, respectively).
RESULTS: The transfer learning model outperformed the custom-made model: overall accuracy of 60.4% and AUROCs of 0.85, 0.90, 0.63, 0.82 for LR-1/2, LR-3, LR-4, LR-5, respectively. On EXT-CT, the model had an overall accuracy of 41.2% and AUROCs of 0.70, 0.66, 0.60, 0.76 for LR-1/2, LR-3, LR-4, LR-5, respectively. On EXT-MR, the model had an overall accuracy of 47.7% and AUROCs of 0.88, 0.74, 0.69, 0.79 for LR-1/2, LR-3, LR-4, LR-5, respectively.
CONCLUSION: Our study shows the feasibility of CNN for assigning LI-RADS categories from a relatively small dataset but highlights the challenges of model development and validation.
引用
- Yamashita R, Mittendorf A, Zhu Z, et al. Deep convolutional neural network applied to the liver imaging reporting and data system (LI-RADS) version 2014 category classification: a pilot study. Abdom Radiol (NY). 2020. PMID: 31696269. doi:10.1007/s00261-019-02306-7
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。