UD-MIL: Uncertainty-Driven Deep Multiple Instance Learning for OCT Image Classification
UD-MIL: Uncertainty-Driven Deep Multiple Instance Learning for OCT Image Classification
复制标题
DOI：
10.1109/jbhi.2020.2983730
复制
发表时间：
2020-12-01
影响因子：
7.7
通讯作者：
Heng, Pheng-Ann
中科院分区：
文献类型：
--
作者：
Wang, Xi;Tang, Fangyao;Heng, Pheng-Ann
深度学习在光学相干断层扫描 (OCT) 图像分类任务中取得了显着的成功，并提供了大量标记的 B 扫描图像。然而，获得如此细粒度的专家注释通常相当困难且昂贵。如何利用体积级标签来开发强大的分类器非常有吸引力。在本文中，我们提出了一种具有不确定性估计的弱监督深度学习框架，以解决 OCT 图像中与黄斑相关的疾病分类问题，并且具有唯一可用的体积级标签。首先，通过使用所提出的不确定性驱动的深度多实例学习方案来迭代细化基于卷积神经网络（CNN）的实例级分类器。据我们所知，我们是第一个将不确定性评估机制纳入多实例学习（MIL）中以训练鲁棒实例分类器的人。分类器能够检测可疑的异常实例，同时抽象出相应的具有高表示能力的深度嵌入。其次，循环神经网络（RNN）将同一包中的实例特征作为输入，并通过考虑单独的本地实例信息和全局聚合的包级表示来生成最终的包级预测。为了进行更全面的验证，我们从不同的设备和成像协议中构建了两个大型糖尿病黄斑水肿 (DME) OCT 数据集来评估我们方法的功效，这些数据集分别由来自 274 名患者的 1,396 卷中的 30,151 个 B 扫描（Heidelberg-DME 数据集）和来自 490 名患者的 3,248 卷中的 38,976 个 B 扫描（Triton-DME 数据集）组成。我们将所提出的方法与最先进的方法进行了比较，并通过实验证明我们的方法优于替代方法，在 Heidelberg-DME 上实现了 95.1%、0.939 和 0.990 的音量级精度、F1 分数和接收者操作特征曲线下面积 (AUC)，在 Triton-DME 上分别实现了 95.1%、0.935 和 0.986。此外，所提出的方法还在另一个公共年龄相关性黄斑变性 OCT 数据集上产生了有竞争力的结果，表明作为临床实践中有效筛查工具的巨大潜力。
Deep learning has achieved remarkable success in the optical coherence tomography (OCT) image classification task with substantial labelled B-scan images available. However, obtaining such fine-grained expert annotations is usually quite difficult and expensive. How to leverage the volume-level labels to develop a robust classifier is very appealing. In this paper, we propose a weakly supervised deep learning framework with uncertainty estimation to address the macula-related disease classification problem from OCT images with the only volume-level label being available. First, a convolutional neural network (CNN) based instance-level classifier is iteratively refined by using the proposed uncertainty-driven deep multiple instance learning scheme. To our best knowledge, we are the first to incorporate the uncertainty evaluation mechanism into multiple instance learning (MIL) for training a robust instance classifier. The classifier is able to detect suspicious abnormal instances and abstract the corresponding deep embedding with high representation capability simultaneously. Second, a recurrent neural network (RNN) takes instance features from the same bag as input and generates the final bag-level prediction by considering the individually local instance information and globally aggregated bag-level representation. For more comprehensive validation, we built two large diabetic macular edema (DME) OCT datasets from different devices and imaging protocols to evaluate the efficacy of our method, which are composed of 30,151 B-scans in 1,396 volumes from 274 patients (Heidelberg-DME dataset) and 38,976 B-scans in 3,248 volumes from 490 patients (Triton-DME dataset), respectively. We compare the proposed method with the state-of-the-art approaches, and experimentally demonstrate that our method is superior to alternative methods, achieving volume-level accuracy, F1-score and area under the receiver operating characteristic curve (AUC) of 95.1%, 0.939 and 0.990 on Heidelberg-DME and those of 95.1%, 0.935 and 0.986 on Triton-DME, respectively. Furthermore, the proposed method also yields competitive results on another public age-related macular degeneration OCT dataset, indicating the high potential as an effective screening tool in the clinical practice.