Sato Junya, Suzuki Yuki, Wataya Tomohiro, Nishigaki Daiki, Kita Kosuke, Yamagata Kazuki, Tomiyama Noriyuki, Kido Shoji
Department of Artificial Intelligence Diagnostic Radiology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, Osaka 565-0871, Japan.
Department of Radiology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, Osaka 565-0871, Japan.
iScience. 2023 Jun 15;26(7):107086. doi: 10.1016/j.isci.2023.107086. eCollection 2023 Jul 21.
In this study, we present a self-supervised learning (SSL)-based model that enables anatomical structure-based unsupervised anomaly detection (UAD). The model employs an anatomy-aware pasting (AnatPaste) augmentation tool that uses a threshold-based lung segmentation pretext task to create anomalies in normal chest radiographs used for model pretraining. These anomalies are similar to real anomalies and help the model recognize them. We evaluate our model using three open-source chest radiograph datasets. Our model exhibits area under curves of 92.1%, 78.7%, and 81.9%, which are the highest among those of existing UAD models. To the best of our knowledge, this is the first SSL model to employ anatomical information from segmentation as a pretext task. The performance of AnatPaste shows that incorporating anatomical information into SSL can effectively improve accuracy.
在本研究中，我们提出了一种基于自监督学习（SSL）的模型，该模型能够实现基于解剖结构的无监督异常检测（UAD）。该模型采用了一种解剖感知粘贴（AnatPaste）增强工具，该工具使用基于阈值的肺部分割预训练任务，在用于模型预训练的正常胸部X光片中创建异常。这些异常类似于真实异常，有助于模型识别它们。我们使用三个开源胸部X光数据集对模型进行评估。我们的模型的曲线下面积分别为92.1%、78.7%和81.9%，在现有UAD模型中是最高的。据我们所知，这是第一个将分割的解剖信息用作预训练任务的SSL模型。AnatPaste的性能表明，将解剖信息纳入SSL可以有效提高准确率。
