School of Information Science and Engineering, East China University of Science and Technology, Shanghai, China.
Department of Pancreas Surgery, Huashan Hospital Fudan University, Shanghai, China.
Med Phys. 2022 Aug;49(8):5523-5536. doi: 10.1002/mp.15708. Epub 2022 Jun 1.
Pancreatic cystic neoplasms (PCNs) are relatively rare neoplasms and difficult to be classified preoperatively. Ordinary deep learning methods have great potential to provide support for doctors in PCNs classification but require a quantity of labeled samples and exact segmentation of neoplasm. The proposed metric learning-based method using graph neural network (GNN) aims to overcome the limitations brought by small and imbalanced dataset and get fast and accurate PCNs classification result from computed tomography (CT) images.
The proposed framework applies GNN. GNNs perform well in fusing information and modeling relational data and get better results on dataset with small size. Based on metric learning strategy, model learns distance from the data. The similarity-based algorithm enhances the classification performance, and more characteristic information is found. We use a convolutional neural network (CNN) to extract features from given images. Then GNN is used to find the similarity between each two feature vectors and complete the classification. Several subtasks consisting of randomly selected images are established to improve generalization of the model. The experiments are carried out on the dataset provided by Huashan Hospital. The dataset is labeled by postoperative pathological analysis and contains region of interest (ROI) information calibrated by experts. We set two tasks based on the dataset: benign or malignant diagnosis of PCNs and classification of specific types.
Our model shows good performance on the two tasks with accuracies of 88.926% and 74.497%. The comparison of different methods' F1 scores in the benign or malignant diagnosis shows that the proposed GNN-based method effectively reduces the negative impact brought by imbalanced dataset, which is also verified by the macroaverage comparison in the four-class classification task.
Compared with existing models, the proposed GNN-based model shows better performance in terms of imbalanced dataset with small size while reducing labeling cost. The result provides a possibility for its application into the computer-aided diagnosis of PCNs.
胰腺囊性肿瘤（PCN）是相对罕见的肿瘤，术前难以分类。普通的深度学习方法在 PCN 分类方面具有很大的潜力，但需要大量的标记样本和肿瘤的精确分割。本研究提出了一种基于度量学习的图神经网络（GNN）方法，旨在克服小数据集和不平衡数据集带来的限制，从 CT 图像中快速准确地对 PCN 进行分类。
本研究提出的框架应用了 GNN。GNN 在融合信息和建模关系数据方面表现出色，在小数据集上能取得更好的效果。基于度量学习策略，模型学习数据之间的距离。基于相似性的算法提高了分类性能，发现更多特征信息。我们使用卷积神经网络（CNN）从给定的图像中提取特征。然后，GNN 用于寻找每个两个特征向量之间的相似性，并完成分类。建立了几个由随机选择的图像组成的子任务，以提高模型的泛化能力。实验是在华山医院提供的数据集上进行的。该数据集由术后病理分析标记，并包含专家标定的感兴趣区域（ROI）信息。我们基于该数据集设置了两个任务：PCN 的良性或恶性诊断和特定类型的分类。
我们的模型在这两个任务上都表现出了良好的性能，准确率分别为 88.926%和 74.497%。不同方法在良性或恶性诊断中的 F1 分数比较表明，基于 GNN 的方法有效地减少了不平衡数据集带来的负面影响，这在四类分类任务的宏平均比较中也得到了验证。
与现有模型相比，所提出的基于 GNN 的模型在小数据集和不平衡数据集方面表现出了更好的性能，同时降低了标记成本。研究结果为其在 PCN 的计算机辅助诊断中的应用提供了可能性。
