与惰性 NHL 患者相比，侵袭性非霍奇金淋巴瘤 (NHL) 患者接受不同的治疗策略。然而，基于正电子发射断层扫描 (PET) 或计算机断层扫描 (CT) 图像的目视检查来估计 NHL 侵袭性具有挑战性。由于弥漫性大 B 细胞淋巴瘤 (DLBCL) 和滤泡性淋巴瘤 (FL) 分别是最典型、最显着的侵袭性和惰性 NHL，因此本研究旨在开发一种人工智能模型，以在 PET/CT 中区分 DLBCL 和 FL图像作为应对这一挑战的第一步。我们提出了一种混合少镜头多实例学习模型来预测 NHL 的攻击性。首先，采用基于旋转的自监督学习 (SSL) 在大规模、公开的 CT 图像数据集上训练编码器。其次，通过将深层特征与 PET 和 CT 模式的放射组学特征相结合，获得每个 NHL 病变的混合实例级特征。第三，实例级特征转换为包级（或患者级）表示。最后，通过小样本学习将袋级表示输入基于距离的分类器，以预测 NHL 攻击性。我们的模型实现了 0.751 ± 0.008 的准确度、0.787 ± 0.012 的灵敏度、0.715 ± 0.013 的特异性、F1 -袋子水平的分数为 0.753 ± 0.009，曲线下面积 (AUC) 为 0.795 ± 0.009。它优于使用放射组学特征、随机森林进行特征选择以及支持向量机 (SVM) 作为分类器的典型对应方法。三个对应项的准确度分别为 0.714 ± 0.023、0.705 ± 0.008 和 0.698 ± 0.008。此外，SSL训练数据集（深部病变）和任务（旋转）的设置、混合CT和放射组学PET特征、最大池层策略以及基于距离的分类器生成最佳模型。实例学习模型可以预测 PET/CT 图像中淋巴瘤的侵袭性，并可能成为确定治疗策略的潜在工具。混合特征以及 SSL、小样本学习和弱监督学习的组合是该模型的两个强大支柱，这些可以扩展到样本有限和注释不完整的其他医疗应用。版权所有 © 2023。由 Elsevier B.V. 出版。

Patients with aggressive non-Hodgkin lymphoma (NHL) undergo distinct therapy strategies compared with indolent NHL patients. However, it is challenging to estimate NHL aggressiveness based on visual inspection of positron emission tomography (PET) or computed tomography (CT) images. Since diffuse large B-cell lymphoma (DLBCL) and Follicular lymphoma (FL) are the most typical and dominant aggressive and indolent NHL, respectively, this study aims to develop an artificial-intelligence-enabled model to distinguish DLBCL from FL in PET/CT images as the first step to tackle this challenge.We propose a hybrid few-shot multiple-instance learning model to predict the aggressiveness of the NHL. First, rotation-based self-supervision learning (SSL) has been employed to train the encoder on a large-scale, publicly available CT image dataset. Second, hybrid instance-level features are obtained for each NHL lesion by combining deep features with the radiomics features from both PET and CT modalities. Third, instance-level features are transformed into bag-level (or patient-level) representations. Finally, bag-level representations are fed into a distance-based classifier through few-shot learning to predict NHL aggressiveness.Our model achieves an accuracy of 0.751 ± 0.008, a sensitivity of 0.787 ± 0.012, a specificity of 0.715 ± 0.013, an F1-score of 0.753 ± 0.009, and an area under the curve (AUC) of 0.795 ± 0.009 at the bag level. It outperforms the typical counterparts that use the radiomic features, random forest for feature selection, and support vector machines (SVMs) as classifiers. The three counterparts yield accuracies of 0.714 ± 0.023, 0.705 ± 0.008, and 0.698 ± 0.008, respectively. Moreover, settings of the SSL training dataset (Deep lesion) and task (rotation), hybrid CT and radiomic PET features, the pool-layer strategy of maximum, and distance-based classifier generate the best model.A hybrid few-shot multiple-instance learning model can predict lymphoma aggressiveness in PET/CT images and could be a potential tool for determining therapy strategies. Hybrid features and the combination of SSL, few-shot learning, and weakly supervised learning are the two powerful pillars of the model, and these can be expanded to other medical applications with limited samples and incomplete annotations.Copyright © 2023. Published by Elsevier B.V.