本研究的目的是开发一种基于放射组学的算法，用于在 CT 上识别小胰腺神经内分泌肿瘤（PanNET），并评估其在手动和自动分割中的稳健性，探索自动筛查的可行性。病理证实的 T1 期 PanNET 的患者和健康患者回顾性地确定了接受双相 CT 成像的对照。对胰腺和肿瘤进行手动分割，然后使用预训练的神经网络生成自动胰腺分割。分别在动脉期和静脉期的两个分割体积中独立提取了总共 1223 个放射组学特征。最终选择了十个特征来训练分类器来识别 PanNET 和控件。该队列被分为训练集和测试集，并使用接受者操作特征曲线 (AUC) 下面积、特异性和敏感性来评估分类器的性能，并与两名对诊断不知情的放射科医生进行比较。共有 135 名患者使用 142 个 PanNET ，并包括 135 名健康对照者。共有 168 名女性和 102 名男性，平均年龄为 55.4 ± 11.6（标准差）岁（范围：20-85 岁）。 PanNET 尺寸中位数为 1.3 厘米（Q1，1.0；Q3，1.5；范围：0.5-1.9）。动脉期 LightGBM 模型在测试集中取得了最佳性能，灵敏度为 90%（95% 置信区间 [CI]：80-98），特异性为 76%（95% CI：62-88），AUC 为 0.87（ 95 % CI: 0.79-0.94)。使用自动分割的特征，该模型的 AUC 为 0.86（95% CI：0.79-0.93）。相比之下，两位放射科医生使用动脉期 CT 图像实现了平均 50% 的灵敏度和 100% 的特异性。放射组学特征可识别小型 PanNET，在使用自动分割提取时具有稳定的性能。这些模型表现出高灵敏度，补充了放射科医生的高特异性，并且可以作为机会性筛选器。版权所有 © 2024 Société française de radiologie。由 Elsevier Masson SAS 出版。版权所有。

The purpose of this study was to develop a radiomics-based algorithm to identify small pancreatic neuroendocrine tumors (PanNETs) on CT and evaluate its robustness across manual and automated segmentations, exploring the feasibility of automated screening.Patients with pathologically confirmed T1 stage PanNETs and healthy controls undergoing dual-phase CT imaging were retrospectively identified. Manual segmentation of pancreas and tumors was performed, then automated pancreatic segmentations were generated using a pretrained neural network. A total of 1223 radiomics features were independently extracted from both segmentation volumes, in the arterial and venous phases separately. Ten final features were selected to train classifiers to identify PanNETs and controls. The cohort was divided into training and testing sets, and performance of classifiers was assessed using area under the receiver operator characteristic curve (AUC), specificity and sensitivity, and compared against two radiologists blinded to the diagnoses.A total of 135 patients with 142 PanNETs, and 135 healthy controls were included. There were 168 women and 102 men, with a mean age of 55.4 ± 11.6 (standard deviation) years (range: 20-85 years). Median PanNET size was 1.3 cm (Q1, 1.0; Q3, 1.5; range: 0.5-1.9). The arterial phase LightGBM model achieved the best performance in the test set, with 90 % sensitivity (95 % confidence interval [CI]: 80-98), 76 % specificity (95 % CI: 62-88) and an AUC of 0.87 (95 % CI: 0.79-0.94). Using features from the automated segmentations, this model achieved an AUC of 0.86 (95 % CI: 0.79-0.93). In comparison, the two radiologists achieved a mean 50 % sensitivity and 100 % specificity using arterial phase CT images.Radiomics features identify small PanNETs, with stable performance when extracted using automated segmentations. These models demonstrate high sensitivity, complementing the high specificity of radiologists, and could serve as opportunistic screeners.Copyright © 2024 Société française de radiologie. Published by Elsevier Masson SAS. All rights reserved.