Kim Minsung, Park Taeyong, Kang Jaewoong, Kim Min-Jeong, Kwon Mi Jung, Oh Bo Young, Kim Jong Wan, Ha Sangook, Yang Won Seok, Cho Bum-Joo, Son Iltae
Department of Surgery, Hallym University Medical Center, Hallym Sacred Heart Hospital, Hallym University College of Medicine, 22 Gwanpyeong-ro 170 beon-gil, Pyeongan-dong, Dongan-gu, Anyang, Gyeonggi-do, Republic of Korea.
Medical Artificial Intelligence Center, Hallym University Medical Center, Anyang, Republic of Korea.
Sci Rep. 2025 Mar 5;15(1):7711. doi: 10.1038/s41598-024-84348-6.
Rapid, accurate preoperative imaging diagnostics of appendicitis are critical in surgical decisions of emergency care. This study developed a fully automated diagnostic framework using a 3D convolutional neural network (CNN) to identify appendicitis and clinical information from patients with abdominal pain, including contrast-enhanced abdominopelvic computed tomography images. A deep learning model-Information of Appendix (IA)-was developed, and the volume of interest (VOI) region corresponding to the anatomical location of the appendix was automatically extracted. It was analysed using a two-stage binary algorithm with transfer learning. The algorithm predicted three categories: non-, simple, and complicated appendicitis. The 3D-CNN architecture incorporated ResNet, DenseNet, and EfficientNet. The IA model utilising DenseNet169 demonstrated 79.5% accuracy (76.4-82.6%), 70.1% sensitivity (64.7-75.0%), 87.6% specificity (83.7-90.7%), and an area under the curve (AUC) of 0.865 (0.862-0.867), with a negative appendectomy rate of 12.4% in stage 1 classification identifying non-appendicitis versus. appendicitis. In stage 2, the IA model exhibited 76.1% accuracy (70.3-81.9%), 82.6% sensitivity (62.9-90.9%), 74.2% specificity (67.0-80.3%), and an AUC of 0.827 (0.820-0.833), differentiating simple and complicated appendicitis. This IA model can provide physicians with reliable diagnostic information on appendicitis with generality and reproducibility within the VOI.
阑尾炎的快速、准确的术前影像诊断对于急诊手术决策至关重要。本研究开发了一种使用三维卷积神经网络（CNN）的全自动诊断框架，以从腹痛患者的腹部增强计算机断层扫描图像中识别阑尾炎及临床信息。开发了一种深度学习模型——阑尾信息（IA），并自动提取与阑尾解剖位置对应的感兴趣体积（VOI）区域。使用具有迁移学习的两阶段二元算法对其进行分析。该算法预测了三类：非阑尾炎、单纯性阑尾炎和复杂性阑尾炎。三维CNN架构融合了ResNet、DenseNet和EfficientNet。利用DenseNet169的IA模型在第一阶段分类中识别非阑尾炎与阑尾炎时，准确率为79.5%（76.4 - 82.6%），灵敏度为70.1%（64.7 - 75.0%），特异性为87.6%（83.7 - 90.7%），曲线下面积（AUC）为0.865（0.862 - 0.867），阴性阑尾切除率为12.4%。在第二阶段，IA模型在区分单纯性阑尾炎和复杂性阑尾炎时，准确率为76.1%（70.3 - 81.9%），灵敏度为82.6%（62.9 - 90.9%），特异性为74.2%（67.0 - 80.3%），AUC为0.827（0.820 - 0.833）。这种IA模型可以为医生提供关于阑尾炎的可靠诊断信息，在VOI内具有通用性和可重复性。
