Zeng Lulu, Li Hui, Huang Tian, Heng Yuting, Liu Jun, Hu Xiangpeng
Department of Gastroenterology, the Second Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, China.
Department of Pathophysiology, School of Basic Medical College, Anhui Medical University, Hefei, Anhui Province, China; Functional Experiment Center, School of Basic Medical College, Anhui Medical University, Hefei, Anhui Province, China.
Gastroenterol Hepatol. 2025 Apr;48(4):502265. doi: 10.1016/j.gastrohep.2024.502265. Epub 2024 Oct 11.
Endoscopists utilize depth-predicting score (DPS) and simplified depth-predicting score (S-DPS) to predict the invasion depth of early gastric cancer based on conventional white-light endoscopic features. The effectiveness of these scores has not been fully elucidated among nonexpert endoscopists. This study aimed to compare the ability of DPS and S-DPS to predict invasion depth of differentiated early gastric cancers by nonexpert endoscopists.
We collected subitem scores of DPS and S-DPS from 19 nonexpert endoscopists for early gastric cancer conventional white-light endoscopy images in the test dataset to predict the invasion depth of the early gastric cancer conventional white-light endoscopy images. Accuracy, specificity, overdiagnosis rate, and underdiagnosis rate were subsequently calculated using the histological invasion depth as the gold standard.
Using 3 as the cutoff line, the overall S-DPS diagnostic accuracy for invasion depth was significantly greater than that of DPS [73.86% (69.32%, 75.00%) vs. 67.05% (62.50%, 71.59%), p=0.005]. The overall S-DPS overdiagnosis rate was significantly lower than that of DPS [7.58% (3.03%, 13.64%) vs. 28.79% (18.18%, 37.88%), p=0.000]. The overall S-DPS under-diagnosed rate was significantly higher than that of DPS [86.36% (68.18%, 90.91%) vs. 45.45% (31.82%, 59.09%), p=0.000]. The specificity of the S-DPS was significantly greater than that of DPS [92.42% (86.36%, 96.97%) vs. 71.21% (62.12%, 81.82%), p=0.000].
The diagnostic accuracy of the S-DPS was greater than that of the DPS among nonexpert endoscopists. Furthermore, S-DPS is simpler than other methods, making it more conducive to clinical application for nonexpert endoscopists.
内镜医师利用深度预测评分（DPS）和简化深度预测评分（S-DPS），基于传统白光内镜特征预测早期胃癌的浸润深度。在非专业内镜医师中，这些评分的有效性尚未得到充分阐明。本研究旨在比较非专业内镜医师使用DPS和S-DPS预测分化型早期胃癌浸润深度的能力。
我们从19名非专业内镜医师处收集了测试数据集中早期胃癌传统白光内镜图像的DPS和S-DPS子项评分，以预测早期胃癌传统白光内镜图像的浸润深度。随后以组织学浸润深度作为金标准，计算准确率、特异性、过度诊断率和漏诊率。
以3为截断值时，S-DPS对浸润深度的总体诊断准确率显著高于DPS[73.86%（69.32%，75.00%）对67.05%（62.50%，71.59%），p=0.005]。S-DPS的总体过度诊断率显著低于DPS[7.58%（3.03%，13.64%）对28.79%（18.18%，37.88%），p=0.000]。S-DPS的总体漏诊率显著高于DPS[86.36%（68.18%，90.91%）对45.45%（31.82%，59.09%），p=0.000]。S-DPS的特异性显著高于DPS[92.42%（86.36%，96.97%）对71.21%（62.12%，81.82%），p=0.000]。
在非专业内镜医师中，S-DPS的诊断准确率高于DPS。此外，S-DPS比其他方法更简单，更有利于非专业内镜医师的临床应用。
