Shen Zhefan, Wei Ying, Liu Kexin, Ma Zhiqi, Zhang Zhiliang, Wang Xuechun, Li Yong, Shi Feng, Ding Zhongxiang
Department of Radiology, Affiliated Hospital of Jiaxing University, Jiaxing, P. R. China.
Department of Research and Development, Shanghai United Imaging Intelligence Co., Ltd., Shanghai, P. R. China.
Am J Rhinol Allergy. 2025 May;39(3):187-196. doi: 10.1177/19458924251313845. Epub 2025 Jan 17.
BackgroundComputed tomography (CT) plays a crucial role in assessing chronic rhinosinusitis, but lacks objective quantifiable indicators.ObjectiveThis study aimed to use deep learning for automated sinus segmentation to generate distinct quantitative scores and explore their correlations with disease-specific quality of life.MethodsFrom July 2021 to August 2022, 445 CT data were collected from 2 medical centers. A deep learning model based on nnU-Net was trained for automatic sinus segmentation and internally validated using 300 cases. The remaining 145 cases were split into an external testing set (74 cases) and an independent testing set (71 cases). Two quantitative scores, the quantitative Lund-MacKay score and the quantitative opacification score (QOS), were derived from the segmentation results. The quantitative scores' efficacy was assessed by comparing them with the Lund-MacKay score (LMS), the 22-item Sinonasal Outcome Test score (SNOT-22), and other clinical variables through correlation analyses. Furthermore, the relationship between quantitative scores and postoperative quality of life improvement was explored using single-factor logistic regression.ResultThe segmentation model achieved average Dice similarity coefficients of 0.993, 0.978, 0.958, and 0.871 for the training, validation, external testing, and independent testing sets, respectively. Both quantitative scores significantly correlated with the LMS (= 0.87 and = 0.70, < .001). Neither score correlated with the total SNOT-22 score, although the modified QOS showed significant correlations with the nasal and sleep subdomains (= 0.26 and = 0.27, <.05). No significant association was found between quantitative score and postoperative improvement in quality of life.ConclusionDeep learning enables the automated segmentation of sinuses on CT scans, producing quantitative scores of sinus opacification. These automatic quantitative scores may serve as tools for chronic rhinosinusitis assessment.
背景
计算机断层扫描（CT）在评估慢性鼻-鼻窦炎中起着关键作用，但缺乏客观可量化指标。
目的
本研究旨在利用深度学习进行鼻窦自动分割，以生成不同的定量评分，并探讨其与疾病特异性生活质量的相关性。
方法
2021年7月至2022年8月，从2个医疗中心收集了445份CT数据。基于nnU-Net的深度学习模型用于鼻窦自动分割，并使用300例病例进行内部验证。其余145例病例被分为外部测试集（74例）和独立测试集（71例）。从分割结果中得出两个定量评分，即定量Lund-MacKay评分和定量混浊度评分（QOS）。通过相关性分析，将定量评分与Lund-MacKay评分（LMS）、22项鼻鼻窦结局测试评分（SNOT-22）及其他临床变量进行比较，评估定量评分的有效性。此外，采用单因素逻辑回归探讨定量评分与术后生活质量改善之间的关系。
结果
分割模型在训练集、验证集、外部测试集和独立测试集中的平均Dice相似系数分别为0.993、0.978、0.958和0.871。两个定量评分均与LMS显著相关（r = 0.87和r = 0.70，P <.001）。两个评分均与SNOT-22总分无相关性，尽管改良QOS与鼻部和睡眠子域显著相关（r = 0.26和r = 0.27，P <.05）。定量评分与术后生活质量改善之间未发现显著关联。
结论
深度学习能够在CT扫描上实现鼻窦的自动分割，生成鼻窦混浊度的定量评分。这些自动定量评分可作为慢性鼻-鼻窦炎评估的工具。
