局部晚期下咽癌新辅助免疫治疗联合化疗不同疗效的喉镜图像差异分析
Analysis of Laryngoscopic Image Differences in Different Responses to Neoadjuvant Immunotherapy Combined With Chemotherapy in Locally Advanced Hypopharyngeal Carcinoma.
文献信息
| PMID | 42948597 |
|---|---|
| 原文 | 在 PubMed 查看原文 ↗ |
| 发表日期 | 2026 |
| 作者 | Wen Gao |
| 作者单位 | Department of Otorhinolaryngology Head and Neck Surgery Beijing Tongren Hospital, Capital Medical University Beijing China. |
| 期刊 | Laryngoscope investigative otolaryngology |
| SCI 分区 | Q2 |
| IF | 2.1 |
| 研究类型 | AI/ML · 临床 |
| 所属专科 | 咽喉科 |
中文摘要
目的： 探讨基于治疗前喉镜图像的影像组学特征在预测局部晚期下咽鳞状细胞癌新辅助免疫化疗疗效中的潜在价值。
材料与方法： 回顾性纳入在本院接受新辅助免疫化疗的局部晚期下咽鳞状细胞癌患者，并收集其治疗前的白光和窄带成像（NBI）喉镜图像。根据治疗疗效将患者分为完全缓解（CR）组和非完全缓解（non-CR）组。在所有图像的原始病灶区域统一勾画感兴趣区后，将图像转换为HSV、RGB和CIELAB色彩空间，共得到九个通道。随后在每个通道上提取104个纹理特征。采用独立样本t检验进行组间差异分析，并通过LASSO回归进一步筛选关键特征。
结果： 本研究共分析499张喉镜图像。统计分析在白光图像中识别出67个差异特征，并最终确定8个关键特征；在NBI图像中识别出52个差异特征，并最终确定11个关键特征。
结论： 本探索性研究初步表明，基于治疗前喉镜图像的影像组学特征在区分新辅助免疫化疗疗效方面具有潜力。所筛选出的关键特征为后续构建预测模型提供了有用线索，但其临床适用性仍需在大样本前瞻性研究中进一步验证。
证据等级： 3。
英文摘要
OBJECTIVE: To investigate the potential value of radiomic features based on pretreatment laryngoscopic images in predicting the efficacy of neoadjuvant immunochemotherapy for locally advanced hypopharyngeal squamous cell carcinoma.
MATERIALS AND METHODS: Patients with locally advanced hypopharyngeal squamous cell carcinoma who received neoadjuvant immunochemotherapy at our hospital were retrospectively enrolled, and their pretreatment white-light and narrow-band imaging (NBI) laryngoscopic images were collected. Patients were divided into a complete response (CR) group and a non-complete response (non-CR) group according to treatment efficacy. After uniformly delineating regions of interest on the original lesion areas of all images, the images were converted into HSV, RGB, and CIELAB color spaces, yielding a total of nine channels. A total of 104 texture features were then extracted on each channel. Intergroup difference analysis was performed using an independent-samples t-test, and key features were further screened by LASSO regression.
RESULTS: A total of 499 laryngoscopic images were analyzed in this study. Statistical analysis identified 67 differential features and ultimately 8 key features in white-light images; 52 differential features and ultimately 11 key features were identified in NBI images.
CONCLUSION: This exploratory study preliminarily demonstrates that radiomic features based on pretreatment laryngoscopic images have potential in differentiating the efficacy of neoadjuvant immunochemotherapy. The screened key features provide useful clues for the subsequent construction of prediction models, although their clinical applicability still requires further validation in large-sample prospective studies.
LEVEL OF EVIDENCE: 3.