Department of Endoscopy, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China.
The First Affiliated Hospital of Harbin Medical University, Harbin, China.
Laryngoscope. 2024 Oct;134(10):4321-4328. doi: 10.1002/lary.31537. Epub 2024 May 27.
Vocal fold leukoplakia (VFL) is a precancerous lesion of laryngeal cancer, and its endoscopic diagnosis poses challenges. We aim to develop an artificial intelligence (AI) model using white light imaging (WLI) and narrow-band imaging (NBI) to distinguish benign from malignant VFL.
A total of 7057 images from 426 patients were used for model development and internal validation. Additionally, 1617 images from two other hospitals were used for model external validation. Modeling learning based on WLI and NBI modalities was conducted using deep learning combined with a multi-instance learning approach (MIL). Furthermore, 50 prospectively collected videos were used to evaluate real-time model performance. A human-machine comparison involving 100 patients and 12 laryngologists assessed the real-world effectiveness of the model.
The model achieved the highest area under the receiver operating characteristic curve (AUC) values of 0.868 and 0.884 in the internal and external validation sets, respectively. AUC in the video validation set was 0.825 (95% CI: 0.704-0.946). In the human-machine comparison, AI significantly improved AUC and accuracy for all laryngologists (p < 0.05). With the assistance of AI, the diagnostic abilities and consistency of all laryngologists improved.
Our multicenter study developed an effective AI model using MIL and fusion of WLI and NBI images for VFL diagnosis, particularly aiding junior laryngologists. However, further optimization and validation are necessary to fully assess its potential impact in clinical settings.
3 Laryngoscope, 134:4321-4328, 2024.
声带白斑（VFL）是喉癌的癌前病变，其内镜诊断具有挑战性。我们旨在开发一种使用白光成像（WLI）和窄带成像（NBI）的人工智能（AI）模型来区分良性和恶性 VFL。
总共使用了 426 名患者的 7057 张图像来开发和内部验证模型。此外，还使用了来自另外两家医院的 1617 张图像进行模型外部验证。基于 WLI 和 NBI 模式的建模学习采用深度学习与多实例学习方法（MIL）相结合的方式进行。此外，还使用了 50 个前瞻性采集的视频来评估实时模型性能。涉及 100 名患者和 12 名喉科医生的人机比较评估了该模型的实际效果。
该模型在内部和外部验证集中的最高受试者工作特征曲线（AUC）值分别为 0.868 和 0.884。视频验证集中的 AUC 为 0.825（95%CI：0.704-0.946）。在人机比较中，人工智能显著提高了所有喉科医生的 AUC 和准确率（p<0.05）。在人工智能的辅助下，所有喉科医生的诊断能力和一致性都得到了提高。
我们的多中心研究使用 MIL 和 WLI 与 NBI 图像融合开发了一种用于 VFL 诊断的有效 AI 模型，特别是对初级喉科医生有帮助。然而，为了充分评估其在临床环境中的潜在影响，还需要进一步优化和验证。
3 级喉镜，134：4321-4328，2024。
