基于R-EBUS和pCLE多模态影像的肺外周病变诊断预测模型研究

径向支气管内超声和探针型激光共聚焦显微内镜引导下的经支气管肺活检术是一项对肺外周病变进行病理诊断的有效方式。两种引导设备可在同一次操作中通过同一支气管镜工作通道或引导鞘管到达病变，根据影像特征半定量、定量分析结果，进行无创影像诊断并引导精准活检，具有实时、快速、无辐射、精确度高、无创伤等优点。.本课题拟构建大数据径向支气管内超声和探针型激光共聚焦显微内镜多模态影像标注库。通过深度学习，建立肺外周病变人工智能影像诊断预测模型，并完成临床验证。从而去除医学影像的伪像干扰，避免医生的主观因素影响，使患者在病理诊断之前客观、高效、精准、全面地获得无创影像诊断。提高病理诊断阳性诊断率，优化诊治策略。使肺外周病变的诊治关口前移，精准治疗，改善患者疾病预后。

Transbronchial lung biopsy guided by radial endobronchial ultrasound (R-EBUS) and probe confocal laser endoscopy (pCLE) is an effective method for pathological diagnosis of peripheral pulmonary lesions (PPLs). R-EBUS and pCLE can reach the lesion through the same bronchoscope working channel or guide sheath in the same procedure. According to the semi-quantitative and quantitative analysis of image characteristics, pulmonary physicians can identify diseases and guide the biopsy accurately, which has the advantages of real-time, rapid, no radiation, high accuracy and noninvasive..Our study is designed to build a large multimodal image annotation database of R-EBUS and pCLE. According to deep learning, artificial intelligence diagnosis and prediction model for PPLs will be established and clinical verification will be completed. In this way, the pseudo-image interference of image would be removed, and the subjective factors of physicians would be avoided, so that patients can obtain the objective, efficient, accurate, comprehensive and noninvasive image diagnosis before the pathological diagnosis. At the same time, the pathological diagnosis yield will be improved and the diagnosis and treatment strategy will be optimized. Therefore, the patients with PPLs can receive the treatments earlier and more precise, which can improve the prognosis of the patients.