Image based prediction of aggressive early lung cancer in lung cancer screening populations

1)简要说明研究背景，包括潜在影响肺癌筛查邀请高危受试者对其肺部进行CT扫描，以确定早期可治疗的肺癌。到2028年，在英国国家肺癌筛查计划中，每年将有大约500-75万人接受肺部CT扫描。2-3%的筛查对象会患上肺癌。肺癌可以表现出不同的生长和扩散到淋巴结的速度，一些肺癌，尽管治疗可能会复发。在早期阶段识别潜在的侵袭性肺癌可能会改变全球的肺癌管理。预计具有侵袭性的癌症可以在手术前进行额外的化疗。我们的研究将分析伦敦大学学院两项研究的数据：顶峰和上升。这项顶峰研究是世界上最大的肺癌筛查研究之一，每年扫描1.3万名受试者以确定肺癌。Ascant研究包括所有被诊断为肺癌的顶峰研究患者。本研究旨在将肺癌生长的影像特征与临床和基因组突变的侵袭性标记物相关联。2)目的和目的确定恶性肺结节在低剂量CT扫描上的图像特征，以预测侵袭性疾病。评估纵隔淋巴结改变作为侵袭性疾病的预测因子。在影像数据上识别可预测侵袭性疾病的基因组特征。3)研究方法的新颖性定义早期肺癌的侵袭性。攻击性目前还没有正式的医学定义，但创建它对许多癌症类型来说可能是有价值的使用考虑肺部和肺外特征的医学成像的时间序列深度学习模型来预测疾病进展新颖的组织病理学-基因组-成像相关性来描绘预测侵袭性早期肺癌的多维风险分数4)与EPSRC的战略和研究领域保持一致EPSRC的战略优先事项：转变健康和医疗保健EPSRC研究组合和优先事项：医疗保健技术5)任何公司或合作者参与N/A

1) Brief description of the context of the research including potential impactLung cancer screening invites high risk subjects to have a CT scan of their lungs to identify early treatable lung cancer. By 2028 approximately 500-750,000 subjects will have a CT scan of their lungs annually in the National UK lung cancer screening program. 2-3% of screened subjects will have a lung cancer. Lung cancers can show differing rates of growth and spread to lymph nodes, and some lung cancers, despite treatment can recur. Identifying potentially aggressive lung cancers at an early stage could transform lung cancer management worldwide. Cancers expected to be aggressive could be treatment with extra chemotherapy prior to surgery. Our study will analyse data from two UCL studies: SUMMIT and ASCENT. The SUMMIT study is one of the largest lung cancer screening studies in the world which has scanned >13,000 subjects annually to identify lung cancer. The ASCENT study comprises all SUMMIT study patients where a lung cancer was diagnosed. The cancers in the ASCENT study have been genotyped and have longitudinal outcome data collected.This study aims to correlate imaging features of lung cancer growth with clinical and genomic mutational markers of aggression.2) Aims and Objectives Identify image-based features of malignant lung nodules on low-dose CT scans that predict aggressive disease.Evaluate mediastinal lymph node change as a predictor of aggressive disease.Identify genomic signatures on imaging data that can predict aggressive disease.3) Novelty of Research MethodologyDefining aggression in early lung cancer. Aggression currently has no formal medical definition, but creating this could be valuable for many cancer typesUse of time-series deep learning models on medical imaging considering lung and extra-lung features to predict disease progressionNovel histopathological-genomic-imaging correlations to delineate a multidimensional risk score predictive of aggressive early lung cancer4) Alignment to EPSRC's strategies and research areasEPSRC Strategic Priorities: transforming health and healthcareEPSRC Research Portfolio and Priorities: healthcare technologies5) Any companies or collaborators involvedN/A