以EGFR-TKI为代表的靶向治疗为晚期非小细胞肺癌（NSCLC）患者带来了曙光，但由于肿瘤细胞演化与异质性，靶向治疗不可避免会耐药，且不同患者出现耐药的时间各异。我们在前期建立了基于影像组学的晚期NSCLC EGFR-TKI靶向治疗耐药预测模型，但其预测准确度尚待提高，且在不同队列中的预测准确性波动较大，推测可能与患者个体差异、肿瘤异质性、复杂分子网络及仅利用常规影像组学方法对胸部CT与临床病理信息进行关联分析，缺乏对不同耐药机制深入探讨有关。为此，本项目拟在前期研究基础上，利用已建立的晚期NSCLC EGFR-TKI靶向治疗临床队列，扩大影像组学研究样本量；同时采用单细胞全外显子测序及转录组测序，多组学筛选介导靶向治疗耐药的关键分子标志物；并利用人工智能深入挖掘影像组学与患者临床、病理、基因多组学大数据之间的关系，构建更加精准的EGFR-TKI靶向治疗耐药预测模型，辅助制定临床决策。

Targeted therapy, represented by EGFR-TKI, brings light to advanced non-small cell lung cancer (NSCLC) patients. However, resistance of targeted therapy is inevitable, and the interval for resistance varies among patients. In previous studies, we have established a radiomics-based model to predict the EGFR-TKI resistance for advanced NSCLC patients. However, the prediction accuracy is not fully satisfied and fluctuated greatly in different cohorts, which might be due to the individual difference, tumor heterogeneity, complex molecular network and the use of conventional radiomics methods to correlate chest CT with clinical pathology information, lack of in-depth discussion of different drug resistance mechanisms. Therefore, in the current research, we intend to screen the key biomarkers which mediate the targeted drug resistance by single-cell whole exon sequencing and transcriptome sequencing. And then to explore the relationship between radiomics and the patients’ clinical, pathological and genetic multi-omics big data by artificial intelligence, to develop a more accurate model for the EGFR-TKI targeted therapy resistance prediction.