摘 要：传统方法预测肺癌患者术后复发通常使用 PET/CT 图像或临床数据等单一模态信息，而文章在卷积神经网络VGG 模型的基础上研究了多模态融合的潜力，通过结合 PET/CT 图像信息、临床数据和影像组学信息对肺癌复发实现了更好的预测。实验结果表明，对 160 名患者的 NSCLC 放射基因组学数据集进行研究时使用三种模态信息预测 NSCLC 患者复发性能达到最佳，其准确率为 84.38%，精确率为 82.76%，召回率为 68.75%，AUC 为 79.69%。
关键词：非小细胞肺癌；PET/CT 图像；临床；影像组学；复发预测
DOI:10.19850/j.cnki.2096-4706.2022.24.020
中图分类号：TP18 文献标识码：A 文章编号：2096-4706（2022）24-0078-04
Recurrence Prediction of Non-small Cell Lung Cancer Based on Multimodal Fusion of VGG Network
SONG Yue, LI Bing, MA Yanan
(School of Information Engineering, Lanzhou University of Finance and Economics, Lanzhou 730020, China)
Abstract: In normal, medical workers use single modal information such as PET/CT images or clinical data to predict postoperative recurrence of lung cancer patients. This paper studies the potential of multimodal fusion based on the convolutional neural network VGG model, and achieves a better prediction of lung cancer recurrence by combining PET/CT image information, clinical data and imageomics information. The experimental results show that the best performance of NSCLC recurrence prediction is achieved by using three modal information when studying NSCLC radiogenomics dataset of 160 patients, with the accuracy rate of 84.38%, the precision rate of 82.76%, the recall rate of 68.75%, and the AUC of 79.69%.
Keywords: non-small cell lung cancer; PET/CT image; clinical; imageomics; recurrence prediction
参考文献：
[1] DUMA N，SANTANA-DAVILA R，MOLINA J R. Non-small cell lung cancer：Epidemiology，screening，diagnosis，and treatment [J]. Mayo Clin Proceedings，2019，94（8）：1623-1640.
[2] CRINO L，WEDER W，VAN MEERBEECK J，et al. Early stage and locally advanced （non-metastatic）non-small-cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis，treatment and follow-up [J].Annals of Oncology，2010，21（5）：103-115.
[3] HUNG J J，WU Y C. Stage I Non-Small Cell Lung Cancer: Recurrence Patterns，Prognostic Factors and Survival [J/OL].Topics in Thoracic Surgery，2012，3（16）：285-292[2022-07-12].https://www. docin.com/p-421240240.html.
[4] SUGIMURA H，NICHOLS F C，YANG P，et al. Survival after recurrent nonsmall-cell lung cancer after complete pulmonary resection [J].The Annals of Thoracic Surgery，2007，83（2）：409-418.
[5] THOMAS P，RUBINSTEIN L. Cancer recurrence after resection：T1 N0 non-small cell lung cancer [J].The Annals of thoracic surgery，1990，49（2）：242-247.
[6] GRIFFETH L K. Use of PET/CT scanning in cancer patients： technical and practical considerations [J].Proc（Bayl Univ Med Cent），2005，18（4）：321-330.
[7] SUBRAMANIAN V，DO M N，MAHMOOD T S. Multimodal Fusion of Imaging and Genomics for Lung Cancer Recurrence Prediction [J/OL].arXiv：2002.01982 [eess.IV].[2022-07-12]. https://arxiv.org/abs/2002.01982.
[8] AONPONG P，IWAMOTO Y，HAN X H，et al. GenotypeGuided Radiomics Signatures for Recurrence Prediction of Non-Small Cell Lung Cancer [J].IEEE Access，2021，9：90244-90254.
[9] CLARK K，VENDT B，SMITH K，et al. The Cancer Imaging Archive（TCIA）：Maintaining and Operating a Public Information Repository [J].J Digit Imaging，2013，26（6）：1045-1057.
作者简介：宋玥（1997—），女，汉族，山西吕梁人，硕士在读，主要研究方向：信息管理与信息系统；李兵（1966—），男，汉族，山西太原人，教授，博士，主要研究方向：计算机、管理信息系统；马亚楠（1997—），女，汉族，河南郑州人，硕士在读，主要研究方向：信息管理与信息系统。