A clinical-information-free method for early diagnosis of lung cancer from the patients with pulmonary nodules based on backpropagation neural network model

• Published in: Computational and structural biotechnology journal Vol. 24; pp. 404 - 411
• Main Authors: Yang, Xin, Wu, Changchun, Liu, Wenwen, Fu, Kaiyu, Tian, Yuke, Wei, Xing, Zhang, Wei, Sun, Ping, Luo, Huaichao, Huang, Jian
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.12.2024; Research Network of Computational and Structural Biotechnology; Elsevier
• Subjects: Backpropagation neural network; Characteristic TCR clone; Early diagnosis; Lung cancer; Special issue articles from "AI-based Medical Imaging and Assistive Diagnostic Technologies" edited by Qi Zhao; TCRβ repertoire; Early diagnosis; TCRβ repertoire; Backpropagation neural network; Characteristic TCR clone; Lung cancer
• ISSN: 2001-0370; 2001-0370
• DOI: 10.1016/j.csbj.2024.05.010

Lung cancer is the main cause of cancer-related deaths worldwide. Due to lack of obvious clinical symptoms in the early stage of the lung cancer, it is hard to distinguish between malignancy and pulmonary nodules. Understanding the immune responses in the early stage of malignant lung cancer patients may provide new insights for diagnosis. Here, using high-through-put sequencing, we obtained the TCRβ repertoires in the peripheral blood of 100 patients with Stage I lung cancer and 99 patients with benign pulmonary nodules. Our analysis revealed that the usage frequencies of TRBV, TRBJ genes, and V-J pairs and TCR diversities indicated by D50s, Shannon indexes, Simpson indexes, and the frequencies of the largest TCR clone in the malignant samples were significantly different from those in the benign samples. Furthermore, reduced TCR diversities were correlated with the size of pulmonary nodules. Moreover, we built a backpropagation neural network model with no clinical information to identify lung cancer cases from patients with pulmonary nodules using 15 characteristic TCR clones. Based on the model, we have created a web server named “Lung Cancer Prediction” (LCP), which can be accessed at http://i.uestc.edu.cn/LCP/index.html. [Display omitted] •TCR diversities in the malignant samples were significantly different from those in the benign samples.•Reduced TCR diversities were correlated with the size of pulmonary nodules.•We built a backpropagation neural network model with no clinical information to identify lung cancer.