期刊
JOURNAL OF MOLECULAR STRUCTURE
卷 1229, 期 -, 页码 -出版社
ELSEVIER
DOI: 10.1016/j.molstruc.2020.129493
关键词
Gastric Cancer; PCA; SVM; KNN; FTIR; Feature selection
资金
- Pharmaceutical Sciences Research Center at Shahid Beheshti University of Medical Sciences (SBMU)
- Faculty of Pharmacy at Shahid Beheshti University of Medical Sciences (SBMU)
- Shohada Hospital
In this study, FTIR-ATR spectroscopy was used to compare gastric samples, and data modeling was performed using PCA, SVM, and KNN algorithms. Specific peaks related to malignancy were identified in malignant tissue, which can be used to distinguish between normal and malignant samples.
Since the diagnosis of gastric cancer in most cases happens in advanced stages and the pathologist judgment plays the major role in the diagnosis, Fourier Transform Infrared (FTIR) attenuated total reflectance (ATR) spectroscopy as a new, fast, non-invasive, and accurate diagnosis and screening tool was used to compare gastric samples in this study.Data modeling was performed based on the Principal Component Analysis (PCA), Support Vector Machines (SVM), and k-nearest neighbor algorithm (KNN) on the spectra of sixty fixed gastric tissue samples. Malignancy was characterized by the peaks that are mainly related to amide III and protein structure at around 1285 cm(-1) and 1339 cm(-1), delta (CH2), lipids, fatty acids, and delta (CH) at around 1439 cm(-1). Spectra comparison also indicates differences in malignant tissue's peak positions for CH2 wagging, amide II and amide I as well as the CH scissoring of the acyl chain of lipids. The statistical analysis results confirm this modeling method to distinguish about 81.7% of the normal and malignant samples just with one feature and suggest that ATR-FTIR spectroscopy may be a potentially useful tool for the diagnosis of gastric cancer. (C) 2020 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据