Colorectal Cancer and Colitis Diagnosis Using Fourier Transform Infrared Spectroscopy and an Improved K-Nearest-Neighbour Classifier

作者：Qingbo Li, Can Hao, Kang Xue, Jialin Zhang, Xuejun Sun, Wenbo Wang, Haishan Zeng · 发表于：Sensors · 年份：2017 · DOI：10.3390/s17122739 · 被引用次数：26 · 研究领域：Spectroscopy and Chemometric Analyses、Spectroscopy Techniques in Biomedical and Chemical Research、Advanced Chemical Sensor Technologies

Combining Fourier transform infrared spectroscopy (FTIR) with endoscopy, it is expected that noninvasive, rapid detection of colorectal cancer can be performed in vivo in the future. In this study, Fourier transform infrared spectra were collected from 88 endoscopic biopsy colorectal tissue samples (41 colitis and 47 cancers). A new method, viz., entropy weight local-hyperplane k-nearest-neighbor (EWHK), which is an improved version of K-local hyperplane distance nearest-neighbor (HKNN), is proposed for tissue classification. In order to avoid limiting high dimensions and small values of the nearest neighbor, the new EWHK method calculates feature weights based on information entropy. The average results of the random classification showed that the EWHK classifier for differentiating cancer from colitis samples produced a sensitivity of 81.38% and a specificity of 92.69%.