Wan Chayan, Cao Wenqing, Cheng Cungui
College of Chemistry and Life Science, Zhejiang Normal University, Jinhua 321004, China.
Shandong Exit-Entry Inspection and Quarantine Technology Center, Qingdao 266002, China.
J Anal Methods Chem. 2014;2014:564801. doi: 10.1155/2014/564801. Epub 2014 Dec 4.
Sprague-Dawley (SD) rats' normal and abnormal pancreatic tissues are determined directly by attenuated total reflectance Fourier transform infrared (ATR-FT-IR) spectroscopy method. In order to diagnose earlier stage of SD rats pancreatic cancer rate with FT-IR, a novel method of extraction of FT-IR feature using discrete wavelet transformation (DWT) analysis and classification with the probability neural network (PNN) was developed. The differences between normal pancreatic and abnormal samples were identified by PNN based on the indices of 4 feature variants. When error goal was 0.01, the total correct rates of pancreatic early carcinoma and advanced carcinoma were 98% and 100%, respectively. It was practical to apply PNN on the basis of ATR-FT-IR to identify abnormal tissues. The research result shows the feasibility of establishing the models with FT-IR-DWT-PNN method to identify normal pancreatic tissues, early carcinoma tissues, and advanced carcinoma tissues.
采用衰减全反射傅里叶变换红外（ATR-FT-IR）光谱法直接测定Sprague-Dawley（SD）大鼠的正常和异常胰腺组织。为了利用傅里叶变换红外光谱（FT-IR）诊断SD大鼠胰腺癌的早期阶段，开发了一种使用离散小波变换（DWT）分析提取FT-IR特征并结合概率神经网络（PNN）进行分类的新方法。基于4个特征变量的指标，通过PNN识别正常胰腺样本与异常样本之间的差异。当误差目标为0.01时，胰腺早期癌和晚期癌的总正确识别率分别为98%和100%。基于ATR-FT-IR应用PNN识别异常组织是可行的。研究结果表明，采用FT-IR-DWT-PNN方法建立识别正常胰腺组织、早期癌组织和晚期癌组织模型是可行的。
