Wang Gufeng, Platz Charles P, Geng M Lei
Department of Chemistry, The Optical Science and Technology Center, and The Center for Biocatalysis and Bioprocessing, University of Iowa, Iowa City, Iowa 52242, USA.
Appl Spectrosc. 2006 May;60(5):545-50. doi: 10.1366/000370206777412220.
Differential normalized fluorescence (DNF) is an efficient and effective method for the differentiation of normal and cancerous tissue fluorescence spectra. The diagnostic features are extracted from the difference between the averaged cancerous and averaged normal tissue spectra and used as indices in tissue classification. In this paper, a new method, probability-based DNF bivariate analysis, is introduced based on the univariate DNF method. Two differentiation features are used concurrently in the new method to achieve better classification accuracy. The probability of each sample belonging to a disease state is determined with Bayes decision theory. This probability approach classifies the tissue spectra according to disease states and provides uncertainty information on classification. With a data set of 57 colonic tissue sites, probability-based DNF bivariate analysis is demonstrated to improve the accuracy of cancer diagnosis. The bivariate DNF analysis only requires the collection of a few data points across the entire emission spectrum and has the potential of improving data acquisition speed in tissue imaging.
差异归一化荧光（DNF）是一种区分正常组织和癌组织荧光光谱的高效方法。诊断特征从平均癌组织光谱与平均正常组织光谱之间的差异中提取，并用作组织分类的指标。本文在单变量DNF方法的基础上，引入了一种新方法——基于概率的DNF双变量分析。新方法同时使用两个区分特征以实现更高的分类准确率。利用贝叶斯决策理论确定每个样本属于疾病状态的概率。这种概率方法根据疾病状态对组织光谱进行分类，并提供分类的不确定性信息。通过一个包含57个结肠组织部位的数据集，证明基于概率的DNF双变量分析可提高癌症诊断的准确率。双变量DNF分析仅需要在整个发射光谱上采集几个数据点，并且具有提高组织成像数据采集速度的潜力。
