Baky A A, Winkler D G, Hunter N R, Subach J A, Greenberg S D, Spjut H J, Estrada R, Kimzey S L
Anal Quant Cytol. 1980 Sep;2(3):175-85.
A statistical procedure was developed for calculating an atypia status index (ASI) for cells from the bronchial mucosa. These indices represent the degree of abnormal changes in these cells and classify them as squamous metaplasia, mild atypia, moderate atypia, servere atypia or carcinoma. The classification accuracy of the procedure was more than 99% on trained data and was accomplished by minimizing the overlapping areas between the adjacent multivariate distributions of cell groups for selected features whose group means represent a monotonic function and for which the cell categories are distinguishable. The calculated ASI may reflect abnormal changes in the cell that may not be clearly evident visually. It appears that progression or reversal of bronchial epithelial atypia can be accurately monitored by studying the changes in the ASI, not only for preneoplasia but also for reactions to chemotherapy and various pulmonary infectious disease processes, such as influenza.
已开发出一种统计程序,用于计算支气管黏膜细胞的异型性状态指数(ASI)。这些指数代表这些细胞的异常变化程度,并将它们分类为鳞状化生、轻度异型性、中度异型性、重度异型性或癌。该程序在训练数据上的分类准确率超过99%,是通过最小化所选特征的细胞组相邻多变量分布之间的重叠区域来实现的,这些特征的组均值代表单调函数,且细胞类别是可区分的。计算出的ASI可能反映细胞中视觉上可能不明显的异常变化。似乎通过研究ASI的变化,可以准确监测支气管上皮异型性的进展或逆转,不仅适用于肿瘤前期,也适用于对化疗的反应以及各种肺部感染性疾病过程,如流感。