胃粘膜癌前病变细胞的核和核仁组成区嗜银蛋白图像分析

IMAGE ANALYSIS OF NUCLEAR AND AGNOR IN CELLS OF PRECANCEROUS LESIONS OF THE STOMACH MUCOSA

以银染方法，用ＶＤＰ一１７５０型自动图像分析仪（美国ＶＩＣＯＭ公司），对１２５例胃粘膜细胞的核和核仁组成区嗜银蛋白，进行２０项几何学参数，构造参数测量。结果经多元判别分析，选择判别作用较大的参数组成判别函数，建立判别模式，取得了与原诊断较好的正确符合率，其回代正确率高达９３．６％；刀切法正确率达９０Ａ％。

With VDP-1750 automatic image analysis system(VICOM CO,U.S.A·),the geometric and texture features of epithelial cell nulear and nucleolar organizerregions were measured from precancerous lesions of 125 cases of stomach mucosa stainedby colloid silver technique. The features Which have greatest contri bution to diagnosiswere selected by step forward discrimination ,then a discriminant functio was establis-hed on these features Computer classification have a good result when compared withthe original diagnosis, the correct rate was 93.6% in replace mathod,and 90.4% by Jack -knife method.