摘要
为探讨大肠癌诊断的客观指标，本文采用电子计算机图象分析技术对大肠腺癌、腺瘤及正常肠粘膜细胞核仁组成区嗜银蛋白进行形态测量，建立了病变判别函数式。研究表明，Ａｇ－ＮＯＲ颗粒的形态参数在各病变组之间重叠较大，采用ＳＰＬＭ统计软件进行多因素判别分析建立的函数判别式虽能对大肠癌和腺瘤进行自动判别，但效果欠理想。究其原因主要是重度不典型增生的腺瘤在形态参数上与腺癌极为接近，此外，随机选择的腺瘤和腺癌病变不适于作为典型病变。进一步研究有待选择更为特征性的指标及选择具有典型形态参数改变的病例作标准，可望提高计算机自动图象诊断的准确率。
In order to make diagnosis of colorectal cancers objectively, Ag-NORs quantitative parameters were evaluated by a computer-assistant image analysis system in colorectal adenocarcinoma,adenoma and normal mucosa.It was showed that the morphological parameters of Ag-NORs were overlap among the different groups. Even the lesion-distinguished formula established by multiple factor analysis with SPLM software were difficult to give a satistactory result. The study showed that the same cytobiological characteristics of adenocarcinoma and adenoma with high-grade dysplasia and case selection at random may be partly responsible for that bad result. It suggested that further studies are necessary to select more sensitive and more specific morphological parameters to meet the computer-assistant diagnosis for colorectal tumors.
出处
《胃肠病学和肝病学杂志》
CAS
1994年第4期265-267,283,共4页
Chinese Journal of Gastroenterology and Hepatology
