摘要
本文探讨了数字图象处理与模式识别技术在胃癌细胞识别中的应用。作者在病理医师的帮助下,从二十多例胃刷涂片上采集了56个单细胞图象。先对细胞图象进行预处理,然后分割细胞的核与浆;在此基础上抽取样本细胞特征训练分类器。在分类方法上,应用了人工神经网络技术,得到了优于传统方法的结果。
<ABSTRACT>
The application of digital image processing and pattern recognition
to the diagnosis of stomach smears. Under the supervision of patholo-
gists, cell images are collected and divided into three classes: normal
ce1ls (I ), cells between normal and cancer (Ⅱ)and cancer cells (Ⅲ).
At first, the cell images are preprocessed. The images are enhanced by
histogram equalization and median filtering. Then by computing the th-
reshold usinn the maximum mean spuare error algorithm, the cytoplasm
and the nucleus of a cell can be segmented. On the basis of the above
(Continued on page .34 )
(Continued from page 60 )
work, six features are extracted from the cell images. A neural net-
work approach for the classifjcation is described. The utilized network
is a multilayer perceptrons (MLP) .The backpropagation learning is us-
ed for its training. The performance of the MLP is compared to tradi-
tional linear classifiers. It is shown that the MLP outperforms tradi-
tional linear classifieis.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
1993年第1期56-60,34,共6页
Chinese Journal of Biomedical Engineering
关键词
图象处理
模式识别
细胞
胃肿瘤
Digital image
Pattern recognition
Cell
Classifiers
Neural network
