摘要
应用计算机数字图像分析技术,对104例乳腺及良、恶性病变的标本进行形态计量研究,提取了六项细胞核参数。它们是核最大和最小灰度、核中心平均距、核周长、核面积和圆偏度。结果证明:在乳腺及病变的上皮细胞中,细胞及细胞核的大小,形状及核物质密度均按正常乳腺—良性肿瘤(或良性病变)—交界性病变—癌的顺序呈规律性变化。各病变彼此之间及同正常乳腺之间,除不典型增生和乳腺癌两者之间外,均可提取有鉴别意义的形态学定量参数。模式识别分类分析结果友明:五类样本之间有明显分尽,说明从组织切片测得前述六项指标,对乳腺疾病进行计算机判别分类,有一定临床意义。图像分析技术能准确地确定不典型增生的程度及与原位癌鉴别。
Morphometrical studies were made on 104 specimens of mammary gland and its lesions by the digital image analysis technique of electronic computers.6 parameters were obtained:the max- imal grey degree,the minimal grey degree,the mean distance of mucleus centre,the girth,the sur- face area and the degree of deviation from a circle(DC).The results showed that the size,the form and the density of the nucleis in the cells of mammary gland and its lesions took on regular in the order of normal mammary gland,benign lesions,atypical hyperplasia and breast cancer. Quantitative morphological parameters could be obtained,which were significance on identifica- tion for every two kinds of lesions ,or the lesions and the normal mammary gland.It was uneasy to distinguish the atypical hyperplasia from the breast carcer.The results from the analysis on the classification of the pattern recognizationg showed that there were clear boundaries between every two kinds of lesions .which meant that there was some clinical significance to classify the mamma- ry gland lesions trhough 6 parameters produced by the computers.The technique of the image analysis could be used to identify the degree of atypical hyperplasia and to divide it from carcino- ma in situ.
关键词
乳腺疾病
上皮细胞核
形态计量法
模式识别
breast disease
breast neoplasm
hyperplasia morphometry
pattern recognization
