摘要
针对目前乳腺X线影像中微钙化点的计算机辅助检测普遍假阳性较高的难点 ,提出一种能发挥差值、小波及神经网络等多种技术优势的综合处理检测方法。对临床实际病例 (10名患者 ,2 44个微钙化点 )的试用结果表明 ,与单纯使用上述检测技术比较 ,该方法不仅操作简单 ,且具有较高的检出率 (TP达 93 % ) ,同时还明显地降低了假阳性 ,值得深入研究。
In order to avoid a high false positive rate in the computer aided detection of microcalcifications in mammograms which is ourrently obtained, a synthetic processing method that exploits the good qualities of several techniques such as difference-image technique, wavelet transform, artificial neural network has been proposed and tested with 10 mammograms (244 microcalcifications). The test results showed that the synthetic method was not only simple, but also could achieve a relative high true positive rate (93%) with a lower false positive which was much better than the results obtained by using either of the above techniques alone.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2002年第6期536-542,共7页
Chinese Journal of Biomedical Engineering
基金
天津市重点学科建设资金资助
