摘要
在乳腺钼靶X片计算机辅助诊断中,感兴趣区域也就是可疑病灶区的自动提取是乳腺图像处理中的重点和难点之一,本研究旨在提出一种准确、有效的算法,以提高乳腺癌早期诊断的效率和准确率。方法:首先对感兴趣区域的多种特征进行了分析,然后提出一种基于神经网络新训练算法的乳腺钼靶x片感兴趣区域的自动提取方法。结果:用我们的算法对山东省医学影像研究所提供的30个临床实际病例,60幅图像进行了检测和分析。实验结果验证了该算法的有效性和准确性。结论:该算法在保持较低的假阳性率的同时,能得到高的阳性检出率。
In mammography processing, the extraction of region of interest is one of the most difficult problem. this paper presents a new efficiency method for diagnosing the early cases of breast cance. We firstly analyze many characteristics of region of interest, then, we suppose an automatic extracting method of region of interest in mammography based on neural network. We analyze 20 cases and 40 image from shandong Medical imaging research institute.experimental results demonstrate better extracting effect.The test results showed that the method could achieve a relative high true positive rate（TPR）with a lower false positive.
出处
《中国医学装备》
2008年第2期27-31,共5页
China Medical Equipment
基金
山东省教育厅科技基金资助(NO.J06G64)
