中文摘要：

目的：乳腺肿块的计算机检测可以帮助医师定位肿瘤，提高乳腺癌诊断的速度和准确率。方法：作者利用AFUM（average fraction under the minimum）^[1]算子和一阶梯度向心率估计检测肿块异常区域。作者对一阶梯度向心率估计做了详细阐述，包括原理和相关参数的选择。结果：作者对40张来自Digital of Screening Mammography（DDSM）乳腺X光图像进行了检测，并将检测结果与图像库的金标准进行比较，画出FROC（false positive receiver operating characteristic）^[1]曲线。平均每幅图像的假阳率约为1.792，肿瘤检出率约为90.63％，每个病例的检测时间约为2min。结论：算法可以检测出大部分的肿瘤，并且每幅图像的假阳率比较低，检测速度非常快。

英文摘要：

Objective： Breast mass computer aided detection may help clinicians to locate breast cancers, and increase its speed and accuracy. Methods： In this paper, the author detects malignant mass employing the method of average fraction under the minimum （AFUM） and the first-order gradient centripetal rate. The author also gives the detailed description of first-order gradient centripetal rate, including principles and selection of parameters. Results： The author detects 40 mammograms from DDSM, compares the results with the golden rules of DDSM, and draws the false positive receiver operating characteristic^[1]（FROC） curves. More than 90.63% of breast cancers are detected, while the average false positives of an image are 1.792 and the average detection time for a case is about 2 minutes. Conclusions： The algorithm can find most breast cancers for clinicians, while the false positives of this algorithm are very low, the speed is very fast.