Baudoin C E, Lay B J, Klein J C
Rev Epidemiol Sante Publique. 1984;32(3-4):254-61.
A computerized method for the detection of microaneurysms (MA) in fluorescein angiograms is proposed, using the concepts of mathematical morphology. The MA which are almost circular particles, are extracted from the image by performing different "top-hat transformations". Some particles, however, may then be detected inside the nonhomogeneous vessels, and it is necessary to be able to extract the vasculary net. The MA which present fuzzy boundaries (due to the leakage of fluorescein) are the most difficult to extract. The algorithm has been tested on 25 angiograms, with 1 045 MA analysed. A comparison between the automatic counting procedure and three manual methods of counting has been made to prove the robustness of the proposed method.
提出了一种利用数学形态学概念在荧光血管造影中检测微动脉瘤（MA）的计算机化方法。微动脉瘤几乎呈圆形颗粒状，通过执行不同的“顶帽变换”从图像中提取。然而，一些颗粒可能会在不均匀的血管内部被检测到，因此有必要能够提取血管网络。具有模糊边界（由于荧光素泄漏）的微动脉瘤是最难提取的。该算法已在25幅血管造影图像上进行了测试，共分析了1045个微动脉瘤。对自动计数程序和三种手动计数方法进行了比较，以证明所提方法的稳健性。
