乳腺超声图像肿瘤全自动定位方法研究

摘要

Tumor detection in breast ultrasound images is a prerequisite for further segmentation and classification in compu-ter-aided diagnosis ( CAD) system. This paper presented a fully automated method for locating tumors. This method didn' t re-ly on fixed reference position and any initial hard rule and could adapt to the changing relative position of tumors in the ultra-sound images to a large extent. This method had higher accuracy compared to several tumor detection methods.%乳腺超声图像肿瘤的定位是计算机辅助诊断(CAD)系统进行肿瘤分割和良恶性分类处理的前提,为此提出了一种全自动定位肿瘤位置的方法.该方法不依赖初始的固定参考位置和强制性后处理规则,能够较大限度地适应肿瘤在超声图像中相对位置的变化.与目前最好的几种自动定位方法相比,该方法具有更高的定位准确率.