论文部分内容阅读
以肺癌患者计算机断层扫描图像中的肿瘤追踪为例,提出一种基于稀疏分解和支持向量机的新型医学影像内目标物体追踪算法被提出。首先,split-Bregman理论被运用在分解医学影像中:每一幅医学图像都被分解成一副含有较多运动信息的低秩图像和一幅含有较多噪声信息的稀疏图像。其次,支持向量机被运用在低秩图像中,使得低秩图像内的像素点被区分为来自肿瘤区域的像素点和来自非肿瘤区域的像素点,从而达到在医学影像中分割和追踪肿瘤的目的。实验表明,该算法在肺癌患者医学影像的肿瘤追踪实例中能取得较其他比较方法更为准确的追踪效果。
Taking tumor tracking in computed tomography of lung cancer patients as an example, a new target tracking algorithm based on sparse decomposition and support vector machines is proposed. First, the split-Bregman theory is used to decompose medical images: each medical image is decomposed into a low-rank image containing more motion information and a sparse image containing more noise information. Secondly, SVM is used in low-rank images so that the pixels in the low-rank image are divided into the pixels from the tumor area and the pixels from the non-tumor area so as to achieve the goal of dividing and tracking the tumor in the medical image purpose. Experiments show that this algorithm can obtain a more accurate tracking result than the other comparative methods in the case of tumor tracking of medical images of lung cancer patients.