【摘要】 医学图像分割是各种医学图像应用的基础,但由于医学图像噪声复杂,纹理低,使得分割极为困难。且目前大多的图像分割算法,忽略像素点的位置信息,应用包含位置信息的Turbo-pixel超像素对CT图像进行处理,后期结合少量的人工标记信息,可完整提取出CT图像的软组织信息,为医生减轻负担,且根据专家目测表明,改进的Turbo-pixel分割提取的软组织较为完全且位置信息正确。更多还原

【Abstract】 Medical image segmentation is the basis for various medical imaging applications. However, it is extremely difficult to segment the medical image due to the complex noise and low grain. Currently, the position information of pixel is ignored by most of the image segmenta-tion algorithm. Turbo-pixel super pixel is used to process the CT image containing location information, and then the soft tissue informa-tion is extracted with little manual tagging information, reduces the burden for doctors. Visual inspection of experts shows that the soft tis-sue extracted by improved Turbo-pixel segmentation is more complete and the location information is correct.更多还原