基于GPGPU的实时结肠虚拟展平技术
GPGPU Based Real-Time Virtual Colon Flattening
-
摘要: 为了在结肠检查过程中实时绘制结肠内壁表面的展平图像,并动态地调整分割阈值,提出一种基于GPGPU的实时结肠虚拟展平技术.首先在结肠中心路径上每一点的切线方向建立圆柱形的光照模型,然后根据当前CT分割阈值,并通过GPGPU加速光线投射算法提取亚像素级的结肠内壁表面,最终得到局部虚拟展平的结果.实验结果表明,文中技术渲染速度达到50帧/s的实时要求,在结肠检查过程中能够通过动态地调整分割阈值获取精细的结肠内壁,提高了虚拟展平算法的适应性和准确性.Abstract: This paper presents a GPGPU-based real-time virtual colon flattening method that allows dynamic modification of the CT segmentation threshold during a colon observation.First,a cylindrical lighting model is built by leveraging the tangent of each point on the colon's center line.Then a GPGPU-based ray-casting volume rendering algorithm is employed to generate the flatten colon surface according to the CT segmentation threshold of the colon's inner surface.Experimental results show that the proposed method achieves real-time performance(up to 50 frames per second),and improves the adaptability and accuracy of the virtual colon flattening.