基于Spark的CT图像FBP重建算法程序并行设计

Parallel Design of FBP Reconstruction Algorithm for CT Image Based on Spark

将常用于CT图像重建的滤波反投影算法程序设计成能够运行在大数据框架Spark中的并行模式,以此来提高计算效率并实现批量图像的重建,缩短图像重建时间。基于分布式计算框架Spark,利用其图像处理工具Thunder,将滤波反投影算法在图像重建过程中设计成并行程序模式,实现图像的片间并行重建。实验结果表明,随着Spark集群规模的不断扩大,在确保重建图像质量的前提下,重建一定数量的CT图像相比单机模式下时间显著缩短,并行滤波反投影算法具有完全加速比,并行效率趋近于1。基于Spark集群实现的滤波反投影算法能够显著提升CT图像重建速度,并实现大量图像并行重建,可扩展其他的CT图像重建算法,对远程医学图像重建平台的建设具有重要参考意义。

To design a filtered back projection algorithm, which is often used in CT image reconstruction and can run in the parallel mode of the big data framework Spark, so as to improve the computational efficiency of the algorithm and realize the reconstruction of batch images and shorten the image reconstruction time. Based on the distributed computing framework Spark, the image processing tool Thunder is used to design the filtered back projection algorithm into a parallel program mode in the image reconstruction process. Experiments show that with the continuous expansion of the Spark cluster size, under the premise of ensuring the quality of reconstructed images, the reconstruction of a certain number of CT images is significantly shorter compared to the stand-alone mode, and the parallel filtered back projection algorithm has almost complete speedup. The filtered back projection algorithm based on Spark cluster can significantly improve the CT image reconstruction speed and achieve a large number of parallel image reconstructions. It can expand other CT image reconstruction algorithms, and has important reference significance for the construction of remote medical image reconstruction platform.