Iterative reconstruction of CT images on GPUs

Model-based Iterative CT Image Reconstruction on GPUs

Computed Tomography (CT) Image Reconstruction is an important technique used in a variety of domains, including medical imaging, electron microscopy, non-destructive testing and transportation security. Model-based Iterative Reconstruction (MBIR) using Iterative Coordinate Descent (ICD) is a CT algorithm that produces state-of-the-art results in terms of image quality. However, MBIR is highly computationally intensive and challenging to parallelize, and has traditionally been viewed as impractical in applications where reconstruction time is critical. We present the first GPU-based algorithm for ICD-based MBIR. The algorithm leverages the recently-proposed concept of SuperVoxels, and efficiently exploits the three levels of parallelism available in MBIR to better utilize the GPU hardware resources. We also explore data layout transformations to obtain more coalesced accesses and several GPU-specific optimizations for MBIR that boost performance. Across a suite of 3200 test cases, our GPU implementation obtains a geometric mean speedup of 4.43X over a state-of-the-art multi-core implementation on a 16-core iso-power CPU.

GPU accelerated voxel-driven forward projection for iterative reconstruction of cone-beam CT

Background

For cone-beam computed tomography (CBCT), which has been playing an important role in clinical applications, iterative reconstruction algorithms are able to provide advantageous image qualities over the classical FDK. However, the computational speed of iterative reconstruction is a notable issue for CBCT, of which the forward projection calculation is one of the most time-consuming components.

Method and results

In this study, the cone-beam forward projection problem using the voxel-driven model is analysed, and a GPU-based acceleration method for CBCT forward projection is proposed with the method rationale and implementation workflow detailed as well. For method validation and evaluation, computational simulations are performed, and the calculation times of different methods are collected. Compared with the benchmark CPU processing time, the proposed method performs effectively in handling the inter-thread interference problem, and an acceleration ratio as high as more than 100 is achieved compared to a single-threaded CPU implementation.

Conclusion

The voxel-driven forward projection calculation for CBCT is highly paralleled by the proposed method, and we believe it will serve as a critical module to develop iterative reconstruction and correction methods for CBCT imaging.