Daniel Castaño Díez | PubMed Health

PubMed2006

Implementation and performance evaluation of reconstruction algorithms on graphics processors.

The high-throughput needs in electron tomography and in single particle analysis have driven the parallel implementation of several reconstruction algorithms and software packages on computing clusters. Here, we report on the implementation of popular reconstruction algorithms as weighted backprojection, simultaneous iterative reconstruction technique (SIRT) and simultaneous algebraic reconstruction technique (SART) on common graphics processors (GPUs). The speed gain achieved on the GPUs is in the order of sixty (60x) to eighty (80x) times, compared to the performance of a single central processing unit (CPU), which is comparable to the acceleration achieved on a medium-range computing cluster. This acceleration of the reconstruction is caused by the highly specialized architecture of the GPU. Further, we show that the quality of the reconstruction on the GPU is comparable to the CPU. We present detailed flow-chart diagrams of the implementation. The reconstruction software does not require special hardware apart from the commercially available graphics cards and could be easily integrated in software packages like SPIDER, XMIPP, TOM-Package and others.

Algorithms↗

PubMed2006

Tilt-series and electron microscope alignment for the correction of the non-perpendicularity of beam and tilt-axis.

In electron tomography the sample is tilted in the electron microscope and projections are recorded at different viewing angles. In the correct geometric setting, the tilt-axis of the object under scrutiny is perpendicular to the beam direction. However, we will demonstrate that this does not necessarily apply to all electron microscopes equipped with the default column alignment. The resulting effect is that a conical tilt is performed, which has to be considered in the reconstruction to avoid artifacts and to improve the resolution. A novel solution, with significantly improved convergence properties, will be introduced for calculating the three-dimensional marker model, which is necessary for the alignment of the tilt-series. Thereby, the angle between the beam direction and the tilt-axis is calculated, together with other geometrical distortions, like magnification and rotation changes, and incorporated in the reconstruction. Hereby, artifacts can be eliminated at the image processing basis, and the resolution can be significantly improved at the medium to high range frequencies. Synthetical and real data are used to demonstrate the obstructions caused by this effect and the quality improvement of the reconstructions. Finally, we also present a way to align the hardware of the microscope to correct for the non-perpendicularity between the beam direction and the tilt-axis, which is specifically tailored for tomographic applications.

Image Processing, Computer-Assisted↗