Reconstruction

For users who appreciate the importance of high-quality images to reliable quantitative analysis, SCANCO Medical systems are the first choice. Our standard acquisition and reconstruction protocols handle very large numbers of projections, with small angular separation, to deliver images with high signal-to-noise ratio and high resolution, essential characteristics for accurate, reproducible quantitative 3D analysis.

Tomographic image reconstruction is an inherently parallel operation. Reconstruction tasks are distributed automatically among all available processors in microCT workstations, and run in parallel to scanning and other tasks. In microCT systems with multiple computers in a cluster, reconstruction workload is spread among more processors, further reducing total time required for reconstruction of all slices in a given scan.

GPU Accelerated Reconstruction

The SCANCO GPU Accelerated Reconstruction System is an optional module for faster image reconstruction. The SCANCO GPU Reconstruction software is fully integrated in the scanner workflow and works automatically without any user intervention.

The system will speed up the calculation time up to 120 times in comparison with the standard reconstruction software running on a single core. It additionally frees resources leaving more computer power for analysis tasks.

Images reconstructed with the SCANCO GPU Accelerated Reconstruction System are fully compatible with images reconstructed with the standard CPU based reconstruction software.

The SCANCO GPU Accelerated Reconstruction software typically runs on the latest NVIDIA GPU Accelerators  (Tesla, Kepler, Maxwell, Pascal, Volta etc. architecture).

Typical configuration
HP Desktop Workstation, 64 bit Z4 Z8
RAM 64 GB 128 GB
CPU (cores) 1 (6) 2 (24)
Disks 1 TB SSD 1 TB SSD
GPU NVIDIA RTX A4000 NVIDIA RTX A6000
Image Reconstruction
Reconstruction
NVIDIA GPU Accelerator
Nvidia GPU
GPU gain

Graph showing the gain in speed between the standard CPU reconstruction operating on a one core CPU in comparison with the GPU reconstruction running on a NVIDIA GPU Accelerator. The speed gain is shown for different sizes of the image matrix.