Skip to content

RECASTX

RECASTX (REConstruction of Arbitrary Slabs in Tomography X) is a GPU-accelerated software written in modern C++(17). This project has been developed based on a successful proof-of-principle test [1] using RECAST3D in 2019 at the TOMCAT, Swiss Light Source. It aims at providing a near real-time streaming data analysis and visualization tool to allow monitoring tomoscopy experiments effectively. It also serves as the foundation of building a smart data acquisition system which has the potential to reduce the recorded data size by removing trivial or repetitive data while preserving the important scientific information which could lead to scientific discoveries.

References

[1] Buurlage, JW., Marone, F., Pelt, D.M. et al. Real-time reconstruction and visualisation towards dynamic feedback control during time-resolved tomography experiments at TOMCAT. Sci Rep 9, 18379 (2019)

Low-latency

Slab can be defined and reconstructed on-the-fly almost instantly without reconstructing the whole 3D volume.

High-throughput

Throughput reaches up to 3 GB/s of 16-bit raw pixel data (a few tomograms/s) on a middle-end GPU node.

Flexible

Different scan modes as well as configurations of slabs/slices and data processing pipeline are provided.

On-demand slab/slice reconstruction for dynamic experiment

Rich graphical user interface

High-resolution 3D reconstruction for static experiment

Scalable architecture

And more will come

Real-time 3D slab reconstruction

As an enhancement to 2D reconstructed slice, 3D slab will give you localised 3D information in real time.

Segmentation

Segmentation is essential to better scene understanding. Machine learning will play a pivotal role here.

Rendering materials

Rendering voxels with real material properties will greatly enhance 3D visualization.

Higher throughput

We use heterogeneous computing strategy to achieve the highest possible reconstruction throughput. Both CPU and GPU algorithms will be further optimised.

Interactive 2D/3D image analysis

We understand the importance of user experience to online analysis softwares.

Smart data acquisition

This will generate huge impact on data reduction. Funded by SDSC data science projects for large-scale infrastructures.