KCT_cbct Version 1.0 Release

Release of Version 1.0
I am thrilled to announce the official release of KCT_cbct Version 1.0 corresponding to 306th commit and the first commit in 2025

It is a significant milestone in the development of this software package for algebraic CT and CBCT reconstruction. This release reflects years of dedicated effort to create a fast, accurate, and reliable software that meets the needs of researchers and engineers in tomographic imaging.

Key Features of Version 1.0
State-of-the-Art Algorithms: Includes robust implementations of iterative reconstruction algorithms such as CGLS, LSQR, PSIRT, and OS-SART, leveraging insights from recent publications and the latest advancements in the field.
Cutting Voxel Projector (CVP): A highly efficient voxel based projector achieving unprecedented accuracy with optimized OpenCL implementation.
Versatile Projectors: Additional support for TT and Siddon projectors ensures flexibility for diverse reconstruction tasks.
Cross-Platform Support: Fully compatible with OpenCL 1.2, tested across a range of hardware platforms, including AMD, NVIDIA, and Intel GPUs and CPUs, ensuring broad applicability.

Over the past years, substantial development work has been undertaken, including:

• Enhancing Krylov-based reconstruction methods with delayed residual computation for improved accuracy and efficiency.
• Extensive optimization of OpenCL implementations to achieve peak performance on modern hardware.
• Several publications detailing our methodologies and recent innovations are currently in preparation, emphasizing our commitment to advancing the state of the art in tomographic imaging.

Acknowledgments
This project is developed mainly by Vojtěch Kulvait to support and complement my research in the field of tomographic reconstruction.
I thank to Otto-von-Guericke-Universität Magdeburg and Georg Rose for funding my research position in 2018-2021.
I thank Helmholtz-Zentrum hereon GmbH and Julian Moosmann for funding my current research position and I wish to acknowledge the project Holistic Data Analysis (HoliDAy) under the I2B Innovation, Information & Biologisation Fund in the period 2022-2024

Future Directions
Looking ahead, we are excited to propose and work on the following features:

• Multi-GPU Processing: Enabling scalable computations across multiple GPUs for accelerated performance.
• Primal-Dual Hybrid Gradient Methods: Integration of advanced optimization techniques to enhance reconstruction quality and robustness.
• Support for Time-Resolved Datasets: Developing capabilities for processing dynamic datasets, namely those from synchrotron tomography.
• Non-Uniform Data Grids: Expanding support to handle non-uniform grids, offering greater flexibility for complex imaging scenarios.
• Multi node processing: For some tasks even utilization of multiple GPUs on single node might not be sufficient. So we wish to work to enable processing large coupled nonlinear problems, where the memory footprint of the data is in the range of 1TB.
• Python bindings and extensive documentation.

We are actively looking for funding opportunities to enable working on and integrating these features.

Final notes
We extend our gratitude to the contributors, testers, and the community for their invaluable feedback and support in shaping KCT_cbct into a powerful toolset for tomographic imaging.

For details, documentation, and source code, visit:
GitHub Repository
Bitbucket Repository
I look forward to your feedback and collaboration as we continue to innovate and expand the capabilities of KCT_cbct.

KCT_cbct is distributed under the terms of GNU GPL3 license.