Most studies in computed tomography cannot be reproduced
Summary
The article discusses reproducibility problems in computer tomography research, especially around low-dose and clinical cone-beam CT reconstruction. It compares Smart Tomo Engine baselines with ASTRA-FDK and other reconstruction methods using the ICASSP-2024 challenge dataset. It also highlights dataset validation, cross-dataset testing, and metric choices such as AMSE and masked AMSE. The piece reads like a research and benchmarking update rather than a commercial product announcement.
Classifications
industries
No industries detected
applications
No applications detected
AskAI Classifications
Labels
No AI classifications detected