Authors: Mengzhou Li, Chuang Niu, Ge Wang, Maya R Amma, Krishna M Chapagain, Stefan Gabrielson, Andrew Li, Kevin Jonker, Niels de Ruiter, Jennifer A Clark, Phil Butler, Anthony Butler, Hengyong Yu
Published on: March 19, 2024
Impact Score: 7.2
Arxiv code: Arxiv:2403.12331
Summary
- What is new: Introduces a deep learning-based approach for PCCT image reconstruction at halved dose and doubled speed, alongside a patch-based network for GPU memory limitations.
- Why this is important: Current X-ray PCCT requires improvements in radiation dose and imaging speed, hindered by GPU memory, data scarcity, and domain gap challenges.
- What the research proposes: A patch-based volumetric refinement network that uses synthetic data and model-based iterative refinement to improve image reconstruction.
- Results: Improved image quality in both simulation, phantom experiments, and a clinical trial with 8 patients, validated by radiologists against full-view datasets.
Technical Details
Technological frameworks used: Patch-based volumetric refinement network
Models used: Deep learning models for few-view reconstruction
Data used: Synthetic and real-world patient data
Potential Impact
Could impact medical imaging companies, healthcare providers, and radiology departments by improving PCCT imaging efficiency and safety.
Want to implement this idea in a business?
We have generated a startup concept here: QuickScan AI.
Leave a Reply