Publications by authors named "Luka Vetoshkin"
Front Med (Lausanne)
September 2024
Article Synopsis
- The study aims to explore AI methods for reducing PET acquisition time, enhancing patient comfort and scanner efficiency while maintaining image quality.
- Researchers evaluated various neural network models for PET denoising using a large dataset of 56,908 images, including different model architectures like 2D ResNet, Unet, and 3D MedNeXt.
- The 3D MedNeXt model was found to be the most effective, significantly improving image quality metrics during PET scans, particularly in 30-second and 60-second acquisitions compared to the original 90-second scans.
