CWI and University College London received an EU ATTRACT grant to deploy and develop AI techniques to improve X-ray CT scanning in such a way that much lower doses of radiation can be used in the future to obtain high-resolution images. The project started in July 2019 and will last a year.
Charlotte Hagen, researcher from the UCL Department of Medical Physics and Biomedical Engineering says in a video on the ATTRACT website: “This project is about developing a new X-ray micro CT-technique that will allow performing high-resolution scans with significantly less radiation dose than what is currently possible. This would be a real break-through, as it would allow way more extensive high-resolution scanning in biomedicine, where samples are typically highly radiation sensitive. To achieve this, we will combine advanced X-ray engineering with machine learning. The X-ray beam in our scanner will be split into an array of beamlets, which will shield large parts of the sample from radiation. We will then combine this setup with a new cycloidal scanning scheme by which the sample is translated and rotated at the same time and exploit new directions in machine learning to reconstruct high-resolution images”.
Daan Pelt, researcher at in the Computational Imaging research group at CWI adds: “Less radiation dose usually means less resolution but with the new learning algorithms we want to develop and combine with the new scanning procedure, we can obtain high-resolution images from low-dose data. In addition to improving our experimental setup of the CWI FleX-Ray lab, this grant will allow us to install a dedicated multi-GPU machine for testing and applying the new machine learning techniques.”
The international research team writes: “The breakthrough character of this project was demonstrated in two proof-of-principle studies. The first study has shown that our cycloidal acquisition approach of obtaining data can increase the spatial resolution in X-ray micro-CT images by at least a factor of three without requiring any increase in radiation dose. The second study has demonstrated that convolutional neural networks are well-suited for application in low-dose computed tomography and lead to a significantly improved image quality. We expect that, by combining both aspects, we will be able to increase the spatial resolution in micro-CT images much further than what we have already demonstrated, with no increase in radiation dose”.
CWI researcher Daan Pelt.
More information
- ML-CYCLO-CT project web page on 'Combining cycloidal computed tomography with machine learning: a mechanism to disrupt the costly relationship between spatial resolution and radiation dose'
- CWI’s Computational Imaging research group
- Daan (Daniël) Pelt’s contact data at CWI
- University College London and the UCL Department of Medical Physics and Biomedical Engineering
- Video Computational Imaging research group
