Motion Correction In Digital Subtraction Angiography
ID U-6842
Category Imaging (Software)
Subcategory CT
Researchers
Brief Summary
Motion Correction in Digital Subtraction Angiography Using Generative Adversarial Networks (GAN)
Problem Statement
Patient motion in digital subtraction angiography (DSA) can render background subtraction difficult, limiting the images' diagnostic quality. As a result, the procedure is often repeated leading to increased healthcare costs, procedure times, radiation dose, and contrast consumption.
Technology Description
Motion artifacts in digital subtraction angiograms (DSA) can be reliably eliminated in real-time by using generative adversarial networks (GAN) to produce subtractions directly from the post-contrast fluoroscopic images, without the use of a pre-contrast mask. The intent is to incorporate the technology into medical imaging equipment to improve resultant diagnostic images during motion sensitive procedures.
Stage of Development
Build
Benefit
- Improved diagnostic image quality
- Reduced healthcare costs
- Reduced procedure times
- Reduced radiation dose and contrast use.
Publications
Crabb, et al. Deep Learning Subtraction Angiography: Improved Generalizability with Transfer Learning. J Vasc Interv Radiol. 2023 Mar;34(3):409-419.e2. doi: 10.1016/j.jvir.2022.12.008. Epub 2022 Dec 16. PMID: 36529442.
Outstanding Laboratory Investigation Award from the Journal of Vascular and Interventional Radiology (JVIR) 2023
Contact Info
Huy Tran
(801) 581-7792
huy.tran@utah.edu