ÄûÃʵ¼º½

The title of the thesis: Deep learning-based metal and scatter artifact reduction in cone-beam computed tomography 

Thesis defender: Harshit Agrawal
Opponent: Prof. Samuli Siltanen, University of Helsinki, Finland
Custos: Prof. Simo Särkkä, Aalto University School of Electrical Engineering

Cone-Beam Computed Tomography (CBCT) is an advanced X-ray imaging technique that produces high-quality three-dimensional images with lower radiation exposure, reduced cost, and a smaller physical footprint compared to conventional CT scanners. These advantages make CBCT valuable across a range of clinical applications, including dentistry, orthopedics, interventional radiology, and image-guided therapies, as well as in mobile and remote healthcare settings. 

However, CBCT images are often degraded by distortions caused by high-density metal objects and scattered radiation, which can limit diagnostic accuracy. This thesis presents deep learning-based methods to address these challenges, leveraging simulated training data and lightweight models designed for seamless integration with existing systems. The proposed methods effectively reduce artifacts and enhance image quality in real time, making CBCT more reliable, practical, and suitable for broader clinical adoption.

Thesis available for public display 7 days prior to the defence at . 

Contact: harshit.agrawal@aalto.fi