Quality assesment in clinical brain CT
In emergency brain imaging, CT is often the first and most time-sensitive imaging tool for detecting stroke, trauma, or bleeding. Any reduction in image quality, whether caused by motion, metallic implants, or incorrect technical settings, can hinder the visibility of critical findings. If such problems are detected only after the patient has left the scanner, another scan may be necessary. This leads to delays in treatment decisions which is especially problematic when seconds count.
To prevent such scenarios, it is highly desirable to automatically evaluate, document, and report image quality immediately after scan acquisition.
In a related but distinct direction, the widespread adoption of AI-based reconstruction and diagnostic algorithms raises a new challenge: the need to objectively assess the quality of AI-generated results. One promising solution is the development and evaluation of advanced clinical imaging phantoms, which can be used to benchmark system performance and ensure reliable and reproducible outcomes across different CT devices.
Potential Thesis Topics
Brain CT image quality:
- Automated clinical image quality assessment using artificial intelligence
- Atlas-based or deep learning-based segmentation methods for on clinical brain
- Efficient segmentation using AI methods
- Noise characterization and analysis in clinical brain CT
Phantom development and system evaluation: - GEANT4-based CT system simulation
- Development of novel CT imaging phantoms for quality control
- Developement and evaluation of AI reconstruction techniques
CT hardware and system optimization:
- Projects related to CT hardware and acquisition optimization are planned but will only be available once the new system is installed. Specific thesis topics in this category will be announced later.
