Turku UAS student’s thesis is nominee for AI Thesis of the Year

AI Finland announced the nominees for the 2025 AI Thesis finalists. Joni Obradovic’s thesis made it to the top three. Obradovic studied Information and Communications Technology Engineering in the Health Technology competence track.

AI Finland, the national AI network, announced the nominees for the AI thesis of the year. Among the top three final thesis nominees was the thesis of Joni Obradovic.

According to AI Finland, the nominees represent a broad range of practical applications, ambitious leadership and groundbreaking research.

In the thesis “Federated brain tumor segmentation with patient-level local differential privacy“, Obradovic developed a method that combines federated learning and local differential privacy for medical imaging.

The solution will train AI models in a distributed way with patient data from different hospitals without the need to transfer sensitive data centrally. The solution improves patient-level privacy by adding controlled noise to the model learning process, and is particularly well suited for applications such as brain tumor segmentation.

Turku UAS develops future-oriented AI research

The Secure and Accurate Federated Learning as a Service (SA-FLaaS) project develops secure and accurate federated learning solutions, especially for health technology and medical data.

Obradovic’s thesis is a good example of how theses that delve into the practices succeed in solving problems or challenges that arise in a real-life environment.