Speaker: Dimitris Tryfonopoulos, PhD Candidate in Biomedical Sciences, University of Antwerp
Date: Thursday, 13-03-2025
Time: 11:30 - 12:30
Location: Online Seminar
Title: Deep CNN-MRI reconstruction using a learnable regularization
Host: George Tzagkarakis, FORTH-ICS, SPL
Abstract:Advancements in biomedical imaging and artificial intelligence (AI) have significantly enhanced MRI reconstruction techniques, leading to faster acquisition times and improved image quality. This work presents an integrated framework combining MRI pulse sequence design with deep learning-based reconstruction methods. By leveraging convolutional neural networks (CNNs), the proposed approach accelerates MRI reconstruction while preserving structural fidelity and enhancing signal-to-noise ratios. The methodology is validated on real-world MRI datasets, demonstrating its effectiveness in producing high-resolution images with reduced scanning times. Comparative analyses highlight the advantages of AI-driven reconstruction over conventional methods, paving the way for more efficient and accurate MRI workflows in clinical and research settings.
Short Bio: Dimitris Tryfonopoulos is a Data Scientist and Technical Project Manager specializing in biomedical engineering, AI-driven medical imaging, and statistical modelling. Dimitris holds an MSc in Biomedical Engineering from Telecom Paris Tech University and in Data Science & Machine Learning from NTUA. He is a PhD candidate in Biomedical Sciences at the University of Antwerp.