Medical Image Reconstruction: From Model-Based to Data-Driven

This lecture by the instructor covers the evolution of medical image reconstruction techniques, starting from classical model-based approaches to the recent data-driven methods using deep neural networks. The presentation delves into topics such as variational formulation, compressed sensing, sparsity evolution, and the application of deep learning in medical imaging. The lecture also explores the challenges in image reconstruction, the concept of sparsity constraints, and the use of iterative algorithms for compressed sensing. Additionally, it discusses the emergence of deep convolutional neural networks for bioimage reconstruction and their potential applications in various imaging modalities.