Perfusion MRI: AI methods for DCE-(DSC-)MRI

Quantitative analysis of perfusion MRI traditionally relies on model fitting, which is often computationally demanding and ill-posed, leading to suboptimal or unstable parameter estimates. Our research focuses on leveraging deep learning (DL) approaches to address these challenges by training neural networks to directly estimate key perfusion and permeability parameters from DCE and DCE-DSC data. 

This AI-driven approach enables fast, robust, and accurate parameter quantification, making it more practical for clinical and research applications. Our work in this area has resulted in two conference publications [1,2], where we demonstrated the feasibility and advantages of deep learning models compared to conventional fitting techniques.