Our student Alex Suer successfully defends his master’s thesis!
We’re pleased to congratulate our graduate student Alexander Suer on successfully defending his master’s thesis, titled “Fourier Neural Operator Pretraining for Physics-Informed Intravitreal Drug Diffusion Prediction.” His research presents a significant advancement in modeling complex drug delivery systems, with a specific focus on intravitreal drug transport. Below, you’ll find an abstract summarizing his findings and approach. We look forward to sharing more comprehensive publications on his work soon - stay tuned!
Abstract:
Complex liquefaction geometries in the vitreous can significantly affect the dynamics of intravitreal drug delivery. This work introduces a Physics-Informed Neural Network (PINN) designed to predict drug transport in the vitreous humor, accounting for these complexities. Additionally, it explores a pretrained foundation model for predicting pressure and velocity fields within a partially liquefied vitreous. Several neural network architectures—including a Feed-forward Neural Network (FNN), ResNet, U-Net, DeepONet, and Fourier Neural Operator (FNO)—were benchmarked for efficiency and accuracy. The proposed pretraining and physics-based fine-tuning strategy using the FNO demonstrated superior accuracy and reduced computation time compared to other models. This approach holds promise for optimizing ocular injection therapies and enabling individualized treatment in intravitreal drug delivery.
