AI in Biophotonics & Optical Systems
Inverse Modeling and Uncertainty-Aware AI for Biophotonics
My research focuses on inverse modeling, uncertainty quantification, and multimodal integration for the analysis of biophotonic data, combining physical modeling with machine learning.
Research Projects
Inverse Reconstruction of Optical Measurements
Problem
Optical measurements are indirect and noisy, making reconstruction an ill-posed inverse problem.
Approach
AI-based and physics-informed models with uncertainty estimation.
Status
Active – code available (prCARS, CARSBench)
This project focuses on solving ill-posed inverse problems arising from indirect and noisy optical measurements. The approach combines physics-informed modeling with AI-based methods to reconstruct underlying structures while accounting for uncertainty. The goal is to achieve more stable and interpretable reconstructions in complex optical systems.
Uncertainty-Aware Inverse Modeling in Biophotonics
Problem
Standard AI models do not quantify uncertainty, limiting reliability in biomedical settings.
Approach
Probabilistic modeling and uncertainty quantification methods for robust predictions.
Status
Active – code available (CARSGuard)
This work addresses the lack of reliability in standard AI models by incorporating uncertainty quantification into inverse modeling. Probabilistic approaches provide not only predictions but also confidence estimates, which are crucial in biomedical applications. This aims to improve trust and robustness in data-driven biophotonic analysis.
Multimodal Integration of Biological Data
Problem
Biological systems are observed through heterogeneous data sources across scales.
Approach
Integration of imaging, signal, and omics data for cross-scale modeling.
Status
Active – exploratory results available
This research direction focuses on connecting optical, spectral, and imaging measurements across scales. Current exploratory work uses reconstruction and representation-learning workflows for biomedical imaging, with the aim of extending these methods toward the integration of molecular and omics data (e.g. genomics, RNA-seq).
Connect for Collaboration
Interested in collaboration in inverse modeling, uncertainty quantification, or multimodal analysis for biophotonic systems?
Open to research collaborations, student projects, and interdisciplinary work.