Research Interests

-

I study neuroinflammatory processes in multiple sclerosis using quantitative PET/MRI neuroimaging, focusing on how reliable representations and quantitative measures can be constructed from high-dimensional, noisy, and data-limited imaging data.

-

My work focuses on defining and robustly measuring clinically meaningful imaging biomarkers from imperfect imaging observations collected under real-world clinical constraints. In particular, I study how modeling choices, data representations, and learning strategies influence what can be inferred about disease biology from complex multimodal imaging data.

Neuroimaging with PET and MRI provides a natural setting for these questions, as these data are inherently high-dimensional and multimodal while clinical practice often limits data quantity and prevents idealized measurement protocols. My research therefore aims to develop quantitative and representation-level approaches that remain reliable, interpretable, and transferable to real clinical environments.

A central goal of my work is enabling biologically meaningful inference when gold-standard measurements are unavailable or impractical. To address this, I integrate quantitative imaging models, domain knowledge, and computational methods, including machine learning, to design measurement frameworks that support biological and clinical questions across PET and MRI.

In my current postdoctoral work on multiple sclerosis, these ideas are applied to studying lesion evolution and inflammatory activity, where representation design and modeling assumptions directly influence biomarker estimation and disease interpretation under clinical data constraints.