Selected research and engineering work. Grounded in first-principles thinking, aimed at real-world impact.
Coronary artery disease is the world's leading cause of death. Current intravascular OCT shows where plaques are — but not what they're made of. Plaque composition (lipid-rich necrotic core, fibrous cap, calcification) is the primary determinant of rupture risk, yet this information is invisible to standard OCT.
Spectroscopic OCT can extract compositional data from wavelength-dependent tissue attenuation, but coherent speckle noise corrupts spectral estimates — a barrier blocking clinical translation for over a decade.
Real-time, label-free, compositionally-specific intravascular imaging could transform interventional cardiology — enabling risk stratification before rupture rather than responding after an acute MI. This work is a direct building block toward spectroscopic OCT as standard-of-care. The computational techniques are also broadly applicable to any coherent imaging system requiring quantitative spectroscopic characterization.
Label-free discrimination of lipids and collagen — key disease markers in cardiovascular and connective tissue pathology — requires spectroscopic imaging in the NIR-II window (1000–1700 nm). No compact, fiber-based, multi-wavelength system existed for this regime, creating a fundamental hardware bottleneck.
This platform enables label-free molecular imaging in a spectral window previously inaccessible to compact systems. The combination of custom hardware and quantitative algorithms opens pathways to clinical diagnostics, industrial materials inspection, and any application requiring chemically-specific optical characterization without contrast agents. The all-fiber architecture is a critical step toward translation from bench to clinic and industry.
Optical coherence tomography uses broadband light interferometry for high-resolution cross-sections at micron scale. Spectroscopic extensions exploit frequency-domain content to extract wavelength-dependent material properties — yielding chemical contrast, not just structure.
Combines the molecular specificity of optical spectroscopy with the penetration depth of ultrasound. Pulsed laser excitation generates acoustic waves from target chromophores. Multi-spectral acquisition enables unmixing of molecular species without exogenous labels.
The second near-infrared window (1000–1700 nm) offers reduced scattering and absorption relative to visible wavelengths, enabling deeper penetration into materials and tissue. Distinct spectroscopic signatures in this regime make it ideal for label-free molecular discrimination.
Coherent imaging modalities suffer from speckle noise — a multiplicative interference artifact that corrupts spectral content. Mitigating speckle while preserving spectroscopic fidelity has been a fundamental barrier to quantitative spectroscopic imaging for over a decade.
These projects are building blocks — the techniques generalize well beyond their original domains.
Extending spectroscopic and photoacoustic methods from controlled environments to real-world conditions — addressing motion artifacts, depth requirements, and real-time computation.
Combining structural imaging with molecular contrast and computational methods to produce co-registered, multi-parametric maps — structural and chemical information in a single acquisition.
Translating spectroscopic imaging algorithms to industrial materials characterization, energy systems, and advanced manufacturing — label-free, non-destructive chemical sensing at scale.