1152 Lightfield Snapshot Hyperspectral Imaging for Quantitative Fluorescence-Guided Neurosurgery

INTRODUCTION: Fluorescence-guided surgery with protoporphyrin (PpIX) has shown improved rates of gross total resection. However, visual fluorescence assessments have diagnostic accuracies as low as <70% for detecting residual tumor due light attenuation and scattering. Hyperspectral imaging (HSI) techniques have been developed to measure absolute PpIX concentrations. However, these methods are slow (seconds to minutes/image) due to existing HSI spectral- or spatial-scanning technologies. METHODS: We built a custom HSI fluorescence system for visible to near infrared imaging. We validated an empirical model for correcting tissue optical properties on the measured fluorescence to estimate tissue PpIX concentrations, requiring both reflectance and fluorescence measurements. RESULTS: We used our novel HSI system to image fluorescence and reflectance of tissue mimicking phantoms of known absorption, scattering, and PpIX concentrations . Our empirical correction model showed high linearity (R2 = 0.95) to correct for optical properties unlike the raw fluorescence with known PpIX concentrations . We then validated our HSI system in a rat glioma model to accurately measure PpIX concentrations, where tumor areas that are invisible with conventional raw fluorescence are visualized with our HSI system with estimated absolute concentrations. CONCLUSIONS: Here we demonstrate a novel HSI system for real time quantitative fluorescence image-guided surgery. We optimized and validated our system in pre-clinical phantom and rat glioma studies demonstrating its ability to quantify PpIX and detect PpIX concentrations otherwise ‘invisible’ to conventional fluorescence, which provide support for translating our results to a clinical system.