Three-dimensional intraoperative light field fluorescence imaging system for PpIX-guided tumour resection

TL;DR

This study introduces a 3D intraoperative light field fluorescence imaging system that accurately measures tumour fluorescence depth and intensity, improving glioma resection guidance.

Contribution

It develops a dual mode light field imaging system using a commercial camera to capture 3D structure and correct fluorescence attenuation in a single snapshot.

Findings

Achieved millimeter scale vertical resolution with high linearity (R^2 > 0.95).

Demonstrated accurate depth estimation with deviations below 2.5%.

Recovered intrinsic fluorescence concentrations mimicking PpIX in glioma.

Abstract

Conventional 2D fluorescence imaging in glioma surgery cannot separate intrinsic fluorophore strength from attenuation with depth, creating depth-intensity ambiguity that can compromise assessment of residual tumour and fluorescence based grading. This study develops and validates a dual mode light field imaging system that could capture 3D structure and depth corrected fluorescence in a single snapshot by adapting a commercial Lytro Illum camera. A custom 3D printed depth standard was used to optimise main lens focal length and to derive a grayscale - distance linearity from Lytro Desktop depth maps. CdSe/ZnS quantum dot targets and fluorescent brain phantoms were imaged to establish fluorescence intensity distance attenuation models and to recover intrinsic fluorescence. In system optimisation, the increasing FU strengthened grayscale depth linearity and achieved millimetre scale vertical resolution ($R^{2}$ > 0.95) for FU $\ge$ 60 mm. Higher concentration quantum dot wells of the fluorescent target showed consistent attenuation. In fluorescence mode, the deviations of distance estimations across six regions of a fluorescent brain phantom were 0.14 to 2.45% with intensity prediction errors from -11.73% to 6.08% based on the fluorescence intensity-distance model, enabling recovery of intrinsic quantum dot concentrations which are mimicking PpIX characteristics in glioma. This research supports light field imaging as a practical approach for depth resolved quantitative fluorescence and improved intraoperative tumour characterisation.