Raman spectroscopy, a photonic modality based on inelastic backscattering of light, is an asset to the intraoperative sensing space, offering non-ionizing potential and highly-specific spectroscopic signatures that can be used for diagnosis of pathological tissue in the dynamic surgical field. Though Raman is an inherently weak process, Surface-Enhanced Raman Spectroscopy (SERS), which uses metal nanostructures to amplify Raman signals, can achieve detection sensitivities that rival traditional photonic modalities. Our lab has  developed SERS-based Gold Nanostars that accumulate preferentially in certain tumors, allowing for discernment of pathological tissue.

However, there remains a need for a robotic Raman system and algorithm that can use Gold Nanostars to efficiently reconstruct a region of interest, to aid surgeons as they seek to improve extent-of-resection and overall patient survival. In this study, we outline a robotic Raman system that can reliably pinpoint the location and boundaries of a tumor embedded in healthy tissue, modeled here as a brain-mimicking phantom with selectively infused Gold Nanostar regions. Further, due to the dearth of collected biological SERS or Raman data, we implement transfer learning and achieve 100% validation classification accuracy for Gold Nanostars compared to Control Agarose, thus providing a proof-of-concept for Raman-based deep learning pipelines.

We reconstruct a surgical field of 30x60mm in 10.2 minutes, and achieve 98.2% accuracy, preserving relative measurements between phantom features. We also achieve an 84.3% Intersection-over-Union, which is the extent of overlap between the ground truth and predicted reconstructions. Lastly, we also demonstrate that the Raman system and classification algorithm do not discern based on sample color, but instead on presence of SERS agents. This study provides a crucial step in the translation of robotic Raman systems in intraoperative oncological spaces.