Development and characterization of a handheld hyperspectral Raman imaging probe system for molecular characterization of tissue on mesoscopic scales

Purpose Raman spectroscopy is a promising cancer detection technique for surgical guidance applications. It can provide quantitative information relating to global tissue properties associated with structural, metabolic, immunological, and genetic biochemical phenomena in terms of molecular species including amino acids, lipids, proteins, and nucleic acid ( DNA ). To date in vivo Raman spectroscopy systems mostly included probes and biopsy needles typically limited to single‐point tissue interrogation over a scale between 100 and 500 microns. The development of wider field handheld systems could improve tumor localization for a range of open surgery applications including brain, ovarian, and skin cancers. Methods Here we present a novel Raman spectroscopy implementation using a coherent imaging bundle of fibers to create a probe capable of reconstructing molecular images over mesoscopic fields of view. Detection is performed using linear scanning with a rotation mirror and an imaging spectrometer. Different slits widths were tested at the entrance of the spectrometer to optimize spatial and spectral resolution while preserving sufficient signal‐to‐noise ratios to detect the principal Raman tissue features. The nonbiological samples, calcite and polytetrafluoroethylene ( PTFE ), were used to characterize the performance of the system. The new wide‐field probe was tested on ex vivo samples of calf brain and swine tissue. Raman spectral content of both tissue types were validated with data from the literature and compared with data acquired with a single‐point Raman spectroscopy probe. The single‐point probe was used as the gold standard against which the new instrument was benchmarked as it has already been thoroughly validated for biological tissue characterization. Result We have developed and characterized a practical noncontact handheld Raman imager providing tissue information at a spatial resolution of 115 microns over a field of view >14 mm 2 and a spectral resolution of 6 cm −1 over the whole fingerprint region. Typical integration time to acquire an entire Raman image over swine tissue was set to approximately 100 s. Spectra acquired with both probes (single‐point and wide‐field) showed good agreement, with a Pearson correlation factor >0.85 over different tissue categories. Protein and lipid content of imaged tissue were manifested into the measured spectra which correlated well with previous findings in the literature. An example of quantitative molecular map is presented for swine tissue and calf brain based on the ratio of protein‐to‐lipid content showing clear delineations between white and gray matter as well as between adipose and muscle tissue. Conclusion We presented the development of a Raman imaging probe with a field of view of a few millimeters and a spatial resolution consistent with standard surgical imaging methods using an imaging bundle. Spectra acquired with the newly developed system on swine tissue and calf brain correlated well with an establish single‐point probe and observed spectral features agreed with previous finding in the literature. The imaging probe has demonstrated its ability to reconstruct molecular images of soft tissues. The approach presented here has a lot of potential for the development of surgical Raman imaging probe to guide the surgeon during cancer surgery.