In situ tissue pathology from spatially encoded mass spectrometry classifiers visualized in real time through augmented reality

Integration between a hand-held mass spectrometry desorption probe based on picosecond infrared laser technology (PIRL-MS) and an optical surgical tracking system demonstrates in situ tissue pathology from point-sampled mass spectrometry data. Spatially encoded pathology classifications are displayed at the site of laser sampling as color-coded pixels in an augmented reality video feed of the surgical field of view. This is enabled by two-way communication between surgical navigation and mass spectrometry data analysis platforms through a custom-built interface. Performance of the system was evaluated using murine models of human cancers sampled in situ in the presence of body fluids with a technical pixel error of 1.0 ± 0.2 mm, suggesting a 84% or 92% (excluding one outlier) cancer type classification rate across different molecular models that distinguish cell-lines of each class of breast, brain, head and neck murine models. Further, through end-point immunohistochemical staining for DNA damage, cell death and neuronal viability, spatially encoded PIRL-MS sampling is shown to produce classifiable mass spectral data from living murine brain tissue, with levels of neuronal damage that are comparable to those induced by a surgical scalpel. This highlights the potential of spatially encoded PIRL-MS analysis for in vivo use during neurosurgical applications of cancer type determination or point-sampling in vivo tissue during tumor bed examination to assess cancer removal. The interface developed herein for the analysis and the display of spatially encoded PIRL-MS data can be adapted to other hand-held mass spectrometry analysis probes currently available.
Integration between a hand-held mass spectrometry desorption probe based on picosecond infrared laser technology (PIRL-MS) and an optical surgical tracking system demonstrates in situ tissue pathology from point-sampled mass spectrometry data.