# The feasibility and accuracy of real-time intra-operative confocal tissue diagnosis in brain and spine cancer surgery

## TL;DR

This study evaluates a fast confocal microscope for real-time brain and spine tumor diagnosis during surgery, showing promising results for quick and accurate intraoperative decisions.

## Contribution

The study introduces and validates the use of an ultra-fast confocal scanner for real-time intraoperative tissue diagnosis in brain and spine cancer surgery.

## Key findings

- 94% of 47 tissue specimens produced interpretable images within 60 seconds using the Histolog® system.
- Histolog® matched gold standard diagnoses in 68% of cases, comparable to existing intraoperative methods.
- The system enabled real-time tumor cell identification in all specimens.

## Abstract

Real-time intra-operative brain tumour tissue analysis can reduce turnaround times and enable repeated sampling, enhancing diagnostic accuracy and guiding resection. We investigated the use of an ultra-fast confocal microscopy scanner (Histolog®, SamanTree Medical SA) for real-time brain tumour tissue diagnosis. This study aims to demonstrate Histolog®’s diagnostic accuracy in a cohort of brain or spinal cord tumour patients. A single-group, observational study included adult patients undergoing tissue biopsies or tumour debulking surgery. Multiple, freshly excised tissue samples were stained and imaged within 60 seconds using Histolog® alongside standard diagnostic methods. A Consultant Neuropathologist performed a blinded concordance analysis. Of 47 specimens, 94% produced interpretable images during surgery. Histolog® images enabled real-time tumour cell identification in all specimens and matched gold standard diagnoses in 68% of cases. Histolog® demonstrated rapid intraoperative imaging with diagnostic performance for tumour presence that was comparable to existing intraoperative methods in this small cohort, while exact histological concordance was lower (68%). Future studies will explore real-time margin zone analysis and machine learning for automatic diagnosis.