Raman spectroscopy in neurosurgery – assistance in decision making of tumor borders and tumor grade

Introduction

Malignant brain tumors constitute 1,7 percent of all cancers worldwide, with 50 000 cases reported annually in northern Europe. Malignant brain tumors are characterized by diffuse infiltration into normal tissue. This transition zone is difficult or almost impossible to demarcate during surgery with available microscopic techniques making a radical removal of neoplasms difficult. The extent of removal is the single most important factor for survival; the one-year survival rate doubles for cases with macroscopically complete resections. Raman spectroscopy has the sensitivity and potential to help neurosurgeons define cancer cell gradients extending from the brain cancer mass. However, it has not yet been implemented in clinical settings. The cross-disciplinary collaboration between physicists, biomedical engineers, neurosurgeons, and neuropathologists in the Swedish Foundation for Strategic Research (SSF) funded project Med-X External link. with the main seat at Linköping’s University gives the unique opportunity to introduce Raman spectroscopy in neurosurgery.

The aims of this project are to:

• evaluate the methods ex-vivo in tissue samples and intraoperatively during surgical resection of brain tumors,
• increase fundamental knowledge and understanding of the Raman data from brain tumors coupled to neuropathology,
• develop a calibration tool for reliable direct analysis of brain tissue in vivo,
• develop data mining methods and a software that gives real-time results from brain tissue classification,
• investigate the combination of the Raman equipment with the presently used optical, and navigation techniques, g. fluorescence, laser Doppler flowmetry (LDF) and iMRI, in the neurosurgical operation room, see Fig 1.

We foresee a valuable novel tool for neurosurgeons that will complement their toolbox for radical removal of neoplasms without interfering with the current workflow. Our vision is to develop tools that drastically reduce the recurrence of brain cancer.

Preliminary results

Patients scheduled for brain tumor resection at Linköping’s University Hospital who volunteered to participate in the study gave written informed consent. Raman spectroscopic investigations were conducted in a room next to the operating theatre. A homebuilt Raman microscope and a Raman spectrometer with a sterilizable probe head were used on fresh biopsy samples obtained directly from the operating room. The microscope was used first, and research is ongoing using the probe. After measurements, the samples were transferred to neuropathology. Fig. 2 shows the two setups used and a spectrum pretreated by a convolutional neural network (CNN) together with a color-coded Raman peak assignment.

Future work

We are now developing a CNN approach that allows us to simultaneously analyze all spectral data, even if they come from different setups. We are also working on a spectral library to screen the processed Raman data to different proteins, lipids or nucleic acids.

Funding

The Swedish Research Council VR

The Swedish Foundation for Strategic Research SSF

References

1. Wahl J, Klint E, Hallbeck M, Hillman J, Wårdell K, Ramser K. Impact of preprocessing methods on the Raman spectra of brain tissue. Biomedical Optics Express [Internet]. 2022;13(12):6763–77. Available from: https://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-95007 External link.

2. Wahl J, Klint E, Hallbeck M, Hillman J, Wårdell K, Ramser K. Raman spectroscopic analysis of fresh tissue samples from brain tumors. In: European Conferences on Biomedical Optics 2021 [Internet]. Optics Info Base, Optical Society of America; 2021. Available from: https://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-94973 External link.

3. Wahl J, Klint E, Hallbeck M, Hillman J, Wårdell K, Ramser K. Raman spectroscopic analysis of fresh tissue samples from brain tumors. In: Diffuse Optical Spectroscopy and Imaging VIII [Internet]. SPIE - International Society for Optical Engineering; 2021. Available from: https://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-87552 External link.

4. Wahl J, Sjödahl M, Ramser K. Single-Step Preprocessing of Raman Spectra Using Convolutional Neural Networks. Applied Spectroscopy [Internet]. 2020;74(4):427–38. Available from: https://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-77138 External link.