Low grade glioma (LGG) is a slowly evolving, highly invasive intrinsic brain tumor displaying only subtle tissue differences with the normal surrounding brain, hampering the attempts to visually discriminate tumor from normal brain, especially at the border interface. This makes anatomical borders hard to define during early maximal resection, which is the initial treatment strategy. Therefore, innovative, robust and easy-to-use real-time strategies for intra-operative detection and discrimination of (residual) LGG tumor tissue would strongly influence on-site, surgical decision making, enabling a maximal extent of resection.

To validate this approach hyperspectral imaging (HSI) - using a SnapScan HSI-Camera (IMEC), stably mounted on an OPMI Pentero 900 microscope (Zeiss) - will be used to generate spectral imaging data patterns that discriminate in vivo low grade glioma tissue from normal brain both on the cortical and subcortical level.

Included patients will undergo a resection of the low grade glioma as standard-of-care. Before, during and after the resection, HSI data ('datacubes') will be acquired by the SnapScan HSI camera on the microscope of all relevant areas of the exposed cortical surface and subcortical cavity walls. The exact points of which the datacubes will be acquired are defined by unequivocal single points on the neuronavigational system (Brainlab). From the points from which the datacubes have been obtained a corresponding tissue sample will be obtained (labeled biopsy) if tumor tissue is to be expected in that particular point, based on the current standard of care assessments intraoperatively using white light illumination on the microscope, intraoperative navigation and intraoperative ultrasound. As such, normally looking brain in the resection cavity wall, will only be biopsied if tumor free margins should be proven as part of the standard-of-care operative procedure (non-critically eloquent brain regions). The objective of this all is to get an initial high quality in vivo dataset to start exploring the potential of the technology.

The project will follow a 'stop and go' design: during the first 9 months, the initially collected spectrally corrected datacubes will be analyzed using machine learning on coded data sets. After this initial phase, an interim analysis will be made from the full list of analyzed datacubes. If a reliable and robust discriminative signal can be detected in low grade glioma tissue, segregating these signals from those in normal tissue (as defined pathologically and/or radiologically), efficacy is demonstrated (proof of concept) and the trial will go on for further collecting of samples in the following 26 months. Within the expanded dataset, the different spectral data patterns will be translated into user's friendly pattern codes for rapid real-time, on-site detection and interpretation through development of dedicated software. If no reliable signal can be retrieved from low grade glioma tissue in vivo during the surgery, further recruitment of patients will be stopped. At that time, the investigators and partners will decide on whether or not relevant amendments to the study will be proposed or not.

Inclusion Criteria:

• Age ≥ 18 years
• Radiologically suspected low grade glioma (newly diagnosed or recurrent)
• Scheduled for tumor resection at UZ Leuven
• Signed informed consent document prior to resection

Exclusion Criteria:

• Children with age < 18 years
• If final pathology reveals other pathological diagnosis than low grade glioma, datacubes will not be included in the final analysis