Spectroscopic optical coherence tomography for ex vivo brain tumor analysis

For neurosurgeries precise tumor resection is essential for the subsequent recovery of the patients since nearby healthy tissue that may be harmed has a huge impact on the life quality after the surgery. However, so far no satisfying methodology has been established to assist the surgeon during surgery to distinguish between healthy and tumor tissue. Optical Coherence Tomography (OCT) potentially enables non-contact in vivo image acquisition at penetration depths of 1-2 mm with a resolution of approximately 1-15 μm. To analyze the potential of OCT for distinction between brain tumors and healthy tissue, we used a commercially available Thorlabs Callisto system to measure healthy tissue and meningioma samples ex vivo. All samples were measured with the OCT system and three dimensional datasets were generated. Afterwards they were sent to the pathology for staining with hematoxylin and eosin and then investigated with a bright field microscope to verify the tissue type. This is the actual gold standard for ex vivo analysis. The images taken by the OCT system exhibit variations in the structure for different tissue types, but these variations may not be objectively evaluated from raw OCT images. Since an automated distinction between tumor and healthy tissue would be highly desirable to guide the surgeon, we applied Spectroscopic Optical Coherence Tomography to further enhance the differences between the tissue types. Pattern recognition and machine learning algorithms were applied to classify the derived spectroscopic information. Finally, the classification results are analyzed in comparison to the histological analysis of the samples.