Invasive brain cancer cells cannot be visualized during surgery and so they are often not removed. These residual cancer cells give rise to recurrences. In vivo Raman spectroscopy can detect these invasive cancer cells in patients with grade 2 to 4 gliomas. The robustness of this Raman signal can be dampened by spectral artifacts generated by lights in the operating room. We found that artificial neural networks (ANNs) can overcome these spectral artifacts using nonparametric and adaptive models to detect complex nonlinear spectral characteristics. Coupling ANN with Raman spectroscopy simplifies the intraoperative use of Raman spectroscopy by limiting changes required to the standard neurosurgical workflow. The ability to detect invasive brain cancer under these conditions may reduce residual cancer remaining after surgery and improve patient survival.
It is often difficult to identify cancer tissue during brain cancer (glioma) surgery. Gliomas invade into areas of normal
brain, and this cancer invasion is frequently not detected using standard preoperative magnetic resonance imaging
(MRI). This results in enduring invasive cancer following surgery and leads to recurrence. A hand-held Raman
spectroscopy is able to rapidly detect cancer invasion in patients with grade 2-4 gliomas. However, ambient light sources
can produce spectral artifacts which inhibit the ability to distinguish between cancer and normal tissue using the spectral
information available. To address this issue, we have demonstrated that artificial neural networks (ANN) can accurately
classify invasive cancer versus normal brain tissue, even when including measurements with significant spectral artifacts
from external light sources. The non-parametric and adaptive model used by ANN makes it suitable for detecting
complex non-linear spectral characteristics associated with different tissues and the confounding presence of light
artifacts. The use of ANN for brain cancer detection with Raman spectroscopy, in the presence of light artifacts,
improves the robustness and clinical translation potential for intraoperative use. Integration with the neurosurgical
workflow is facilitated by accounting for the effect of light artifacts which may occur, due to operating room lights,
neuronavigation systems, windows, or other light sources. The ability to rapidly detect invasive brain cancer under these
conditions may reduce residual cancer remaining after surgery, and thereby improve patient survival.
Cancer tissue is frequently impossible to distinguish from normal brain during surgery. Gliomas are a class of brain cancer which invade into the normal brain. If left unresected, these invasive cancer cells are the source of glioma recurrence. Moreover, these invasion areas do not show up on standard-of-care pre-operative Magnetic Resonance Imaging (MRI). This inability to fully visualize invasive brain cancers results in subtotal surgical resections, negatively impacting patient survival. To address this issue, we have demonstrated the efficacy of single-point in vivo Raman spectroscopy using a contact hand-held fiber optic probe for rapid detection of cancer invasion in 8 patients with low and high grade gliomas. Using a supervised machine learning algorithm to analyze the Raman spectra obtained in vivo, we were able to distinguish normal brain from the presence of cancer cells with sensitivity and specificity greater than 90%. Moreover, by correlating these results with pre-operative MRI we demonstrate the ability to detect low density cancer invasion up to 1.5cm beyond the cancer extent visible using MRI. This represents the potential for significant improvements in progression-free and overall patient survival, by identifying previously undetectable residual cancer cell populations and preventing the resection of normal brain tissue. While the importance of maximizing the volume of tumor resection is important for all grades of gliomas, the impact for low grade gliomas can be dramatic because surgery can even be curative. This convenient technology can rapidly classify cancer invasion in real-time, making it ideal for intraoperative use in brain tumor resection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.