Photonics/Optics

An optical probe that detects cancerous brain cells in real time is impressive enough. Scientists in Montreal say they’ve developed one that is “infallible.”

A team, led by researchers at Polytechnique Montréal, the University of Montreal Hospital Research Center (CRCHUM), the Montreal Neurological Institute and Hospital (The Neuro), and McGill University, has updated the handheld spectroscopy device to additionally detect colon, lung, and breast cancer cells.

Dr. Kevin Petrecca and Dr. Frédéric Leblond.

The technology uses a multimodal combination of spectroscopy techniques to interpret the molecular composition of the cells. The measurement methods are integrated into a single sensor, in combination with stimulating lasers, a highly sensitive camera, and a spectrometer.

In 2015, Dr. Kevin Petrecca, Chief of Neurosurgery at the Montreal Neurological Institute, and Dr. Frédéric Leblond, Professor of Engineering Physics at Polytechnique Montréal and researcher at the University of Montreal Hospital Research Center, created a company, ODS Medical, to commercialize the probe.

Tech Briefs: Take me through the technology of the multimodal optical spectroscopy probe.

Leblond: It’s a little bit like a pen, but smaller. The surgeon will take the tool and put it in soft contact with the tissue he or she wants to analyze. There are optical fibers within the tool itself, which are guiding laser light to the surface. The laser light essentially excites the molecules, making them vibrate. Those vibrations induce a frequency shift in the light that is detected by the optical fibers. We’re detecting this light at more than 1000 different wavelengths, colors essentially. Based on that reflected light, we’re forming the spectra, and the spectra’s information is molecular in nature. The intensity of specific peaks within the spectrum relates to the concentration of proteins, lipids, DNA, or certain amino acids, for example.

Tech Briefs: How is the probe “multimodal?"

Leblond: The signal we’re collecting is multimodal because we’re collecting a Raman spectrum [a form of spectral sample identification based on the detection of the vibration of certain molecular bonds] and we’re also collecting for fluorescence. There are multiple spectra, with thousands of data points within that. With the neurosurgical tool, we’ve developed artificial intelligence/machine learning algorithms trained to recognize, based on this multimodal data, whether a spectrum belongs to cancer or to normal cells.

Tech Briefs: What are the challenges that this probe is designed to address?

Schematic depiction of the probe being used to interrogate brain tissue during surgery. The image shows a photograph of the probe held by a surgeon, Raman spectral associated with normal brain and cancer, as well as a magnetic resonance image (MRI) of a brain cancer patient with the red area representing the tumor.

Leblond: The first approach for all solid tumors is essentially surgery. Before you do chemo, radiotherapy, or immunotherapy, you typically need to go in and remove the tumor. The reason why a tumor will often come back or will not respond well to chemo or radiotherapy is that there is too much cancer left behind. The probe allows us to detect these tiny nests of cancer cells that cannot be picked up by CT or MRI, or the surgeon as he or she is operating. Our first finding, in 2015, demonstrated that this tool is more powerful than MRI in terms of its sensitivity to detect tissue where there is only small densities of cancer cells.

Tech Briefs: In your press release, I saw the probe referred to as “infallible.” How do you ensure that the probe is never wrong?

Leblond: To ensure that the models are working properly, we collected data – histopathology and optical measurements – for patients. We were taking a small chunk of tissue right underneath where we were doing each optical measurement. This tissue was then sent to a pathologist, and the pathologist would say: “This is cancer; this is normal; this is a benign tumor; this cell contains a certain density of cancer cells.” We had all that information that we could use to train the algorithm, from which we were able to make highly accurate statistical models for tissue classification. We found that, close to 100 percent of the time, it was correct in predicting cancer when interrogating cancer tissue. We’ve developed these algorithms that can determine, live during the cancer surgery, the nature of tissue type.

Tech Briefs: What’s next regarding the development?

Leblond: The first track is commercialization. We’ve raised money, and we’re going through FDA approval. At the same time, we started a clinical trial to assess the impact on patient outcome for neurosurgery.

We’re also demonstrating the use of the technology for other types of cancer, and adapting the technology so that it fits within the surgical workflow for other technologies. For prostate cancer, we’ve used the technology on more than 40 patients. We’ve shown that the probe can detect cancer in the brain as well. We’re working with a robotics company to integrate the tool into its surgical robots, which are more and more frequently being used for prostate cancer surgery.

Tech Briefs: What is most exciting to you about this probe?

Leblond: Our results can be generalized to other types of cancer cells. We’re developing a new suite of instruments that are specifically adapted to ovarian cancer surgery, to prostate cancer surgery, and to other types of surgeries. What made this whole field explode for us was that realization that we can use this for other cancers.

RELATED CONTENT:

  • An early-detection tool analyzes blood samples for a specific pancreatic cancer marker, or protein.

The U.S. Government does not endorse any commercial product, process, or activity identified on this web site.