Since internal cancerous tumors usually look exactly like the healthy tissues around them, frozen section analysis is used during and after surgeries to confirm that the entirety of a tumor has been removed. This is a slow process that often leads to patients requiring follow-up procedures.
Researchers at the University of Texas at Dallas have developed a hyperspectral surgical microscope that may be able to spot cancer cells in real-time during surgeries. It not only images across nearly the entire optical spectrum, from UV to infrared, but it also uses machine learning techniques to analyze the images it produces and identify tell-tale signs of cancer.
The current prototype is a bench-top instrument, but it can be developed into a small and portable microscope that doesn’t require contrast agents and which will not emit harmful radiation beyond UV. The device analyzes the entirety of the spectrum within which it works to assess how different light frequencies are reflected and absorbed by cells. Hyperspectral imaging was originally used in astronomy and space applications.
The researchers were already able to use the new smart microscope to spot cancer cells with about an 85% accuracy in 293 tissue samples from 102 head and neck surgery patients. Further work will be needed to better train the system, and the researchers are already beginning to do that by building a large database of samples for the system to study.
They hope that one day surgeons will be able to see cancer tissue in real time, quickly and accurately resecting what is needed while sparing healthy tissues.
Via: UT Dallas