Enhancing Cancer Diagnostic Tools with Deep Learning

Beckham Institute Biophotonics Imaging Laboratory researchers applied deep learning to polarization-sensitive optical coherence tomography (PS-OCT) to improve cancer diagnostic tools.

OCT systems are clinically common and are used to create high-resolution cross-sectional images of regions in the human body. The research team developed a new method of applying software to the OCT tool to provide polarization-sensitive capabilities without the cost and complexity of hardware-based PS-OCT imaging systems.

“We’re trying to replace the hardware associated with PS-OCT,” PhD candidate in electrical and computer engineering at the University of Illinois Urbana-Champaign and member of the Beckman Institute’s Biophotonics Imaging Laboratory, Yi “Edwin” Sun, said in a press release.

“However, [it] is still in the stage of development and research. By adding a deep learning model on top of an OCT system, suddenly we arrive at PS-OCT capabilities without the traditional added hardware.”

OCT is a non-invasive imaging test that uses light waves to determine the properties of a biological sample. However, by enabling the tool to use polarization sensitivity, scientists can find relevant information that OCT cannot capture on its own.

OCT can precisely differentiate tissue, while PS-OCT can detect abnormalities in microstructural features, including collagen fiber orientations that change in cancer-infected areas compared to uninfected areas.

“We proved that applying our method to other systems can generate a PS-OCT contrast, and that this model can be used on many OCT systems to help us differentiate cancer tissues and other types of tissues much better than OCT systems alone,” Sun said. “This is a huge improvement, making this system better for cancer diagnoses.”

Deep learning allowed the research team to create software that can work with OCT systems to deliver polarizations sensitivity.

“Deep learning enabled a more advanced method of picking up subtle features in images, which can be used for more accurate segmentation and classification. It also allows the imaging tool to use multiple layers to pick up spatial features in an image,” Sun said.

With the use of historical data, deep learning methods assist with accurate diagnoses and predictive analytics. The researchers tested the model by predicting what a photo of a lush summer forest might look like in December.

According to the researchers, synthetic PS-OCT may take a few years and more data to reach the clinical stages. However, clinicians could apply the model to commercial systems once it does, assisting in cancer detection and diagnosis.

The research was supported in part by grants from the National Cancer Institute and the National Institutes of Health.