Imaging Combines With ML for Improved Colon Cancer Diagnosis

Colon Cancer Image — Servier Medical Art/Flickr

A cross-departmental collaboration between researchers at Washington University in St. Louis, Missouri, could lead to better and earlier diagnoses of colon cancer. A research team, led by biomedical engineering Ph.D. student Hongbo Luo from the McKelvey School of Engineering’s Department of Biomedical Engineering, worked with physicians in the School of Medicine to combine optical coherence tomography (OCT) and machine learning to create a tool that penetrates tissue, providing physicians deeper exploration capabilities when searching for abnormalities.

Currently, doctors perform colonoscopies using an endoscopy OCT and depend on visual identification to diagnose abnormalities. However, the new imaging tool developed by Luo’s team (under the guidance of Edwin H. Murty Professor of Engineering Quing Zhu and Associate Professor of Biomedical Engineering Chao Zhou) combines a small OCT catheter with a longer wavelength of light, which allows exploration of the underlying tissue – 1-2 mm deep.

The OCT catheter created by the team provided greater information than current, white-light imaging used by physicians. Further, using the imaging data collected with the device, Shuying Li, a biomedical engineering Ph.D. student and member of the research team, was able to train a machine learning algorithm to distinguish cancerous tissue from healthy tissue. Using these two tools together, the team could detect and classify cancerous tissue with 93% accuracy.

Colorectal cancer was the third most common cancer and had the second lowest survival rate worldwide in 2020. While the overall survival rate for colorectal cancer is 64%, if diagnosed before it spreads beyond the surrounding tissue, organs, and lymph nodes, the survival rate increases to 72%, and when discovered while localized, the survival rate jumps to 91%.

The combination of the algorithm’s success and that of the OCT catheter offers great potential for early diagnosis of colorectal cancer in the future. Zhu and her team, along with their partners in the School of Medicine, began patient trials in July.

Source¹

Source²