Colon polyps diagnosed more accurately by adding OCT to colonoscopy
Quing Zhu leads team that developed custom probe to better analyze polyps
Colonoscopy is the standard of care for colorectal cancer screening, yet it still falls short in allowing clinicians to access the depth that polyps invade the colon noninvasively. Now, a team of researchers at Washington University in St. Louis has added another imaging technique that could help clinicians determine whether a colon polyp is benign or cancerous with more accuracy during routine colonoscopy.
Quing Zhu, the Edwin H. Murty Professor of Engineering in the McKelvey School of Engineering at WashU, and Vladimir M. Kushnir, MD, professor of medicine in the Division of Gastroenterology at WashU Medicine, led a team of physicians and doctoral students that implemented high-resolution optical coherence tomography (OCT) into routine colonoscopy in a pilot study at WashU Medicine. Not only was the device safe, but it added only a few minutes to the colonoscopy procedure. In addition, their deep learning model predicted whether a polyp was benign or malignant with excellent accuracy. Results of the pilot study were published online in Scientific Reports Nov. 14.
“Our pilot data show that endoscopic OCT should be further investigated as a diagnostic tool for use during colonoscopy,” Zhu said. “When combined with deep learning, OCT can offer clinicians increased confidence to identify subsurface invasive depth of complex polyps and malignant lesions, potentially improving clinical decision-making. Compared with previous studies, ours is the first in vivo OCT imaging coupled with a deep-learning diagnostic model for the evaluation of benign and malignant lesions.”
While most colon polyps are not cause for concern, over time, some can develop into colon cancer, the fourth most common cancer and the fourth leading cause of death related to cancer in the United States. Routine colonoscopy is unable to determine the depth of complex polyps in the colon tissue.
To remedy this, Zhu and her team developed a portable OCT system and custom catheter probe that allowed clinicians to discern subtle differences in tissue below the surface. Biomedical engineering doctoral student Haolin Nie; Hongbo Luo, who earned a doctorate in electrical & systems engineering at WashU in 2023; and former WashU Medicine gastroenterology resident Vladimir Lamm, MD, are the main contributors on the device and machine learning algorithm development and implementation in the research.
The team’s pilot study involved 32 patients with 35 polyps throughout the colon. Adding the endoscopic OCT imaging increased the time of the colonoscopy by an average of 3 minutes and 40 seconds, though nearly 83% of patients were finished within 5 minutes. Two of the lesions evaluated with the custom OCT probe were diagnosed by biopsy as invasive cancer; the remainder were determined to be benign types of polyps.
After acquiring the images, the team analyzed more than 11,000 images, combining 7,250 images taken from live patients in this pilot study and more than 4,000 images taken from a prior study from colorectal tissue specimens. In determining malignant from benign lesions, the deep learning model achieved an area under the receiver operating characteristic curve (AUC) of 0.984 when biopsy results were used as a gold standard. An AUC of one is 100% accurate.
While the pilot study is promising in assessing complex polyps, Zhu said, there are some limitations.
“Our patient cohort was small, and our OCT catheter device can be further improved on speed and depth in human patients,” she said. “However, this pilot study has demonstrated that OCT can capture high-resolution, depth-wise images of the mucous and submucous membranes and carries enough information for machine learning to differentiate malignant from benign lesions.”
Nie H, Luo H, Lamm V, Li S, Thakur S, Zhou C, Hollander T, Cho D, Sloan E, Liu J, Navale P, Bazarbashi AN, Reyes Genere JP, Kushnir VM, Zhu Q. In vivo evaluation of complex polyps with endoscopic optical coherence tomography and dep learning during routine colonoscopy: A feasibility study. Scientific Reports, online Nov. 14, 2024. DOI: https://doi.org/10.1038/s41598-024-78891-5
Funding for this research was partially provided by the National Institutes of Health (NCI R01CA237664 and R01EB034398).
Click on the topics below for more stories in those areas
Back to NewsFaculty in this story
