Thyroid cancer is the most common type of cancer in the endocrine system, and the number of cases keeps rising as more people are being diagnosed every year.
One of the big challenges during thyroid cancer surgery is knowing exactly how much tissue to remove. Surgeons need to carefully separate cancerous tissue from healthy areas, but this is very difficult to do during the operation because the thyroid is surrounded by very delicate structures.
Right now, doctors usually rely on methods like fine-needle aspiration (FNA) and traditional pathology to diagnose thyroid cancer and see if all the cancer has been removed.
While these methods can be accurate, they are slow, not always clear, and cannot guide the surgeon in real time. This means some patients may have healthy tissue removed unnecessarily, while others might need a second surgery if cancer is missed.
A new technology called Dynamic Optical Contrast Imaging (DOCI) may offer a better way. DOCI works by shining light on tissue and capturing the faint natural glow that cells give off—called autofluorescence.
Because cancerous and healthy cells glow differently, they create unique patterns or “optical signatures” that DOCI can detect. Each DOCI scan collects data from 23 different light channels, giving a very detailed view of the tissue’s biology without the need for dyes or contrast agents.
In a recent study published in Biophotonics Discovery, researchers from Duke University and UCLA teamed up to test how well DOCI works when combined with artificial intelligence (AI).
At Duke, Tyler Vasse developed a two-part AI system in the lab of Tuan Vo-Dinh. Meanwhile, Yazeed Alhiyari and a team led by Maie St. John at UCLA handled the clinical testing.
In the first part of the study, the researchers used a basic machine learning model to classify each sample of tissue as healthy, papillary thyroid cancer, or follicular thyroid cancer—the two most common types of thyroid cancer.
The AI system analyzed the 23 channels of DOCI data and picked out a few key features to make its decision. It was able to correctly identify all the samples in an independent test, even identifying the aggressive anaplastic subtype as cancerous, showing the method’s wide sensitivity.
The second part of the study focused on a crucial question for surgeons: exactly where is the cancer located? To solve this, the team used a deep-learning model called U-Net, which is good at finding shapes and patterns in medical images.
This model created maps showing which parts of the tissue were likely cancerous. The system worked especially well for papillary thyroid cancer and made very few mistakes in areas without cancer.
Although the study used tissue that had already been removed, the results are promising. In the future, DOCI and AI could help surgeons during surgery by giving them real-time information. This could reduce mistakes, protect healthy tissue, and help avoid unnecessary procedures.
This technology could change the way thyroid cancer is treated. By using light and AI together, doctors may soon be able to “see” cancer clearly while they operate, making surgeries safer and more effective.
This exciting new method could give both doctors and patients more confidence during treatment, and it may one day become a standard tool in the operating room.
If you care about cancer, please read studies that artificial sweeteners are linked to higher cancer risk, and how drinking milk affects risks of heart disease and cancer.
For more health information, please see recent studies about the best time to take vitamins to prevent heart disease, and results showing vitamin D supplements strongly reduces cancer death.
Copyright © 2025 Knowridge Science Report. All rights reserved.
