Researchers at Washington University in St. Louis and Siteman Cancer Center have developed a 3D optical coherence tomography (OCT) imaging system that uses machine learning to detect endometrial cancer. By capturing high-resolution images of the entire endometrial cavity in under three seconds, the technology aims to address the 10% false-negative rate associated with traditional, invasive biopsy methods, according to findings published in npj Imaging on June 3, 2026.
How does 3D OCT imaging improve cancer detection?
The new diagnostic tool functions as an “optical biopsy,” providing a detailed view of tissue microstructure without the need for physical tissue removal. According to Quing Zhu, the Edwin H. Murty Professor of Engineering at Washington University, the system detects differences in how tissue reflects light, allowing for 3D visualization up to 2 millimeters deep. In a study of 57 post-hysterectomy uteri, the team’s machine learning model successfully categorized tissue into benign or pre-cancerous groups with 94% sensitivity and 87% specificity. This performance compares favorably to standard biopsies, which face limitations due to random sampling and interpretive variability.
Endometrial cancer remains the most common gynecologic malignancy in the United States, with over 69,000 cases diagnosed in 2025 and an annual incidence increase of up to 3%.
Why is this technology a shift from current biopsy standards?
Traditional endometrial biopsies often result in false negatives because they capture only a small, random portion of the uterine lining. Dr. Lindsay Kuroki, an associate professor of obstetrics and gynecology at WashU Medicine, notes that this “blind” sampling approach misses high-risk lesions. By using a custom catheter to scan the entire cavity, the OCT system provides a comprehensive assessment. Dr. David Mutch, the Ira C. and Judith Gall Professor of obstetrics and gynecology, states that this technology addresses a critical gap in medical care, as there is currently no reliable, non-invasive screening method for endometrial cancer.
What are the next steps for clinical adoption?
While the initial research utilized ex-vivo samples, the research team is now preparing to transition the technology into live patient settings. According to Zhu, demonstrating the effectiveness of the catheter in clinical environments is the essential next step to validate the translational potential of the AI-assisted system. If successful, this could reduce the physical trauma and diagnostic delays currently experienced by patients undergoing investigation for potential malignancy.
Frequently Asked Questions
- How fast is the OCT imaging process? The system images the entire endometrial cavity in 2 to 3 seconds.
- Is the procedure invasive? It uses a specialized catheter probe, which is designed to be safer and less invasive than traditional surgical biopsy methods.
- What role does AI play? Machine learning models analyze 26 extracted imaging features to categorize tissue as either normal/benign or pre-cancer/cancer.
Have you or a family member navigated the challenges of a cancer diagnosis? Share your thoughts on the future of AI-assisted medical imaging in the comments below.
