Innovative Optical Imaging Technique Promises Earlier Detection of Colorectal Cancer

Colorectal cancer (CRC) remains one of the most prevalent and deadly cancers worldwide, primarily because it often goes undiagnosed until it reaches an advanced stage. While advancements in colonoscopy have improved early detection of precancerous lesions, accurately identifying which lesions are likely to develop into cancer during the procedure still presents a challenge.

Recent research published in Biophotonics Discovery introduces a promising solution through an advanced, non-invasive imaging method. This technique leverages the natural fluorescence properties of tissues—called autofluorescence—when exposed to specific wavelengths of light. Notably, it does not require any dyes or contrast agents, making it a safer and more straightforward approach.

The key aspect of this method involves analyzing the 'fluorescence lifetime'—the duration that autofluorescence persists after excitation. Variations in this parameter reflect subtle biochemical differences in tissues, which can help distinguish benign from malignant lesions when combined with machine learning algorithms.

In the study, researchers from the Champalimaud Foundation in Portugal examined fresh tissue samples from 117 patients undergoing colorectal surgery. Using a fiber-optic probe coupled with a dual-laser autofluorescence lifetime system, they illuminated tissue specimens at two wavelengths (375 nm and 445 nm). These wavelengths targeted molecules like collagen and cellular coenzymes, capturing multiple optical responses over time. This data was then matched with pathological diagnoses to train an AI-based classification model.

Applying ensemble learning techniques, specifically Adaptive Boosting (AdaBoost), the model learned to identify patterns associated with cancerous tissue. It achieved a high accuracy of 87% on the training set, with sensitivity of 83% and specificity of 90%. When tested on new samples, it maintained strong performance with 85% accuracy, 85% sensitivity, and 85% specificity.

This technology demonstrates potential for real-time, in vivo applications, providing doctors with immediate feedback during colonoscopy or surgery. It can pinpoint tumor regions based solely on autofluorescence signatures, streamlining decision-making and potentially reducing unnecessary biopsies.

Researchers also explored simplifying the system by focusing on the most informative optical channels, which could reduce device complexity and costs without sacrificing accuracy. While further work is needed to improve sensitivity for early or borderline lesions and to validate results across larger, more diverse populations, these findings mark a significant advancement toward integrating optical diagnostics and artificial intelligence into routine colorectal cancer care. Ultimately, this approach could lead to earlier detection, better targeted interventions, and improved patient outcomes.

Source: https://medicalxpress.com/news/2025-08-optical-imaging-technique-colorectal-cancer.html