Posted in | News | Imaging

Deep Learning and OCT: Advancing Colorectal Cancer Detection

In a study published in Scientific Reports, researchers examined the use of endoscopic optical coherence tomography (OCT) combined with deep learning for real-time characterization of colorectal polyps during routine colonoscopy.

Doctor holding human Colon anatomy model

Image Credit: Jo Panuwat D/Shutterstock.com

This approach aims to enhance diagnostic accuracy, particularly for polyps with potential for invasive cancer. By improving the real-time evaluation of colorectal lesions, which are precursors to colorectal cancer (CRC), the technology could contribute to better clinical decision-making and early detection.

Advancements in Imaging Technology

CRC is a leading cause of cancer-related deaths, making early detection crucial for effective treatment. Since most CRC cases develop from colorectal polyps, identifying and evaluating these precursors is important.

OCT is a non-invasive imaging technique that provides high-resolution, cross-sectional views of biological tissues. Using light scattering principles, it enables real-time visualization of subsurface tissue structures without requiring contrast agents. OCT has shown potential in diagnosing gastrointestinal conditions, including Barrett’s esophagus and inflammatory bowel diseases.

Traditional imaging methods, like narrow-band imaging (NBI) and white-light colonoscopy (WLC), rely heavily on surface morphology and often fail to determine the invasion depth of colorectal polyps. OCT addresses this limitation by providing detailed visualizations of tissue architecture, significantly enhancing diagnostic accuracy.

The integration of artificial intelligence (AI), particularly deep learning, further boosts OCT's potential. Neural networks trained on OCT images can automate lesion classification, improve diagnostic precision, and support clinical decision-making. Previous studies have demonstrated that machine learning can distinguish between benign and malignant lesions, paving the way for broader adoption.

Endoscopic OCT: A Novel Approach

This study assessed the feasibility and diagnostic accuracy of a custom-designed endoscopic OCT catheter for evaluating colorectal polyps in vivo during routine colonoscopy. The pilot clinical study, conducted at Washington University School of Medicine, involved 36 patients (mean age 64 years, range 46 between 84) referred for endoscopic treatment of large colorectal polyps.

The imaging process was performed immediately before standard polyp treatment, using a side-viewing OCT catheter specifically designed for adult colonoscopes. This device enabled efficient imaging throughout the colon, adding an average of 3 minutes and 40 seconds (range: 1:54-8:20) to the procedure time without causing any complications.

To analyze the acquired OCT images, the researchers employed a Vision Transformer (ViT) deep learning model, which was trained to classify lesions as benign or malignant. Then, they validated the diagnostic outcomes against the gold standard of histopathology to ensure the accurate evaluation of the method's performance.

Key Findings and Insights

The pilot study demonstrated the feasibility of using OCT during routine colonoscopy, achieving the predefined imaging criterion within 5 minutes for 82.9 % of cases (29 out of 35 polyps). The histopathological analysis identified various polyp types, including tubular adenomas, tubulovillous adenomas, sessile serrated polyps, and invasive cancer. The polyps had an average maximum dimension of 27.5 mm (range: 5-65 mm).

A total of 11351 images, comprising 7250 in vivo images from this research and 4101 ex vivo images from prior studies, were analyzed. The ViT-based deep learning algorithm achieved an area under the receiver operating characteristic curve (AUROC) of 0.984, with an accuracy of 95.0 %. It reported a sensitivity of 95 %, a specificity of 94 %, and a Cohen’s Kappa score of 0.845, indicating strong agreement with histopathological diagnoses.

Furthermore, OCT imaging provided detailed subsurface visualization that closely correlated with histological features. For example, tubular adenomas exhibited distinct "teeth-like" patterns, tubulovillous adenomas displayed "finger-like" projections, and invasive cancers appeared as homogenous, poorly differentiated structures. These detailed patterns supported nuanced differentiation among polyp types, enhancing diagnostic precision and informing clinical decision-making in managing complex colorectal polyps.

Potential Implications

Integrating endoscopic OCT with deep learning has potential for improving the management of colorectal polyps. Real-time, accurate assessments can aid in risk stratification and treatment planning, assisting clinicians in deciding between surgical intervention and endoscopic resection. These advancements could also expand OCT's use to other gastrointestinal conditions, such as Barrett’s esophagus and inflammatory bowel diseases, where detailed tissue imaging is important.

Additionally, OCT’s lower cost compared to traditional imaging methods makes it more feasible for broader clinical use. As the technology advances, it could enable the development of automated diagnostic systems, reducing the workload of endoscopists and streamlining workflows.

Future Directions

This pilot study demonstrated the feasibility and safety of integrating endoscopic OCT imaging into routine colonoscopy for evaluating complex polyps. The high diagnostic accuracy achieved with a custom deep-learning model marks progress in gastroenterology.

Future studies with larger cohorts are needed to validate these findings and assess OCT’s potential as a standard diagnostic tool. Further advancements in imaging technology and machine learning could improve OCT’s diagnostic capabilities, contributing to better patient care and outcomes in CRC prevention and management.

Journal Reference

Nie, H., et al. (2024). In vivo evaluation of complex polyps with endoscopic optical coherence tomography and deep learning during routine colonoscopy: a feasibility study. Sci Rep. DOI: 10.1038/s41598-024-78891-5, https://www.nature.com/articles/s41598-024-78891-5

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Muhammad Osama

Written by

Muhammad Osama

Muhammad Osama is a full-time data analytics consultant and freelance technical writer based in Delhi, India. He specializes in transforming complex technical concepts into accessible content. He has a Bachelor of Technology in Mechanical Engineering with specialization in AI & Robotics from Galgotias University, India, and he has extensive experience in technical content writing, data science and analytics, and artificial intelligence.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Osama, Muhammad. (2024, November 20). Deep Learning and OCT: Advancing Colorectal Cancer Detection. AZoOptics. Retrieved on November 20, 2024 from https://www.azooptics.com/News.aspx?newsID=30058.

  • MLA

    Osama, Muhammad. "Deep Learning and OCT: Advancing Colorectal Cancer Detection". AZoOptics. 20 November 2024. <https://www.azooptics.com/News.aspx?newsID=30058>.

  • Chicago

    Osama, Muhammad. "Deep Learning and OCT: Advancing Colorectal Cancer Detection". AZoOptics. https://www.azooptics.com/News.aspx?newsID=30058. (accessed November 20, 2024).

  • Harvard

    Osama, Muhammad. 2024. Deep Learning and OCT: Advancing Colorectal Cancer Detection. AZoOptics, viewed 20 November 2024, https://www.azooptics.com/News.aspx?newsID=30058.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.