Posted in | News | Medical Optics

Multiphoton Microscopy Technique Improves Pancreatic Diagnoses

Researchers at the University of Arizona have developed a light-based imaging technique that helps surgeons detect pancreatic cancer tissue in real time, improving surgical precision and reducing delays in diagnosis.

Scan of upper abdomen including pancreas. 
Image Credit: Radiological imaging/Shutterstock.com

The study, published in Biophotonics Discovery, describes how multiphoton microscopy can distinguish between healthy and cancerous tissue with remarkable accuracy.

Pancreatic neuroendocrine neoplasms (PNENs) represent a rare type of cancer that impacts the hormone-secreting cells within the pancreas. Despite their rarity, the frequency of these neoplasms has been gradually increasing over recent decades. Available treatment methods encompass chemotherapy and targeted therapies; however, surgery is still the sole potential cure.

Nevertheless, the determination of surgical options frequently relies on pathology findings, which can take several hours or even days. This postpones and heightens the likelihood of incomplete tumor excision.

Their multiphoton microscopy technique uses light-based imaging to illuminate naturally fluorescent molecules within tissue. In contrast to conventional microscopy, MPM inflicts less harm on samples and delivers clearer images, rendering it a valuable tool for real-time analysis during surgical procedures.

The research team employed MPM to examine pancreatic tissue samples for naturally occurring fluorescent markers such as collagen, NADH, FAD, lipofuscins, and porphyrins. These markers are instrumental in differentiating between healthy and cancerous tissues. The researchers used learning (ML) and deep learning methodologies to analyze the data. One ML analyzed the image algorithm, and four convolutional neural networks (CNNs) were developed to categorize the types of tissue.

The ML algorithm attained an accuracy of 80.6 % in detecting cancerous tissue, whereas the CNNs exhibited even greater performance, with accuracies ranging from 90.8 % to 96.4 %. These impressive results are particularly significant as the samples were sourced from various biorepositories, showing that the method is reliable across different origins.

Although the CNNs surpassed the ML algorithm in performance, the latter provided greater transparency. By examining factors that impacted the ML model's decisions, the researchers discovered that collagen content and image attributes such as contrast and correlation were significant indicators of cancer. This understanding may assist in enhancing future models and deepening the comprehension of PNEN tissue structure.

The study also demonstrated the speed of MPM imaging in comparison to conventional histology. The researchers are optimistic that additional enhancements could accelerate the process even further. They intend to evaluate the method on fresh tissue samples obtained during surgery and investigate its potential to assist in identifying the grade and type of PNENs. This data could more accurately inform treatment choices.

This study suggests a future in which cancer diagnosis and surgical planning may occur in almost real-time, which could reduce the need for repeat surgeries and enhance overall outcomes for patients suffering from pancreatic cancer.

Journal Reference:

Daigle, N., et al. (2025) Investigating machine learning algorithms to classify label-free images of pancreatic neuroendocrine neoplasms. Biophotonics Discovery. doi.org/10.1117/1.BIOS.2.4.045001

Source:

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.