Novel imaging and AI technology predicts lung cancer gene mutations to guide targeted treatment

Scientists from the University of Edinburgh and NHS Lothian have developed a new fluorescence imaging and artificial intelligence method to accurately predict gene changes that cause lung cancer, offering a faster, more efficient and cheaper alternative to traditional laboratory testing.

The research¹, conducted in the UK, shows that this approach can identify key EGFR gene mutations from lung tissue samples, helping clinicians select the most appropriate targeted therapies for patients more quickly and using limited biopsy material more effectively.

New method for predicting lung cancer gene changes

The new technology uses fluorescence lifetime imaging microscopy (FLIM) to predict genetic changes in lung cancer without the need for conventional gene sequencing or tissue staining. Instead of relying on multiple, time-consuming molecular tests, the method captures natural light signals emitted by untreated tissue samples.

These signals are then analyzed by artificial intelligence algorithms to detect patterns associated with specific DNA mutations that drive lung cancer, including changes in the EGFR gene. By avoiding complex molecular workflows, the approach has the potential to streamline diagnostic pathways and reduce costs.

How the FLIM and AI platform works

FLIM measures how long molecules in tissue remain in an excited state before emitting light, generating detailed maps of tissue biochemistry. In this study, researchers applied FLIM to lung tissue samples and used AI to interpret the resulting data.

The method was able to:

• Predict the presence of EGFR mutations with very high accuracy
• Distinguish between the two most common types of EGFR mutations that are critical for treatment decisions
• Analyze untreated tissue, preserving biopsy material for further tests if needed

Because the technique does not require staining or destructive processing, it can extract more diagnostic information from small biopsy samples, which are often the only material available from patients with suspected lung cancer.

Addressing pressures on lung cancer diagnostics

Lung cancer is the leading cause of cancer-related death worldwide. Many lung cancers carry specific DNA changes, such as EGFR mutations, that determine whether patients are likely to benefit from targeted therapies.

Currently, detecting these mutations typically involves gene sequencing and other molecular tests that are expensive, time-consuming and consume valuable tissue from small biopsies. As lung cancer screening programs expand and detect more suspected cancers at earlier stages, diagnostic services face increasing pressure to deliver rapid, accurate results from limited samples.

Experts involved in the study say the new FLIM and AI approach could help address these challenges by speeding up diagnosis, reducing costs and preserving biopsy material.

Building on previous lung cancer imaging research

The findings build on earlier work from the same research team, which showed that FLIM could accurately distinguish between major types of non-small cell lung cancer and non-cancerous tissue. The new study extends this capability to predicting targetable genetic mutations, moving closer to a single imaging-based platform that can provide multiple layers of diagnostic information from one biopsy.

The researchers are now working towards clinical validation of these approaches. Future work aims to:

• Extend the platform to additional targetable mutations
• Apply the method to other cancer types
• Integrate FLIM and AI analysis into routine clinical workflows

Professor Ahsan Akram, co-lead of the study from the Institute for Regeneration and Repair at the University of Edinburgh, said, “This is a significant step towards a future where a single, non-destructive fluorescence scan of a biopsy could quickly inform clinicians whether a patient has cancer, what type of cancer they have and now, with this work, if it is likely to respond to targeted treatment, helping to ensure the right treatment reaches the right patient more quickly.”

Dr. David Dorward, Consultant Thoracic Pathologist at NHS Lothian, said: “Clinicians are increasingly seeing more patients with earlier-stage disease and dealing a growing number of biopsy samples, placing significant pressure on diagnostic services. Technologies like this, which can deliver more information from smaller tissue samples at speed, will be essential for developing clinically effective diagnostic pathways.”

References

1. Zang Z, Dorward DA, Ihuoma S, Akram AR, et al. Label-Free Prediction of EGFR Mutation Status Using Fluorescence Lifetime Imaging and Deep Learning in Lung Adenocarcinoma. (2026) Cancer Research (a journal of the American Association for Cancer Research) OF1–OF11. https://doi.org/10.1158/0008-5472.CAN-25-5589

Want the latest science news straight to your inbox? Become a SelectScience member for free today>>