Artificial Intelligence System Detects Regularly Missed Cancer Tumors with 95% Efficiency 

The new AI system has a significantly higher success rate than trained specialists.

Engineers and scientists from the Computer Vision Research Center at the University of Central Florida have developed an Artificial Intelligence (AI) System that can detect often missed tiny specs of tumors in lung cancer patients. 

The AI system claims to have a success rate of 95% in identifying cancer tumors, compared to 65% when done by human eyes.  

The new AI system can detect potentially cancerous nodules with 95% accuracy

The news has the potential to boost the survival rates of lung cancer patients by a significant degree.  

“We used the brain as a model to create our system,” said Rodney LaLonde, one of the researchers. “You know how connections between neurons in the brain strengthen during development and learn? We used that blueprint, if you will, to help our system understand how to look for patterns in the CT scans and teach itself how to find these tiny tumors.” 

The successful diagnosis and treatment of lung cancer is highly dependent on the early detection of lung nodules. And that’s a problem, as even trained specialists miss around 35% of markers. 

How does it work?

Surprisingly, the AI platform has learned to identify the small tumors in a similar fashion to the way algorithms are used in facial-recognition software. The AI was given the key characteristics of what affected lung nodules looked like, and was then given a 1,000 CT scans to study. 

As the AI ‘learned’ to differentiate between the hundreds of images it was given to study, it was also taught to ignore other internal parts of the human body that could have rendered false-positives.  

Next steps

The researchers behind the project are now attempting to fine tune the AI to tell the difference between malignant and benign growths. 

The next phase of the project will be to move the AI system into a hospital for real-world testing. According to the team, the technology may only be two-three years away from being available throughout health care systems across the world.  

Keep up to date with all the latest tech news on FileHippo.