Scientists Train Computers to Detect Mesothelioma on Scans

Mesothelioma imaging scans help doctors diagnose the cancer and review the effectiveness of treatment.

However, it’s a time-consuming process susceptible to human error.

The University of Glasgow hopes to change that.

Researchers at the Scottish institution are streamlining the review of mesothelioma imaging scans. They created an artificial intelligence to find and measure tumors on scans. The computer programs were “extremely accurate,” according to the University of Glasgow press release.

This is not the first time doctors created an AI system for diagnosing mesothelioma. Google researchers teamed with medical institutions to train computers to detect lung cancer. The program was 94% accurate at detecting a disease and knowing the type.

The Scottish system goes a step further, training computers on mesothelioma scans. If this system expands beyond the university and to other countries, it may quicken the diagnostic and treatment processes.

 

Creating the AI System for Mesothelioma Scans

The study team, which includes data scientists and artificial intelligence experts, trained the program. They showed 100 computer tomography (CT) scans, which are the most common for mesothelioma. The team drew an outline around the tumor radius to train the AI program.

When the AI saw a new set of CT scans, it found and measured the tumors without any human assistance.

The initial focus is mesothelioma due to the rates in Scotland, which reportedly has the highest incidence of mesothelioma in the world. Mesothelioma also is a challenging cancer to locate and measure on imaging scans.

Mesothelioma grows and spreads like a sheet, or an avalanche, with many individual tumors rather than one mass. This creates complex and unorthodox tumor shapes. Many doctors either misdiagnose mesothelioma or don’t locate all tumors.

 

Helping With Treatment for Mesothelioma

CT scans show how tumors respond to mesothelioma treatment: If they’ve shrunk, the treatment is working. The AI program can outline the tumors and quickly show doctors a comparison between pre- and post-treatment imaging scans.

This helps doctors decide whether to continue a treatment, like chemotherapy, or switch to another option such as immunotherapy.

“To our knowledge, this study is world-leading in its successful use of AI to assess treatment response in mesothelioma,” said Kevin Blyth, professor of respiratory medicine at the University of Glasgow. “Using external data sets to validate our results, we have shown that tumors can be accurately measured by AI, giving us a new tool that will help us make better decisions for patients on treatment and reducing barriers to development of new treatments in clinical trials.”

Clinical studies may become less expensive and time-consuming, speeding up the approval process for a potential cure. Most important is accuracy, and the AI program seems to exceed expectations.

“The speed and accuracy of the AI algorithm could have a wide-reaching impact on Mesothelioma treatment,” said Keith Goatman, principal scientist at Canon Medical, which partnered with the university. “Accurate tumor volume measurements are much too time-consuming to perform by hand. Automating these measurements will open the way for clinical trials of new treatments, by detecting even small changes in the tumor size. Ultimately, it could be used routinely in hospitals to decide the best treatment for each individual.”