Lung cancer is one of the most common and serious types of cancer, taking more than 1.8 million lives per year – that’s more than breast cancer, prostate and colorectal cancer combined. Like other cancers, it can go undetected in its early stages as there are virtually no signs and symptoms.
Unfortunately, the majority of
cancers aren’t caught until later stages, when interventions are more invasive and
tend to be less successful. There could, however, be a glimmer of hope on the
horizon and it’s AI we will have to thank for it.
A US study conducted by Google and
researchers at Northwestern University in Illinois suggests that AI is better
than specialist doctors at diagnosing lung cancer.
The new AI-based tool by Google was used in a study to see if the analysis of scans could be more accurate and more accessible than current screening protocols.
The typical screening process for
lung cancer is demanding, radiologists have to analyse hundreds of images from
a single CT scan. With this new AI model, Google says it can generate an
overall lung cancer malignancy prediction and identify subtle malignant tissue,
or lung nodules, which are often difficult to see.
The first step in the study was to
train the computer software with 42,290 CT lung scans from nearly 15,000
patients. The AI was not told what to look for, just which patients went on to being
diagnosed with cancer and which did not.
The AI-based tool was then compared
against a team of six board-certified radiologists. In this preliminary study,
Google’s AI detected 5% more cancer cases than the radiologists, while also
reducing false-positive (people falsely diagnosed with cancer) exams by over
Dr Mozziyar Etemadi, from Northwestern University, told the BBC that AI and doctors working side by side would be even more effective and that AI had a ‘huge’ role to play in the future of medicine.
The current software is not yet
ready for clinical use, but Google says the initial results are encouraging.