Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

The AI System That Can Detect Lung Cancer

DZone 's Guide to

The AI System That Can Detect Lung Cancer

Explore about AI that can improve the detection of lung cancer using deep learning.

· AI Zone ·
Free Resource

A central tranche of AI in healthcare today has been in spotting signs of problems in medical imaging faster than humans currently can.  The latest example of this comes via a new study from Google and Northwestern Medicine, which proposes to improve the detection of lung cancer using deep learning.

“Radiologists generally examine hundreds of two-dimensional images or ‘slices’ in a single CT scan, but this new machine learning system views the lungs in a huge, single three-dimensional image,” the researchers explain. “AI in 3D can be much more sensitive in its ability to detect early lung cancer than the human eye looking at 2-D images. This is technically ‘4D’ because it is not only looking at one CT scan, but two (the current and prior scan) over time.”

By partnering with Google the research team gained the computational heft to power such tasks, which aim to provide earlier detection of a disease that causes over 150,000 deaths per year in the United States alone.

Fast and Accurate

Early detection is not enough on its own of course, and the team were striving to ensure their system was as accurate as possible.  Screenings provided today are not only unavailable to large portions of the population, but also come with unacceptably high error rates.

The system utilizes the primary CT scan and whenever possible the prior CT scan from the patient to act as the inputs in the system.  The prior scans are invaluable as they allow the team to test for the growth of suspicious lung nodules.  The system was trained using de-identified biopsy scans, and is capable of identifying both specific regions of interest and the likelihood of lung cancer existing in that region.  In tests against radiologists, it performed at a comparable level.

“The system can categorize a lesion with more specificity. Not only can we better diagnose someone with cancer, we can also say if someone doesn’t have cancer, potentially saving them from an invasive, costly and risky lung biopsy,” the team explain.

The model was trained on around 2,700 CT scans, and it emerged that the system was able to spot even the tiniest malignant lung nodules with a high degree of sensitivity and specificity.

“Most of the software we use as clinicians is designed for patient care, not for research,” the team explain. “It took over a year of dedicated effort by my entire team to extract and prepare data to help with this exciting project. The ability to collaborate with world-class scientists at Google, using their unprecedented computing capabilities to create something with the potential to save tens of thousands of lives a year is truly a privilege.”

Suffice to say, the system still needs to be clinically validated, and the team plan to do so on a larger population sample, but they’re enthused by these initial results and optimistic that it can successfully improve the management and outcome for those with lung cancer in future.

Topics:
ai ,ai in healthcrae ,lung cancer ,ai systems

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}