Observational Study Shows Value in Using AI to Plan Radiotherapy for Lung Cancer
How can coordination between humans and artificial intelligence (AI) improve outcomes for lung cancer patients?
A recent study used CT (Computed Tomography) imaging data from NCI’s National Clinical Trials Network (NCTN) to investigate the usefulness of AI in radiation therapy planning for lung cancer treatment.
Doctors use radiation therapy to target cancer in nearly half of lung cancer cases. However, planning for this therapy is a time-consuming, manual process. Researchers investigated how the use of deep learning algorithms for targeting non-small cell lung cancer (NSCLC) could save resources and lead to better outcomes for patients.
Researchers trained a deep learning algorithm on NCTN CT imaging data sets. The algorithm identified and outlined, or segmented, a NSCLC tumor within seconds.
To test how well the algorithm worked, researchers asked radiation oncologists to perform, rate, and edit lung cancer segmentations.
Some of the promising results from the study include the following:
- Radiation oncologists using the algorithm performed as well as those who did not.
- Radiation oncologists using the algorithm performed 65% quicker than those who did not.
- Researchers saw 32% less variation when editing an AI-produced segmentation compared to a manually produced one.
- Physicians rated the quality of AI-drawn segmentations more highly than the human-expert drawn ones.
The researchers are continuing to study how physicians interact with AI and how that coordination can potentially improve outcomes in a clinical setting. Future study also includes developing additional algorithms to verify results.
For more information on the study and its authors, visit The Lancet.