Could the use of AI for detecting cancer be closer to routine clinical practice than you think? That was the basis for a new feasibility study using NCI’s in-house, AI-assisted model for detecting prostate cancer on MRI scans.

In this study, Dr. Stephanie Harmon, Dr. Baris Turkbey (both from NCI’s Center for Cancer Research), and their colleagues share their experiences deploying a real-time, AI-based prostate cancer pipeline that leverages a publicly available platform (i.e., MONAI Deploy Express).

To give radiologists easy access to the model, the researchers deployed their pipeline using existing infrastructure (Picture Archiving and Communications Systems, or “PACS”) that’s found in hospitals and clinics around the world. Radiologists can run the algorithm and visualize the outputs directly from the clinical PACS system they use on a daily basis.

As noted by Dr. Harmon, “Our open source pipeline worked seamlessly within the PACS system, allowing radiologists to view and interact with the results—all within a system they know and are comfortable with.”

Dr. Turkbey added, “Because our algorithm is open source, and readily available in PACS, it’s available to more people, in more locations, including those who lack access to commercial or third-party software. Our study not only shows AI’s potential for detecting cancer but also offers a way to put this technology within easier reach.”