How AI is Redefining Diagnostic Imaging and Acquisition

Artificial intelligence (AI) is driving a fundamental shift in radiation oncology in how clinicians acquire, interpret, and utilize medical images. While early iterations of computer-assisted diagnosis were limited to basic pattern recognition, the current generation of machine learning is transforming diagnostic imaging.

In this highlight from the 2026 ASCO Gastrointestinal Cancers Symposium, Caroline Chung, MD, MSc, FRCPC, CIP, vice president, chief data and analytics officer, and co-director of the Institute for Data Science in Oncology at MD Anderson Cancer Center, broke down the multifaceted role of AI in the imaging suite. Chung discussed how AI-driven image generation is drastically reducing scan times while offering a lifeline to pediatric or anxious patients who struggle to remain still. However, the quantitative measurements may differ between traditional physical data acquisition and the accelerated AI-based data acquisition and interpolation. Beyond speed of image acquisition, she highlighted the potential of AI to "see" what the human eye cannot, extracting sub-visual features that reflect certain biological signals, but in order to utilize these tools effectively for predictive modeling, there needs to be intentional effort toward more consistent measurements.

Transcript:

The initial introduction of AI came into medicine [through] the diagnostic imaging space. The very first ones were used mostly in mammography and early computer-assisted diagnosis type tools that eventually took on AI aspects. I would say that there are many ways that AI can be used in the diagnostic imaging space. One is in terms of image generation. There’s a lot of AI that’s being built into the actual acquisition of the imaging to speed up that imaging. Now, there’s challenges of understanding what’s been done when you’re taking less information physically and interpolating it with AI. We need to do our job as clinicians to make sure that we understand what that means. At the same time, it is allowing us to accelerate how we can take images. That can benefit, for instance, a patient who is anxious or is not able to participate and stay still. You can [now] get a clear image when, before, you [would] get a blurry image every single time. There are benefits to that technology emerging. When it comes to interpreting the images themselves, there are many ways that AI can be used to get consistent measurements, extract features that we can’t see, as well as potentially use it to model and predict what is going to happen moving forward. It is an exciting space.