HAI Weekly Seminar with Akshay Chaudhari
Beyond Image Interpretation in Radiology: Data-Efficient AI for Accelerating MRI Acquisition
Get the latest news, advances in research, policy work, and education program updates from HAI in your inbox weekly.
Sign Up For Latest News
Beyond Image Interpretation in Radiology: Data-Efficient AI for Accelerating MRI Acquisition
The possibility that AI will automate most cognitive labor is worth taking seriously. How should we adapt to this transformation? I start from the perspective, articulated in the essay “AI as normal technology”, that the true bottlenecks lie downstream of capabilities and that AI’s impacts will unfold gradually over decades. If this is true, there are major gaps in our current evidence infrastructure, because it over-emphasizes the capability layer.
.png&w=1920&q=100)
The possibility that AI will automate most cognitive labor is worth taking seriously. How should we adapt to this transformation? I start from the perspective, articulated in the essay “AI as normal technology”, that the true bottlenecks lie downstream of capabilities and that AI’s impacts will unfold gradually over decades. If this is true, there are major gaps in our current evidence infrastructure, because it over-emphasizes the capability layer.
The AI Index, currently in its ninth year, tracks, collates, distills, and visualizes data relating to artificial intelligence.

The AI Index, currently in its ninth year, tracks, collates, distills, and visualizes data relating to artificial intelligence.
Strategic stability exists when neither side thinks it can improve its strategic outcome by striking first.

Strategic stability exists when neither side thinks it can improve its strategic outcome by striking first.
Recent applications of artificial intelligence (AI) in radiology have focused on image interpretation tasks such as image classification, segmentation, or detection. However, a fundamental challenge in radiology is to acquire these medical images in a safe and efficient manner. New AI techniques have been proposed to solve the inverse problem of image reconstruction wherein only a limited set of measurements are used to reconstruct medical images with high diagnostic quality. Specifically, in this talk, I will describe how physics-guided AI is currently being used to improve the speed of magnetic resonance imaging (MRI). I will further describe how we may eschew requiring large extents of paired datasets required for supervised model training by using novel unsupervised and semi-supervised approaches for accelerated MRI. Beyond data efficiency, these approaches can help mitigate the challenge of distribution shifts for trained models. I will conclude by describing a 1.5TB dataset that we have made publicly available to help evaluate MRI reconstructions with clinically-relevant metrics.
