Mert Sabuncu: In screening ‘change over time’ is the most important signal

Mert Sabuncu, Professor at Cornell Tech, shared on X:

“In screening, e.g. for breast or lung cancer, ‘change over time’ is the most important signal, which can be assessed with longitudinal imaging. For example, in the US, women over 40 are recommended to receive an annual mammogram.

Yet, most AI systems developed for analyzing the imaging data only consider the latest scan. In our work, to be presented at this year’s MICCAI, we developed an AI system that can extract relevant information from all (recent and past) mammograms.

Our results show that historical images are critical to quantify cancer risk beyond the immediate future (e.g., within 5 years). We describe our Longitudinal Mammogram Risk (LoMaR) model in our pre-print.”

Source: Mert Sabuncu/X