Reconsidering Explicit Longitudinal Mammography Alignment for Enhanced Breast Cancer Risk Prediction

One-line Summary

This study shows that image-level alignment of mammograms improves breast cancer risk prediction over representation-level alignment, optimizing both alignment quality and predictive accuracy.

Plain-language Overview

Mammograms are X-ray images used to detect breast cancer early. Researchers are using deep learning to predict a person's risk of developing breast cancer by analyzing these images over time. This study compares different ways to align mammograms taken at different times to improve risk prediction. The researchers found that aligning the images themselves, rather than the features extracted from them, leads to better predictions of breast cancer risk.

Technical Details