# Screening Mammogram Classification with Prior Exams

## TL;DR

This paper introduces neural network models that compare current and prior mammograms to improve breast cancer screening accuracy, trained on a large dataset, achieving high AUC scores.

## Contribution

The study presents novel neural network architectures designed specifically for comparing pairs of mammograms, reflecting radiologists' diagnostic practice.

## Key findings

- Achieved an AUC of 0.866 in malignancy prediction.
- Reduced error rate compared to baseline models.
- Utilized over 665,000 pairs of mammogram images.

## Abstract

Radiologists typically compare a patient's most recent breast cancer screening exam to their previous ones in making informed diagnoses. To reflect this practice, we propose new neural network models that compare pairs of screening mammograms from the same patient. We train and evaluate our proposed models on over 665,000 pairs of images (over 166,000 pairs of exams). Our best model achieves an AUC of 0.866 in predicting malignancy in patients who underwent breast cancer screening, reducing the error rate of the corresponding baseline.