All projects

Research · 2022 · EEF-Net · Research at KJ Somaiya College of Engineering · SSRN preprint

Breast Tumour Classification in Mammograms

Which shadow matters?

EEF-Net, an enhanced EfficientNet for classifying breast tumours in mammograms, where dense tissue and lesions share the same grey.

mammogram · region of interestEEF-Netenhanced EfficientNettumour class97.14%accuracy · F1 98.30%

Problem

In a mammogram, dense tissue and a tumour share the same grey. The classifier has to find the shadow that matters.

Approach

  • EEF-Net: an EfficientNet backbone extended for mammogram tumour classification.

Stack

  • Python
  • TensorFlow
  • Keras

Outcome

  • 97.14% accuracy and a 98.30% F1 score, as reported by the study