Explainability of a CNN for breast density assessment

Scapicchio, C. and Lizzi, F. and Fantacci, M. E. (2021) Explainability of a CNN for breast density assessment. Il nuovo cimento C, 44 (4-5). pp. 1-4. ISSN 1826-9885

Text ncc12317.pdf - Published Version Download (147kB)

Abstract

Deep neural network explainability is a critical issue in Artificial Intelligence (AI). This work aims to develop a method to explain a deep residual Convolutional Neural Network able to automatically classify mammograms into breast density classes. Breast density, a risk factor for breast cancer, is defined as the amount of fibroglandular tissue compared to fat tissue visible on a mammogram. We studied the explainability of the classifier to understand the reasons behind its predictions, in fact with a deep multi-layer structure, it acts like a blackbox. As there is no well-established method, we explored different possible analyses and visualization techniques. The main obtained results were the achievement of a performance improvement in terms of accuracy and a contribution to assess trust in the model. This is fundamental for a potential application in clinical practice.

Item Type:	Article
Subjects:	500 Scienze naturali e Matematica > 530 Fisica
Depositing User:	Marina Spanti
Date Deposited:	20 Sep 2021 09:55
Last Modified:	20 Sep 2021 09:55
URI:	http://eprints.bice.rm.cnr.it/id/eprint/21370

Actions (login required)

View Item