This is an early access version, the complete PDF, HTML, and XML versions will be available soon.
A Deep Learning Approach for the Classification of Fibroglandular Breast Density in Histology Images of Human Breast Tissue
by
, , , and
Hanieh Heydarlou
1,2, 
Leigh J. Hodson
1,2,
Mohsen Dorraki
3,4,
Theresa E. Hickey
5,
Wayne D. Tilley
5,
Eric Smith
6,
Wendy V. Ingman
1,2
and
Ali Farajpour
1,2,*
1
Discipline of Surgical Specialties, Adelaide Medical School, University of Adelaide, The Queen Elizabeth Hospital, Woodville South, SA 5011, Australia
2
Robinson Research Institute, University of Adelaide, Adelaide, SA 5006, Australia
3
School of Computer and Mathematical Sciences, The University of Adelaide, Adelaide, SA 5005, Australia
4
Australian Institute for Machine Learning (AIML), Adelaide, SA 5000, Australia
5
Dame Roma Mitchell Cancer Research Laboratories, Adelaide Medical School, University of Adelaide, Adelaide, SA 5005, Australia
6
Medical Oncology, Basil Hetzel Institute, The Queen Elizabeth Hospital, Woodville South, SA 5011, Australia
*
Author to whom correspondence should be addressed.
Cancers 2025, 17(3), 449; https://doi.org/10.3390/cancers17030449
Submission received: 13 December 2024
/
Revised: 21 January 2025
/
Accepted: 24 January 2025
/
Published: 28 January 2025
(This article belongs to the Special Issue New Insights in Mammographic Density in Breast Cancer Risk, Screening and Detection)
Simple Summary
Mammographic breast density is an important risk factor for breast cancer. Women with dense breasts have a high abundance of fibroglandular breast tissue, which can be seen on a mammogram and is associated with a greater risk of developing breast cancer. To progress research into the biological mechanisms that link mammographic density to breast cancer risk, fibroglandular density can be used as a surrogate measure. Fibroglandular density can be evaluated using thin formalin-fixed paraffin-embedded breast tissue sections stained with hematoxylin and eosin. To date, the classification of fibroglandular breast density is not automated and relies on visual assessment by researchers. Hence, this study explored the use of deep learning models to automate the classification of fibroglandular breast density.
Abstract
Background: To progress research into the biological mechanisms that link mammographic breast density to breast cancer risk, fibroglandular breast density can be used as a surrogate measure. This study aimed to develop a computational tool to classify fibroglandular breast density in hematoxylin and eosin (H&E)-stained breast tissue sections using deep learning approaches that would assist future mammographic density research. Methods: Four different architectural configurations of transferred MobileNet-v2 convolutional neural networks (CNNs) and four different models of vision transformers were developed and trained on a database of H&E-stained normal human breast tissue sections (965 tissue blocks from 93 patients) that had been manually classified into one of five fibroglandular density classes, with class 1 being very low fibroglandular density and class 5 being very high fibroglandular density. Results: The MobileNet-Arc 1 and ViT model 1 achieved the highest overall F1 scores of 0.93 and 0.94, respectively. Both models exhibited the lowest false positive rate and highest true positive rate in class 5, while the most challenging classification was class 3, where images from classes 2 and 4 were mistakenly classified as class 3. The area under the curves (AUCs) for all classes were higher than 0.98. Conclusions: Both the ViT and MobileNet models showed promising performance in the accurate classification of H&E-stained tissue sections across all five fibroglandular density classes, providing a rapid and easy-to-use computational tool for breast density analysis.
