Breast cancer is the most common cancer among people worldwide. According to World Health Organisation data, 2.3 million women were diagnosed with breast cancer in 2020. In India, every four minutes, a woman receives a diagnosis of breast cancer, as reported by Hindustan Times in February 2023.
Although it is not entirely preventative, medical organizations advise regular screening for early detection. A new deep-learning model has been developed to estimate breast density which helps predict cancer risk. Breast density, defined as the proportion of fibro-glandular tissue within the breast, is often used to assess cancer risk.
According to the research team led by Professor Susan M Astley from the University of Manchester, UK, the automatic feature extraction from the training data enabled by a deep-learning-based approach makes it appealing for breast density estimations. The findings were published in Journal of Medical Imaging.
Generally, training deep learning models for medical image analysis is challenging due to limited datasets. However, in this study, the researchers used two independent deep-learning models that were initially trained on ImageNet, a non-medical imaging dataset with over a million images. This approach, known as “transfer learning,” allowed them to train the models more efficiently with fewer medical imaging data, according to SPIE, the international society for optics and photonics.
Experts including radiologists, advanced practitioner radiographers, and breast physicians assigned density values to 160,000 full-field digital mammogram images from 39.357 women on a visual analogue scale. A procedure for estimating the density score for each mammogram image was developed by the researchers to take in a mammogram image as input and present a density score as output.
The procedure involved preprocessing the images to make the training process computationally less intensive, extracting features from the processed images with the deep learning models, mapping the features to a set of density scores, and combining the scores using an ensemble approach to produce a final density estimate, according to SPIE.
The team highly accurate models for estimating breast density and its correlation with cancer risk, while conserving computation time and memory.
The model's performance is comparable to those of human experts within the bounds of uncertainty," said lead researcher Astley. “Moreover, it can be trained much faster and on small datasets or subsets of the large dataset.”
This can also be used for training other medical imaging models based on breast tissue density estimations. This can enable improved performance in tasks such as cancer risk prediction or image segmentation, according to the SPIE.
Also read: How to get more out of your daily walk