Breast density is defined as the proportion of fibro glandular tissue in relation to fat within a breast, and has been shown to be a risk factor for development of breast cancer. Dense breasts, a term that encompasses both the heterogeneously dense and the extremely dense breasts according to the Breast Imaging Reporting and Data System (BI-RADS) assessment categories, approximately account for 23.9%–35.0% of breast cancer in premenopausal women and 13.0%–16.7% in postmenopausal women (Chalfant and Hoyt, 2022).

Breast density is considered to be one of the predictors of risk for developing breast cancer. The risk is 4 - 5 times more in women having dense breast tissue than those with less dense breast tissue (Natesan et al., 2019). Mammography is recognized to be the gold standard for screening of breast cancer. However, mammography is challenging due to overlapping glandular tissue. This can cause false positive or false negative results and consequently, increased recall rates, and can affect both the specificity and sensitivity of the procedure (Badawy et al., 2023). Reduced accuracy in almost 50% of cases with dense breasts, is caused by fibro-glandular tissue veiling the presence of tumors on mammography. Controversy also remains as a result of radiation exposure and painful compression of the breasts (Accuracy et al., 2023).

Advances in artificial intelligence (AI) led to techniques that aid radiologists in increasing breast cancer detectability, estimating density and improving performance. AI is the ability of a computerized system to interpret accurately and learn from external data and to adapt obtained knowledge flexibly to perform certain tasks. Using AI methodology in medical imaging has acquired increasing interest in research. Progress has been made in using AI algorithms, especially deep learning (DL) algorithms, in tasks involving image recognition (Al-Karawi et al., 2024).

This study included 592 adult female patients (age range:29-83 years), coming to our department for either screening or diagnostic mammography, between October 2022 and April 2024. Pregnant and lactating ladies, as well as those with breast implants were excluded from participation. After ethical committee approval and consenting, full-field digital mammography was done in both the Craniocaudal (CC) and mediolateral oblique (MLO) views for each breast. It was then interpreted by two experienced breast imaging radiologists, blinded to the results of each other, where each radiologist gave an ACR breast density score, ranging from A to D for each mammogram. AI images were automatically software-generated (Lunit Insight, Korea) from the mammographic images. The results of AI for estimating ACR breast density were compared to the subjective results of both radiologists.