Purpose: Breast cancer is the commonest cancer in women. Mammographic density is a risk factor and is classified using BI-RADS (A almost entirely fat, B scattered areas of fibroglandular density, C heterogeneously dense, or D extremely dense) where approximately 20% of women are A and D and 80% are C and D density [1]. Conventional mammography density can be estimated either qualitatively or quantitatively. X-ray phase contrast computed tomography (PCT) provides high-contrast 3D soft-tissue visualisation, enabling precise quantitative measurement. Method: Fresh mastectomy specimens were scanned using PCT at the Australian Synchrotron. A 3D nnUNet deep learning framework[2] was selected for PCT image segmentation. Ten cases were used for training, with ≥10 slices per case manually annotated to delineate glandular and adipose tissues. The trained model was applied to 17 additional PCT scans from different patients. An objective breast density score (BDS) was derived from segmented glandular and adipose volumes. Four radiologists independently assigned BI-RADS density to corresponding mammograms. Inter-radiologist agreement was assessed using Krippendorff’s alpha (Kα) and concordance between PCT-derived BDS and BI-RADS density using pooled Kendall’s tau (Kτ). Results: Concordance of BDS and BI-RADS density was moderate (Kτ 0.54, 95%CI 0.27-0.73, p < 0.001). Inter-observer agreement was strong (Kα 0.72, 95%CI 0.44-0.88). Consensus BI-RADS categories were B (n=5), C (n=6), and D (n=6); BDS ranged from 0.07-0.50. Conclusion: PCT enabled objective, quantitative measurement of breast density that showed moderate concordance with radiologist-assigned BI-RADS density. By providing 3D structural assessment of breast tissue, PCT may offer a pathway towards a reproducible reference standard for breast density measurement. Ref 1. Sprague, B.L., et al. 2019. JNCI 111, 629-632 2.Isensee, F., et al. 2021. Nature methods 18, 203-211