Automated, reference-free quantification of cortical bone parameters detects impairments in postmenopausal osteoporosis

While ultrashort echo time (UTE) measures of pore water have shown promise in assessing cortical bone porosity, most are hindered by the need for complicated processing and reference samples. The suppression ratio (SR) is a marker of porosity which is simply calculated as the voxel-wise ratio of two UTE magnitude images, one without and with long-T2 suppression. Automated cortical bone segmentation via deep learning showed elevated SR in postmenopausal women with osteoporosis (P=0.001) and was strongly associated with pore water density (R=0.93) and with pQCT BMD (R=-0.88). Results suggest that SR can detect elevated porosity in postmenopausal osteoporosis.