Back to the list
Congress: ECR25
Poster Number: C-20996
Type: Poster: EPOS Radiologist (scientific)
Authorblock: R. Meertens1, M. Gundry1, M. Evans1, W. Vigers1, B. T. Lopez2, B. Crone2; 1Exeter/UK, 2Sedgefield/UK
Disclosures:
Robert Meertens: Nothing to disclose
Michael Gundry: Nothing to disclose
Michelle Evans: Nothing to disclose
Wavell Vigers: Nothing to disclose
Ben Thomas Lopez: Employee: Ibex Innovations Ltd
Ben Crone: Employee: Ibex Innovations Ltd
Keywords: Bones, Absorptiometry / Bone densitometry, Digital radiography, Computer Applications-Detection, diagnosis, Experimental investigations, Osteoporosis
Results

Mean (SD) age is 23.5 (3.31) and 65% of the cohort are female. Mean (SD) aBMD of the UD and TD region are 0.374 (0.0609) and 0.699 (0.707) respectively. Correlations between DXA aBMD and IBEX BH aBMD at the UD and TD ROIs are 0.901 (95% CI [0.889,0.924]) and 0.852 (95% CI [0.823,0.877]) respectively.

Mean error between DXA aBMD and IBEX BH aBMD is 0.00954 (95% CI [0.00703, 0.0124]) and 0.00259 (95% CI [0.000199, 0.00497]) for 20-30 and over 50s respectively. Standard deviation of the error is 0.0361 (95% CI [0.0344, 0.0380]) and 0.0375 (95% CI [0.0358, 0.0395]) for 20-30 and over 50s respectively.

There are no statistically significant differences in the standard deviation (p = 0.0983) of the error between DXA aBMD and IBEX BH aBMD for the 20-30 and over 50 population. There is evidence of a bias in the mean error (p<0.01).

The population mean DXA aBMD is higher for 20–30-year-olds compared to over 50s. The standard deviation DXA aBMD is significantly smaller for 20–30-year-olds. Therefore, as expected, younger people have on average more dense bones and the range of densities is smaller. The correlation between Ibex BH aBMD and DXA aBMD is strong for all regions. It is statistically significantly smaller for 20–30-year-olds in the TD region than over 50s. However, there is no statistically significant difference in RMSE between 20–30-year-olds and over 50s for any regions. This indicates the differences in correlation are due to the smaller standard deviation in aBMD rather than a mismatch between DXA and IBEX BH. The mean difference in IBEX BH and DXA aBMD is statistically significantly larger for 20-30-year-olds compared to over 50s indicating a bias for younger individuals. However, the bias is less than one sixth of a standard deviation and therefore is unlikely to be clinically significant. 

Figure 6 demonstrates that both hip and AP Spine T-score distributions have a similar mean and standard deviation. As T-score is calculated from a NHANES database of American 20–30-year-olds, if our sample is from a similar distribution of bone density as NHANES, the mean for our population should be zero and the standard deviation should be 1. The mean is close to 0 and the standard deviation is close to 1 indicating that the central DXA measurements are close to that of the NHANES database.  The forearm measurements use manufacturer values, which are a combination of proprietary data and NHANES, to calculate the T-score. Even though our central DXA measurements match the NHANES database well, the forearm measurements differ significantly showing a significant bias towards low density, more variable UD region and less variable TD region. This indicates the population ranges on the manufacturer equipment are high for the population of 20–30-year-olds we sampled. 

FRAX risk factors were present in a small number of individuals in our cohort. 1 had a parental hip fracture, 12 were smokers, 1 was using glucocorticoids, 1 had rheumatoid arthritis, 2 had secondary osteoporosis and 3 had high alcohol use.  
GALLERY