Image Quality of Digital Direct Flat-Panel Mammography Versus an Indirect Small-Field CCD Technique Using a High-Contrast Phantom

Table 2

Explorative analysis using the Kruskal-Wallis and Mann-Whitney tests. The 𝑃 values were not corrected for multiple test scenarios. 𝑃 values <.05 are labeled in bold typeface.

(a) Number of visible silica beads. The explorative analysis of interrater variability from the experimental gold standard did not produce any significant differences (𝑃<.05) between the imaging techniques, neither in the global analysis of the three imaging methods (Kruskal-Wallis Test) nor in the paired comparisons of the two imaging methods (Mann-Whitney Test)

Test	Raters
	1	2	3	4
Kruskal-Wallis test	.85	.71	.41	.63
Mann-Whitney tests
1024 CCDa versus 512 CCD	.68	.60	.18	.54
FPMb versus 512 CCD	.59	.41	.41	.72
FPM versus 1024 CCD	.89	.76	.63	.35

aCCD, charge-coupled device.
bFPM, direct digital flat-panel detector mammography system.

(b) Size of simulated microcalcifications. The explorative analysis of radiologists’ ratings measured against the experimentally preset reference values produced no global difference between the 3 imaging techniques (Kruskal-Wallis test). Explorative analysis using the Mann-Whitney test for raters 1 and 2 produced notable differences in comparing digital mammograms with the CCD images with 1024 and 512 matrix (𝑃<.05), but not for the other two raters

Test	Raters
	1	2	3	4
Kruskal-Wallis test	.00	.00	<.31	.41
Mann-Whitney tests
1024 CCDa versus 512 CCD	.00	.05	.22	.75
FPMb versus 512 CCD	.00	.00	.17	.21
FPM versus 1024 CCD	.02	.00	.87	.30

aCCD, charge-coupled device.
bFPM, direct digital flat-panel detector mammography system.