Skip to main content

Table 2 Comparison between mammographic density assessment using different visual scales (1 rater), DM-Scan (3 raters) and Cumulus (1 rater)

From: Validation of DM-Scan, a computer-assisted tool to assess mammographic density in full-field digital mammograms

   DM-Scan Rater 1 DM-Scan Rater 2 DM-Scan Rater 3 Cumulus (Rater 1)
Classification N Mean P05-P95 Mean P05-P95 Mean P05-P95 Mean P05-P95
Qualitative scales         
Wolfea          
N1 43 1.8% 0.0%-5.8% 3.6% 0.0%-12.5% 1.6% 0.0%-6.5% 2.3% 0.0%-6.5%
P1 327 8.2% 0.0%-19.2% 10.4% 1.5%-24.8% 7.9% 0.0%-17.1% 5.4% 0.0%-17.1%
P2 166 25.6% 12.3%-1.3% 26.7% 11.7%-45.8% 25.7% 5.7%-38.7% 19.3% 5.7%-38.7%
DY 101 33.5% 15.2%-55.7% 34.1% 13.8%-60.1% 32.7% 9.3%-61.4% 29.2% 9.3%-61.4%
Tabarb          
II 43 1.8% 0.0%-5.8% 3.6% 0.0%-12.5% 1.6% 0.0%-6.5% 2.3% 0.0%-6.5%
III 381 10.1% 0.0%-26.0% 12.2% 1.7%-12.5% 9.8% 1.1%-23.2% 6.8% 0.1%-22.6%
IV 155 30.8% 13.4%-53.2% 31.5% 12.3%-54.7% 30.6% 13.0%-50.5% 24.9% 5.9%-50.9%
V 58 29.4% 12.8%-51.6% 29.8% 10.4%-52.5% 28.7% 12.3%-53.8% 25.6% 9.4%-51.4%
BIRADS densityc          
1 340 6.8% 0.0%-17.5% 8.8% 0.4%-21.5% 6.5% 0.0%-15.8% 4.6% 0.0%-13.9%
2 183 21.2% 10.8%-36.6% 21.9% 10.1%-38.3% 20.8% 10.2%-34.6% 15.4% 4.2%-30.6%
3 89 34.1% 24.0%-48.3% 35.4% 22.3%-50.4% 33.9% 21.8%-49.2% 28.2% 15.4%-46.6%
4 25 47.6% 30.9%-68.0% 51.3% 32.0%-69.4% 47.6% 32.5%-61.8% 45.4% 16.8%-68.3%
Semiquantitative scale         
Boyd categories          
0% 46 1.0% 0.0%-5.3% 2.1% 0.0%-8.7% 1.1% 0.0%-3.8% 1.6% 0.0%-5.8%
<10% 186 5.7% 0.0%-14.3% 7.1% 1.5%-14.5% 5.3% 0.8%-12.3% 3.4% 0.0%-8.2%
10%-25% 195 14.1% 3.8%-27.1% 16.0% 7.0%-28.8% 13.6% 5.1%-23.5% 9.2% 2.4%-20.4%
25-50% 139 26.2% 13.6%-40.2% 27.6% 14.9%-44.1% 26.2% 12.5%-41.5% 21.4% 7.3%-37.4%
50-75% 59 37.5% 28.2%-48.8% 39.5% 27.5%-51.1% 38.0% 26.1%-53.4% 31.4% 0.5%-47.9%
>75% 13 56.0% 41.3%-71.7% 57.0% 46.8%-69.4% 51.6% 40.3%-61.8% 55.0% 44.2%-77.2%
Agreement Boyd – DM-Scan       Agreement Boyd –Cumulus
Weighted kappa   0.801   0.789   0.805   0.697
(95% Confidence Interval)   (0.777-0.823)   (0.764-.812)   (0.783-.825)   (0.652-.738)
Agreement Cumulus – DM-Scan        
Mean Difference (DM-Scan – Cumulus) Mean Difference CCC e Mean Difference CCC e Mean Difference CCC   
(P05-P95d) +3.7% 0.841 +5.3% 0.803 +3.5% 0.842
Concordance Correlation Coefficient (CCC) (+0.0% to +42.3%) (0.820-0.863) (+1.3% to+45.6%) (0.777-0.828) (+0.8% to +43.3%) (0.820-0.864)
(95% Confidence Interval)
  1. a Wolfe classification:
  2. N1: Breast composed almost completely of fat, with perhaps just a few fibrous connective tissue strands.
  3. P1: Breast composed mainly of fat, although up to a quarter of the sub-areolar area may show beaded or cord-like areas of ducts.
  4. P2: More severe involvement of the breast, with a prominent duct pattern occupying more than one quarter of breast volume.
  5. DY: Breast typically contains extensive regions of homogeneous mammographic densities. The proportion of density is greater than that of the fat.
  6. b Tabár classification:
  7. I Mammogram composed of scalloped contours with some lucent areas of fatty replacement and 1 mm evenly distributed nodular densities (none of our mammograms were classified in this category).
  8. II Mammogram composed almost entirely of lucent areas of fatty replacement and 1-mm evenly distributed nodular densities.
  9. III: Prominent ducts in the retroareolar area.
  10. IV: Extensive, nodular and linear densities with nodular size larger than normal lobules.
  11. V: Homogeneous ground-glass-like appearance with no perceptible features.
  12. c Breast Imaging Reporting and Data System (BIRADS) classification:
  13. 1 Predominantly fatti breast.
  14. 2 Scattered fibroglandular densities.
  15. 3 Heterogeneously dense breast.
  16. 4 Extremely dense breast.
  17. d 5th & 95th percentiles of the differences between PD estimated with DM-Scan and Cumulus.
  18. e Concordance correlation coefficient and its 95% Con.