Skip to main content

Table 2 Comparison of mammographic factors between pure DCIS and DCIS with invasive upgrade

From: AI analytics can be used as imaging biomarkers for predicting invasive upgrade of ductal carcinoma in situ

 

Total DCIS (n = 440)

Mammographically detected DCIS (n = 341)

Pure DCIS (n = 323)

DCIS with invasive upgrade (n = 117)

p value

Pure DCIS (n = 238)

DCIS with invasive upgrade (n = 103)

p value

Imaging features on mammography

  

< 0.001

  

0.007

 Occult

85 (26.3)

14 (21.0)

 

-

-

 

 Calcifications only

123 (38.1)

47 (40.2)

 

123 (51.7)

47 (45.6)

 

 Mass/asymmetry/distortion

58 (18.0)

15 (12.8)

 

58 (24.4)

15 (14.6)

 

 Combined

57 (17.7)

41 (35.0)

 

57 (23.9)

41 (39.8)

 

Final assessment on mammography

  

< 0.001

  

< 0.001

 BI-RADS 1–2

85 (26.3)

14 (12.0)

 

-

-

 

 BI-RADS 3

12 (3.7)

3 (2.5)

 

12 (5.0)

3 (2.9)

 

 BI-RADS 4a

65 (20.1)

12 (10.3)

 

65 (27.3)

12 (11.7)

 

 BI-RADS 4b

55 (17.0)

16 (13.7)

 

55 (23.1)

16 (15.5)

 

 BI-RADS 4c

74 (22.9)

33 (28.2)

 

74 (31.1)

33 (32.0)

 

 BI-RADS 5

32 (9.9)

39 (33.3)

 

32 (13.4)

39 (37.9)

 

Raw numerical AI-CAD score (%, median [Q1,Q3])

56.9 (5.9, 97.3)

96.6 (44.8, 99.5)

< 0.001

88.9 (33.1, 98.7)

98.1 (83.0, 99.6)

< 0.001

Dichotomized AI-CAD score

  

< 0.001

  

0.002

 AI-CAD score < 50%

155 (48.0)

30 (25.6)

 

76 (31.3)

16 (15.5)

 

 AI-CAD score ≥ 50%

168 (52.0)

87 (74.4)

 

167 (68.7)

87 (84.5)

 

Graded AI-CAD score

  

< 0.001

  

0.016

 AI-CAD score < 25%

123 (38.1)

22 (18.8)

 

47 (19.7)

9 (8.7)

 

 AI-CAD score 25–50%

32 (9.9)

8 (6.8)

 

24 (10.1)

7 (6.8)

 

 AI-CAD score 50–75%

27 (8.4)

8 (6.8)

 

26 (10.9)

8 (7.8)

 

 AI-CAD score ≥ 75%

141 (43.7)

79 (67.5)

 

141 (59.2)

79 (76.7)

 
  1. Percentages are in parentheses. DCIS Ductal carcinoma in situ, Q1 First quartile, Q3 Third quartile, BI-RADS Breast Imaging Reporting And Data System, AI-CAD Artificial intelligence-based computer-aided detection/diagnosis