Skip to main content

Table 2 Comparison of changes in the whole frame reading and frame reduction reading after artificial intelligence filtration

From: Reading of small bowel capsule endoscopy after frame reduction using an artificial intelligence algorithm

Variable

Reading method

Remark

Whole frame (n = 90)

Frame reduction (n = 90)

Mean number of total images (range)

46627.7 ± 20684.7

27711.3 ± 18145.7 (331 ~ 92,877)

40.6%a

Mean number of lesion images (range)

3008.5 ± 9964.9 (4 ~ 70,512)

1401.7 ± 4811.3 (2 ~ 32,327)

53.4%a

inflamed lesion (n = 38)

808.3 ± 1339.6 (11.3%)

414.6 ± 587.5 (12.5%)

48.7%a

hemorrhagic lesion (n = 19)

11265.6 ± 19730.6 (79.1%)

5213.1 ± 9647.4 (78.5%)

53.7%a

polypoid lesion (n = 13)

1661.0 ± 2939.7 (8.0%)

582.0 ± 1273.8 (6.0%)

65.0%a

vascular lesion (n = 15)

216.1 ± 345.6 (1.2%)

194.0 ± 321.6 (2.3%)

10.2%a

nonspecific lesion (n = 5)

234.4 ± 255.6 (0.4%)

175.6 ± 181.6 (0.7%)

25.1%a

Mean reading times, mins (range)

120.9 ± 45.1 (25 ~ 236)

77.9 ± 49.4 (9 ~ 258)

P < 0.001

  1. aRemoval rate