Skip to main content

Table 3 Subgroup analyses by lesion size and location

From: A single-center prospective study evaluating the usefulness of artificial intelligence for the diagnosis of esophageal squamous cell carcinoma in a real-time setting

 

Accuracy

Sensitivity

Specificity

 

AI

Endoscopists

AI

Endoscopists

AI

Endoscopists

Size

      

< 10 mm (n = 192)

82.8% [76.7–87.9]

86.5% [80.8–91.0]

61.5% [40.6–79.8]

46.2% [26.6–66.6]

86.1% [79.9–91.0]

91.7% [86.4–95.4]

≥ 10 mm (n = 45)

71.1% [55.7–83.6]

82.2% [67.9–92.0]

77.8% [52.4–93.6]

83.3% [58.6–96.4]

66.7% [46.0-83.5]

81.5% [61.9–93.7]

Location

      

Ce-Ut (n = 39)

82.1% [66.5–92.5]

84.6% [69.5–94.1]

100% [51.8–100]

87.5% [47.3–99.7]

77.4% [58.9–90.4]

83.9% [66.3–94.5]

Mt (n = 102)

82.4% [73.6–89.2]

85.3% [76.9–91.5]

61.9% [38.4–81.9]

57.1% [34.0-78.2]

87.7% [78.5–93.9]

92.6% [84.6–97.2]

Lt-Ae (n = 96)

78.1% [68.5–85.9]

86.5% [78.0-92.6]

60.0% [32.3–83.7]

53.3% [26.6–78.7]

81.5% [71.3–89.2]

92.6% [84.6–97.2]

  1. Ae, abdominal esophagus; AI, artificial intelligence; Ce, cervical esophagus; Lt, lower thoracic esophagus; Mt, middle thoracic esophagus; Ut, upper thoracic esophagus