Table 1 Accuracy, sensitivity, and specificity for pathologist and SpindleMesoNET assessment of the referral set.

From: Deep-learning based classification distinguishes sarcomatoid malignant mesotheliomas from benign spindle cell mesothelial proliferations

Method

Accuracy

Sensitivity for malignancy

Specificity for malignancy

Pathologist #1

87.5%

85.7%

89.5%

Pathologist #2

95.0%

90.5%

100%

Pathologist #3

92.5%

85.7%

100%

Average of pathologists

91.7%

87.3%

96.5%

SpindleMesoNET (using 0.95 threshold, with 95% confidence intervals)

92.5% (85.0−97.5)

85.7% (70.6−95.7)

100% (100−100)

Back to article page