Skip to main content
. 2022 Jan 15;64(7):1359–1366. doi: 10.1007/s00234-022-02898-w

Table 3.

Diagnostic accuracy, sensitivity, and specificity with 95% confidence intervals for each rater, disease group, and software package

Rater Disease Software package Accuracy (%) Sensitivity (%) Specificity (%)
1 Dementia Software 1 85.0 (73.4; 92.9) 97.5 (86.8; 99.9) 60.0 (36.1; 80.9)
Software 2 90.0 (79.5; 96.2) 90.0 (76.3; 97.2) 90.0 (68.3; 98.8)
AD Software 1 76.7 (64.0; 86.6) 50.0 (27.2; 72.8) 90.0 (76.3; 97.2)
Software 2 78.3 (65.8; 87.9) 40.0 (19.1; 64.0) 97.5 (86.8; 99.9)
FTD Software 1 81.7 (69.6; 90.5) 95.0 (75.1; 99.9) 75.0 (58.8; 87.3)
Software 2 86.7 (75.4; 94.1) 100.0 (83.2–100.0) 80.0 (64.4; 91.0)
2 Dementia Software 1 83.3 (71.5; 91.7) 95.0 (83.1; 99.4) 60.0 (36.1; 80.9)
Software 2 88.3 (77.4; 95.2) 90.0 (76.3; 97.2) 85.0 (62.1; 96.8)
AD Software 1 76.7 (64.0; 86.6) 50.0 (27.2; 72.8) 90.0 (76.3; 97.2)
Software 2 75.0 (62.1; 85.3) 30.0 (11.9; 54.3) 97.5 (86.8; 99.9)
FTD Software 1 83.3 (71.5; 91.7) 95.0 (75.1; 99.9) 77.5 (61.6; 89.2)
Software 2 81.7 (69.6; 90.5) 80.0 (56.3; 94.3) 82.5 (67.2; 92.7)

For the disease dementia, measures represent how well raters determined whether a subject’s volumetric profile was “normal” or “abnormal,” based on software report interpretation without visual assessment, and knowing that the sample consisted of subjects with and without dementia. For AD and FTD, measures represent how well raters determined whether or not a subject had that specific subtype. MCI patients excluded

AD Alzheimer’s, FTD frontotemporal dementia