Table 3.

Accuracy, sensitivity, and specificity of the amyloid status readings (since a significant fraction of the low-dose images were uninterpretable [71/160 reads], they are not included in the analysis).

Value	Method A	Method B	Method C	Method D
(95% Cl)
Accuracy	78.8%	96.3%	98.1%	96.3%
	(71.6%, 84.8%)	(92.0%, 98.6%)	(94.6%, 99.6%)	(92.0%, 98.6%)
Sensitivity	60.5%	100%	98.7%	98.7%
	(48.7%, 71.6%)	(95.3%, 100%)	(92.9%, 100%)	(92.9%, 100%)
Specificity	95.2%	92.9%	97.6%	94.1%
	(88.3%, 98.7%)	(85.1%, 97.3%)	(91.7%, 99.7%)	(86.7%, 98.0%)

CI: confidence interval

Network training methods: A-direct application of pre-trained network; B-transfer learning starting with pre-trained network; C-training new network from scratch; D-training new network with combined datasets