Table 2 Comparison of M-1 and M-2 performance in holdout test set with other deep learning-based studies.

From: Brain metabolic imaging-based model identifies cognitive stability in prodromal Alzheimer’s disease

Study

PMCI vs. SMCI

AUC (%)

ACC (%)

SEN (%)

SPE (%)

Katako (2018)39

~

~

67.3

79.6

Popuri (2018)26

80.0

~

~

~

Yee (2020)40

81.1

74.7

74.0

75.0

Pan (2020)41

86.8

83.1

72.1

88.1

Ma (2021)42

84.2

72.8

84.5

67

Lu (2018)43

~

82.5

81.4

82.9

M-1 (ours)

88.2

84.1

73.7

92

M-2 (ours)

88.6

88.6

73.7

100

  1. Studies which used any information beyond unimodal FDG PET were excluded from comparison. Metrics in bold script are the highest.
Back to article page