Table 2 Classification performance of models trained and tested on images with different compression ratios.

From: Impact of image compression on deep learning-based mammogram classification

Model/Data

CR1

CR15

CR20

CR25

CR50

CR100

CR500

CR1K

CR5K

CR10K

CR11K

(a) AUROC with 95% confidence interval

M-CR1

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.84 (0.82, 0.87)

0.73 (0.69, 0.76)

0.71 (0.68, 0.74)

M-CR15

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.9)

0.87 (0.85, 0.89)

0.85 (0.82, 0.87)

0.73 (0.7, 0.76)

0.72 (0.69, 0.75)

M-CR20

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.84 (0.81, 0.86)

0.72 (0.69, 0.75)

0.70 (0.66, 0.73)

M-CR25

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.86 (0.84, 0.89)

0.85 (0.82, 0.87)

0.73 (0.71, 0.76)

0.71 (0.69, 0.74)

M-CR50

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.87 (0.85, 0.9)

0.85 (0.83, 0.88)

0.73 (0.7, 0.76)

0.71 (0.68, 0.74)

M-CR100

0.88 (0.85, 0.9)

0.88 (0.85, 0.9)

0.88(0.85, 0.9)

0.88 (0.85, 0.9)

0.88 (0.85, 0.9)

0.88 (0.85, 0.9)

0.87 (0.85, 0.9)

0.84 (0.81, 0.86)

0.71 (0.67, 0.74)

0.70 (0.66, 0.73)

M-CR500

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87 (0.85, 0.89)

0.87(0.85, 0.89)

0.87 (0.85, 0.9)

0.86 (0.83, 0.88)

0.73 (0.7, 0.76)

0.72 (0.69, 0.75)

M-CR1K

0.86 (0.83, 0.88)

0.86 (0.83, 0.88)

0.86 (0.83, 0.88)

0.86 (0.83, 0.88)

0.86 (0.83, 0.88)

0.86 (0.83, 0.88)

0.86 (0.84, 0.88)

0.84 (0.81, 0.87)

0.73 (0.7, 0.76)

0.72 (0.69, 0.75)

M-CR5K

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.84, 0.89)

0.87 (0.85, 0.89)

0.78 (0.76, 0.81)

0.76 (0.73, 0.79)

M-CR10K

0.82 (0.78, 0.84)

0.82 (0.78, 0.84)

0.82 (0.78, 0.84)

0.82 (0.78, 0.84)

0.82 (0.78, 0.84)

0.82 (0.78, 0.84)

0.82 (0.79, 0.84)

0.82 (0.79, 0.84)

0.83 (0.79, 0.85)

0.79 (0.76, 0.82)

M-CR11K

0.80 (0.76, 0.82)

0.80 (0.77, 0.82)

0.80 (0.77, 0.82)

0.80 (0.76, 0.82)

0.80 (0.77, 0.82)

0.80 (0.77, 0.82)

0.80 (0.77, 0.82)

0.80 (0.77, 0.82)

0.80 (0.77, 0.83)

0.79 (0.76, 0.81)

(b) AUPRC with 95% confidence interval

M-CR1

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.68 (0.63, 0.72)

0.40 (0.35, 0.46)

0.38 (0.33, 0.43)

M-CR15

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.70 (0.65, 0.74)

0.41 (0.36, 0.46)

0.39 (0.34, 0.44)

M-CR20

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.67 (0.62, 0.71)

0.41 (0.36, 0.46)

0.38 (0.33, 0.43)

M-CR25

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.75 (0.7, 0.78)

0.70 (0.65, 0.74)

0.42 (0.37, 0.47)

0.39 (0.34, 0.44)

M-CR50

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.70 (0.65, 0.74)

0.41 (0.36, 0.46)

0.38 (0.33, 0.43)

M-CR100

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.68 (0.62, 0.72)

0.39 (0.34, 0.44)

0.37 (0.32, 0.42)

M-CR500

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.76 (0.71, 0.79)

0.70 (0.65, 0.74)

0.40 (0.35, 0.46)

0.37 (0.32, 0.42)

M-CR1K

0.74 (0.68, 0.76)

0.74 (0.68, 0.76)

0.74 (0.68, 0.77)

0.74 (0.68, 0.76)

0.74 (0.68, 0.76)

0.74 (0.68, 0.77)

0.74 (0.68, 0.77)

0.69 (0.64, 0.73)

0.41 (0.36, 0.46)

0.39 (0.34, 0.44)

M-CR5K

0.74 (0.69, 0.78)

0.74 (0.69, 0.78)

0.74 (0.69, 0.78)

0.74 (0.69, 0.78)

0.74 (0.69, 0.78)

0.74 (0.69, 0.78)

0.74 (0.7, 0.78)

0.75 (0.7, 0.78)

0.49 (0.43, 0.55)

0.44 (0.38, 0.49)

M-CR10K

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.66 (0.59, 0.69)

0.65 (0.58, 0.69)

0.48 (0.42, 0.53)

M-CR11K

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.59 (0.53, 0.63)

0.57 (0.51, 0.62)

0.47 (0.42, 0.53)

  1. For each table, the row “M-CR15” denotes classification performance for a model trained on images compressed with compression ratio 15, while each cell is the relevant performance metric and 95% confidence interval for data when tested on data compressed with the CR denoted in the column header.