Skip to main content
. 2021 Oct 19;21(20):6936. doi: 10.3390/s21206936

Table 2.

Results obtained by varying the image input size and masks of the HYPER-RETINO system on 1400 retinal images.

No. Data set SE% SP% ACC% PR% MCC% F1% AUC E
1 600 × 700 93 90.5 91.5 92.6 60 90 0.92 0.60
2 512 × 512 92.5 89 88.5 89.2 59 87 0.89 0.65
3 500 × 500 91 88 87.5 86.4 58 84 0.83 0.75
4 450 × 450 89 86 86.5 85.2 56 82 0.82 0.80
5 400 × 400 85 83 83.5 80.6 54 79 0.78 0.90
6 380 × 380 75 71.5 72.5 73.6 51 71 0.70 0.905

MCC: Matthews correlation coefficient, SE: Sensitivity, SP: Specificity, F1: F1 score, ACC: Accuracy, E: Training errors and AUC: Area under the receiver operating curve.