Skip to main content
. 2019 Feb 2;5(2):26. doi: 10.3390/jimaging5020026

Table 6.

Model performance measures from cross-training.

Methods SN SP Pr Acc AUC kappa G MCC F1
Test images from: DRIVE
Model
trained on:
STARE 		Soares et al. [10] - - - 0.9397 - - - - -
Ricci et al. [3] - - - 0.9266 - - - - -
Marin et al. [47] - - - 0.9448 - - - - -
Fraz et al. [7] 0.7242 0.9792 - 0.9456 0.9697 - 0.8421 - -
Li et al. [51] 0.7273 0.9810 - 0.9486 0.9677 - 0.8447 - -
Liskowski et al. [52] - - - 0.9416 0.9605 - - - -
Mo et al. [55] 0.7412 0.9799 - 0.9492 0.9653 - 0.8522 - -
PixelBNN 0.5110± 0.0362 0.9533± 0.0094 0.7087± 0.0554 0.8748± 0.0126 0.7322± 0.0199 0.5193± 0.0404 0.6974± 0.0258 0.5309± 0.0422 0.5907± 0.0348
Model
trained on:
CHASE_DB1 		Li et al. [51] 0.7307 0.9811 - 0.9484 0.9605 - 0.8467 - -
Mo et al. [55] 0.7315 0.9778 - 0.9460 0.9650 - 0.8457 - -
PixelBNN 0.6222± 0.0441 0.9355± 0.0085 0.6785± 0.0383 0.8796± 0.0090 0.7788± 0.0204 0.5742± 0.0282 0.7622± 0.0254 0.5768± 0.0279 0.6463± 0.0237
Test images from: STARE
Model
trained on:
DRIVE 		Soares et al. [10] - - - 0.9327 - - - - -
Ricci et al. [3] - - - 0.9464 - - - - -
Marin et al. [47] - - - 0.9528 - - - - -
Fraz et al. [7] 0.7010 0.9770 - 0.9493 0.9660 - 0.8276 - -
Li et al. [51] 0.7027 0.9828 - 0.9545 0.9671 - 0.8310 - -
Liskowski et al. [52] - - - 0.9505 0.9595 - - - -
Mo et al. [55] 0.7009 0.9843 - 0.9570 0.9751 - 0.8306 - -
PixelBNN 0.7842± 0.0552 0.9265± 0.0196 0.6262± 0.1143 0.9070± 0.0181 0.8553± 0.0323 0.6383± 0.0942 0.8519± 0.0343 0.6465± 0.0873 0.6916± 0.0868
Model
trained on:
CHASE_DB1 		Li et al. [51] 0.6944 0.9831 - 0.9536 0.9620 - 0.8262 - -
Mo et al. [55] 0.7387 0.9787 - 0.9549 0.9781 - 0.8503 - -
PixelBNN 0.6973± 0.0372 0.9062± 0.0189 0.5447± 0.0957 0.8771± 0.0157 0.8017± 0.0226 0.5353± 0.0718 0.7941± 0.0245 0.5441± 0.0649 0.6057± 0.0674
Test images from: CHASE_DB1
Model
trained on:
DRIVE 		Li et al. [51] 0.7118 0.9791 - 0.9429 0.9628 - 0.8348 - -
Mo et al. [55] 0.7003 0.9750 - 0.9478 0.9671 - 0.8263 - -
PixelBNN 0.9038± 0.0196 0.8891± 0.0089 0.3886± 0.0504 0.8901± 0.0088 0.8964± 0.0116 0.4906± 0.0516 0.8963± 0.0116 0.5480± 0.0413 0.5416± 0.0513
Model
trained on:
STARE 		Fraz et al. [7] 0.7103 0.9665 - 0.9415 0.9565 - 0.8286 - -
Li et al. [51] 0.7240 0.9768 - 0.9417 0.9553 - 0.8410 - -
Mo et al. [55] 0.7032 0.9794 - 0.9515 0.9690 - 0.8299 - -
PixelBNN 0.7525± 0.0233 0.9302± 0.0066 0.4619± 0.0570 0.9173± 0.0059 0.8413± 0.0132 0.5266± 0.0482 0.8365± 0.0143 0.5475± 0.0412 0.5688± 0.0475