Automated Detection of Infection in Diabetic Foot Ulcer Images Using Convolutional Neural Network

Table 4

Performance of DFINET in infection classification from DFU images.

Model/classifier	Train/test split	ACC	SEN	SPE	PRE	F1-score	MCC
GoogLeNet	90/10	75.34	73.27	77.82	79.80	76.39	0.50
VGG16	90/10	83.11	81.45	84.95	85.74	83.54	0.66
AlexNet	90/10	76.74	74.96	78.79	80.31	77.54	0.53
DFINET	90/10	91.98	90.57	93.49	93.72	92.12	0.84
DFINET	80/20	89.29	88.46	90.63	93.88	91.09	0.77
DFINET	70/30	87.50	92.78	80.28	86.54	89.55	0.74
Ensemble CNN [22]	90/10	72.70	70.90	74.40	73.50	72.20	0.45
Res7Net [23]	90/10	80.00	80.40	80.20	79.70	79.80	0.60
Inception ResNetV2 [22]	90/10	67.60	68.80	66.40	67.20	68.00	0.35
Bayes Net [22]	90/10	63.90	61.90	66.00	65.30	62.20	0.29