Application of deep learning in wound size measurement using fingernail as the reference

Table 4 Experimental results of the U-net model with different pairs of loss and optimization functions

Loss function	Optimization function	mDSC¹	mPA²	mPS³	mPP⁴
Dice Loss	Adam⁵	0.9250	0.9523	0.9439	0.9068
Dice Loss	SGD⁶	0.8849	0.9256	0.9164	0.8554
Binary Cross Entropy	Adam	0.9232	0.9510	0.9455	0.9055
Binary Cross Entropy	SGD	0.8713	0.9143	0.9302	0.8194

(1) Mean Dice coefficient, (2) mean pixel accuracy, (3) mean pixel sensitivity, (4) mean pixel precision, (5) adaptive moment estimation, (6) stochastic gradient descent

ISSN: 1472-6947