Table 14.

Severity quantification performance of the proposed model with and without pre-trained backbone weights on CheXpert dataset.

Methods	Internal validation	External 1 (CNUH)	External 2 (YNU)	External 3 (KNUH)
	MSE	MSE	MSE	MSE
w pre-train	0.528	1.441	1.435	1.458
w/o pre-train	0.607	1.737	2.319${}^{**}$	1.835${}^{*}$

Note:${}^{*}$, ${}^{**}$ denote the better performance of ours with statistical significance ($p<0.05$, $p<0.001$).