Dose prediction of CyberKnife Monte Carlo plan for lung cancer patients based on deep learning: robust learning of variable beam configurations

Table 2 The critical indicators of the target area and SSIM in lung cases (mean ± SD)

Input	nCI	HI	GI	SSIM(Body)	SSIM(PTV)
Clinical plan	1.0173 ± 0.0226	1.3435 ± 0.0210	1.0171 ± 0.0221	–	–
MaskO	1.4943 ± 0.2675	1.4095 ± 0.0414	1.4924 ± 0.2644	0.8836 ± 0.0636	0.9873 ± 0.0069
Mask	1.0201 ± 0.0232	1.3265 ± 0.0266	1.0201 ± 0.0232	0.9088 ± 0.0378	0.9954 ± 0.0027
MaskR	1.0455 ± 0.0623	1.3402 ± 0.0318	1.0453 ± 0.0618	0.9363 ± 0.0240	0.9951 ± 0.0016
ABO	1.0209 ± 0.0232	1.3299 ± 0.0307	1.0208 ± 0.0230	0.9636 ± 0.0107	0.9953 ± 0.0018
AB	1.0141 ± 0.0185	1.3057 ± 0.0276	1.0140 ± 0.0182	0.9554 ± 0.0128	0.9952 ± 0.0019
ABR	1.0085 ± 0.0141	1.3093 ± 0.0312	1.0083 ± 0.0135	0.9794 ± 0.0043	0.9955 ± 0.0020

Annotation The target area’s nCI, HI, and GI are critical indicators to evaluate plan quality in clinical practice. The closer the nCI is to 1, the better the result is. The structural similarity index measure (SSIM) indicates the similarity quantifier between the observed and target images, between 0 and 1. The closer to 1, the better the result is

ISSN: 1748-717X