Pretreatment patient-specific quality assurance prediction based on 1D complexity metrics and 3D planning dose: classification, gamma passing rates, and DVH metrics

Table 2 Evaluation indicators of various predictive models with various inputs

		TPS deviation	Combined Model	1D Model	3D Model (Transformer)	3D Model (Res-Net)	3D Model (U-Net)
MAPE (%)	PTV D₉₅	2.55	1.23	1.52	1.66	1.69	2.13
	PTV D₂	3.67	2.36	2.67	2.45	2.62	2.73
	PTV D_mean	2.36	1.46	1.70	1.71	1.85	2.12
MAE	GPR	/	0.039	0.046	0.040	0.040	0.044
	PTV HI	0.041	0.018	0.021	0.019	0.020	0.019
	PTV CI	0.125	0.075	0.078	0.084	0.094	0.091
AUC	Classfication	/	0.92	0.88	0.86	0.85	0.81
Number of parameters		/	1.9 M	44 K	1.9 M	3.6 M	3.5 M

Note: MAPE: mean absolute percentage error, MAE: mean absolute error, AUC: area under the receiver operator characteristic curve, TPS deviation: ground truth deviations between DVH metrics calculated by TPS and measure-based QA tools

ISSN: 1748-717X