Table 2 Mean absolute error for the prediction of planning target volume by each model.

From: Prescription‑dose stratification improves deep learning‑based VMAT dose prediction in locally advanced NSCLC

Matrices

Mean ± SD

Model 1 (50 Gy)

Model 2 (54 Gy)

Model 3 (60 Gy)

Model 4 (50 & 60 Gy)

D99% (Gy)

3.08 ± 2.72

3.21 ± 3.06

4.01 ± 3.37*

3.62 ± 3.36

D98% (Gy)

2.30 ± 2.46

2.53 ± 2.63

2.96 ± 3.09

3.47 ± 2.60*

D95% (Gy)

1.87 ± 2.10

1.96 ± 2.16

2.23 ± 2.37

2.73 ± 1.81*

Dmean (Gy)

0.48 ± 0.49

0.52 ± 0.35

0.58 ± 0.49

1.27 ± 0.57*

D2cc (Gy)

0.49 ± 0.54

0.54 ± 0.40*

0.57 ± 0.48

1.27 ± 0.58*

D5cc (Gy)

0.74 ± 0.88

0.75 ± 0.74

0.83 ± 0.90

1.44 ± 0.48*

V95% (%)

4.68 ± 5.44

4.01 ± 5.01*

4.87 ± 5.89

7.70 ± 4.20*

HI

0.09 ± 0.08

0.09 ± 0.09

0.09 ± 0.08

0.09 ± 0.10

  1. *Indicates a statistically significant difference between predicted and reference values (p < 0.05)
Back to article page