Skip to main content
. 2017 Aug 18;7:8717. doi: 10.1038/s41598-017-08969-w

Table 1.

Quantitative evaluation of representation-based registration on the clinical and synthetic testing sets.

Evaluation methods Intensity-based Registration MACMI MSERg SIFT
tumor lesion DSC 0.59 ± 0.17 0.52 ± 0.18 0.66 ± 0.10 0.54 ± 0.16
capsule boundary DSC 0.93 ± 0.02 0.95 ± 0.03 0.96 ± 0.01 0.84 ± 0.04
RSMD 3.97 ± 0.70 3.64 ± 1.02 2.96 ± 0.76 3.89 ± 1.17
MDD (T1 to PD) 33.78 ± 17.80 2.53 ± 1.05 1.45 ± 0.43 3.51 ± 0.73
MDD (T2 to PD) 41.89 ± 16.19 1.82 ± 1.00 0.50 ± 0.28 1.36 ± 0.62

(N = 25 for both datasets) The table shows the mean and the standard derivation of tumor DSC, prostate boundary DSC and landmark RMSD for clinical dataset and MDD for synthetic dataset. Note that the greater the DSC, the lower the RMSD and MDD indicate the more accurate result.