TABLE 2.
Pipeline optimization pyramid @ 45 μm resolution.
| Experiments | Optimization composition | Score |
| Stage 1 Global |
To optimize the combination of multiple transforms | |
| P1_01 | Affine (Default) + Syn (Default) | 0.3467 |
| P1_02 | Affine (Default) + B-spline Syn (Default) + Syn (Default) | 0.303 |
| P1_03 | Rigid (Default) + Affine (Default) + Syn (Default) | 0.4269 |
| P1_04 | Rigid (Default) + Affine (Default) + B-spline Syn (Default) + Syn (Default) | 0.3644 |
| P1_05 | Affine (Default) + B-spline Syn (Default) | 0.3333 |
| P1_06 | B-spline Syn (Default) + Syn (Default) | 0.3131 |
| Stage 2 Similarity |
To optimize the similarity metrics | |
| P2_01 | Affine (MI) + B-spline Syn (CC) + Syn (MI) | 0.303 |
| P2_02 | Affine (MI) + B-spline (CC) + Syn (CC) | 0.3606 |
| P2_03 | Affine (MI) + B-spline (MI) + Syn (MI) | 0.3385 |
| P2_04 | Affine (MI) + B-spline (MI) + Syn (CC) | 0.3752 |
| P2_05 | Affine (CC) + B-spline (CC) + Syn (MI) | 0.3186 |
| Stage 3 B-spline |
To tune the multiresolution setting in b-spline stage | |
| P3_11 | - -shrink-factor 10- -smoothing 5 | 0.332 |
| P3_12 | - -shrink-factor 1- -smoothing 5 | 0.323 |
| P3_13 | - -shrink-factor 1- -smoothing 1 | 0.326 |
| P3_21 | - -shrink-factor 10 × 1- -smoothing 2 × 1 | 0.280 |
| P3_22 | - -shrink-factor 10 × 1- -smoothing 10 × 2 | 0.285 |
| P3_23 | - -shrink-factor 10 × 1- -smoothing 10 × 10 | 0.350 |
| P3_24 | - -shrink-factor 2 × 1- -smoothing 2 × 1 | 0.308 |
| P3_31 | - -shrink-factor 10 × 5 × 1- -smoothing 3 × 2 × 1 | 0.277 |
| P3_32 | - -shrink-factor 10 × 5 × 1- -smoothing 10 × 5 × 1 | 0.312 |
| P3_33 | - -shrink-factor 10 × 5 × 1- -smoothing 10 × 10 × 10 | 0.383 |
| P3_34 | - -shrink-factors 3 × 2 × 1- -smoothing 3 × 2 × 1 | 0.300 |
| P3_41 | - -shrink-factor 10 × 7 × 4 × 1- -smoothing 1 × 1 × 1 × 1 | 0.274 |
| P3_42 | - -shrink-factor 10 × 7 × 4 × 1- -smoothing 4 × 3 × 2 × 1 | 0.268 |
| P3_43 | - -shrink-factor 10 × 7 × 4 × 1- -smoothing 10 × 7 × 4 × 1 | 0.362 |
| P3_44 | - -shrink-factor 10 × 7 × 4 × 1- -smoothing 10 × 10 × 10 × 10 | 0.495 |
| P3_45 | - -shrink-factor 4 × 3 × 2 × 1- -smoothing 4 × 3 × 2 × 1 | 0.278 |
| P3_51 | - -shrink-factor 9 × 7 × 5 × 3 × 1- -smoothing 9 × 7 × 5 × 3 × 1 | 0.292 |
| P3_52 | - -shrink-factor 9 × 7 × 5 × 3 × 1- -smoothing 5 × 4 × 3 × 2 × 1 | 0.355 |
| Stage 4 B-spline distance |
To tune b-spline spline distance | |
| P4_00 | Spline distance default to 26 | 0.268 |
| P4_01 | Spline distance = 10 | 0.341 |
| P4_02 | Spline distance = 40 | 0.268 |
| P4_03 | Spline distance = 60 | 0.268 |
| Stage 5 Syn |
Tuning the multiresolution setting in SyN stage | |
| P5_01 | - -smoothing 3 × 2 × 1 × 0- -shrink 4 × 3 × 2 × 1 | 0.2679 |
| P5_02 | - -smoothing 10 × 7 × 4 × 1- -shrink 10 × 7 × 4 × 1 | 0.3543 |
| P5_03 | - -smoothing 10 × 7 × 4 × 1- -shrink 4 × 3 × 2 × 1 | 0.3084 |
| P5_04 | - -smoothing 1 × 1 × 1 × 1- -shrink 4 × 3 × 2 × 1 | 0.2703 |
| P5_05 | - -smoothing 0 × 0 × 0 × 0- -shrink 4 × 3 × 2 × 1 | 0.2637 |
| P5_06 | - -smoothing 0 × 0 × 0 × 0- -shrink 6 × 4 × 2 × 1 | 0.2626 |
| P5_07 | - -smoothing 0 × 0 × 0 × 0- -shrink 10 × 7 × 4 × 1 | 0.2625 |
| P5_08 | - -smoothing 0 × 0 × 0 × 0 × 0- -shrink 20 × 15 × 10 × 5 × 1 | 0.2633 |
| P5_09 | - -smoothing 3 × 2 × 1 × 0- -shrink 10 × 7 × 4 × 1 | 0.2656 |
Specimen is 191209-1-1. The steps and parameters for the pipes that were tested are summarized for each stage. For each stage only the parameters to be optimized will change, and one optimal pipe will be selected among the pipes within one stage. The aim of the pipeline initialization is to select an optimal registration variables for certain contrasts in MRH/LSM. The pipeline optimization has been performed using one specimen. The application to additional specimens and contrast combinations has been demonstrated in Supplementary Table 2.