Summary of the results of the quantitative image analysis of the in vitro and in vivo datasets
| Original CBCT images (no MAR)* | Post-processed CBCT images (with MAR)* | P values† | |
| Tubes | 56.36±25.90 DU | 42.01±6.96 DU | p<0.001 |
| Patients AVMs+dAVFs | 38.30±22.03 DU | 19.83±12.31 DU | |
| Patients AVMs | 37.82±20.31 DU | 17.95±12.72 DU | |
| Patients dAVFs | 39.19±25.43 DU | 23.33±10.96 DU | |
| Patients Artifact-free layer | 8.12±1.39 DU | 8.11±1.54 DU | p=0.908 |
The MAR software was able to reduce the LEA-related artifacts in CBCT significantly in the standardized in vitro models and the patient datasets in the quantitative analysis. There was no difference between the original and the post-processed images on the layers which were not affected by Onyx-related artifacts.
*Data are provided as mean SD±SD.
†P values of the Wilcoxon matched-pairs signed rank test to evaluate statistical differences between the original (no MAR) and the post-processed (with MAR) CBCT images.
AVM, arteriovenous malformation; CBCT, cone-beam CT; dAVF, dural arteriovenous fistula; DU, density units; LEA, liquid embolic agent; MAR, metal artifact reduction.