In-vitro quantification of the impact of artifacts on metal segmentation and image quality with varying CBCTs, materials and field of views

Objective: To quantify the amount of metal artifact in CBCT images in varying CBCT devices, metal materials and FOVs. Materials and Methods: Three phantoms made with acrylic resin and containing each one 3 metal cylinders of different materials (titanium, cupper-aluminum alloy and amalgam) were imaged scanned using 3 different CBCT devices and 2 FOVs, with a spatial resolution of 0.2 mm. DICOM images were exported for further image processing with MATLAB software. After the segmentation of the metal cylinders and registration of the image volumes, the same ROI was defined for each pair images acquired with different FOV. The difference between the segmented and the real volume of the metal cylinder was evaluated. Moreover, artifacts on the background were assessed by measuring the normalized standard deviation of voxel values in three different ROIs around the metal object. Results: Overestimation of the segmented volume was observed for all CBCTs devices and materials. Within the materials, amalgam and titanium generated, respectively, more and less artifacts in all images. Standard deviation values varied differently between the three ROIs in each device. Conclusion: Different CBCT devices, materials and FOVs should be considered while evaluating CBCT images when metal objects are scanned. In particular, the segmentation of the metal object is highly influenced by the device and material factor. Regarding the background noise, the CBCT devices and the FOV size have more influence on the amount of artifact then the materials.