Surface reconstruction of humanoid phantoms using photogrammetry techniques

Background and Purpose The fabrication of surface moulds is considered laborious and time consuming that often result in repeated attempts due to incorrect catheter positioning or the presence of air gaps. The introduction of 3-dimensional (3D) printing using relatively inexpensive and reliable materials has allowed the rapid creation of patient-specific surface mould applicators to be achieved using patient imaging data obtained from a CT scan. In this study we investigate whether an alternative approach using photogrammetry techniques can further improve this process. Photogrammetry is the process of generating 3D models from a set of 2D photographic images. By using photographic images to reconstruct a 3D model of the target area, the need for an initial CT scan can be removed, thus reducing unnecessary dose to patients. This study examined how the individual camera parameters, colour and texture of the object affected the accuracy of the photogrammetry models. Methods Two differently textured humanoid phantoms, an anthropomorphic rando phantom and a Laerdal Little Anne CPR training manikin (CPR phantom) were used in this study. Both were imaged using a Nikon D5600 DSLR and Nokia 3.1 smartphone camera and reconstructed using the photogrammetry software Agisoft Metashape. CT scans of both phantoms were taken as references for comparing the photogrammetry models. Camera parameters were changed to extreme values to emphasize their effects on the reconstruction quality. Models (totalling 46) were reconstructed from different photo sets and their quality assessed by distance to agreement between the photogrammetry and CT models. Results Agisoft Metashape effectively reconstructed both phantoms for all experiments with a good level of accuracy. We found that increasing the ISO produced the worst reconstruction quality while slow shutter speeds could be tolerated by the photogrammetry software if motion blur in the images was noticeably absent. Reconstructions using video data can be comparable to those using photographic images but at a lower resolution. The CPR phantom was reconstructed to a similar level of accuracy to the rando phantom. Background light that caused undesirable reflections from the phantoms significantly reduced the quality of the reconstructions. Applying a non-reflective micropore tape to the affected regions provided a suitable method for reducing the effects of reflections. Conclusion Photogrammetry techniques were effectively able to reconstruct 3D models of the rando and CPR phantom. The camera settings and lighting did have a profound effect on the reconstruction quality and should be chosen appropriately depending on the scene.

Corey Bridger

School of Physical Science, University of Adelaide
Higher Degree by Research Candidate (PhD)

Corey is a first year higher degree by research candidate at the University of Adelaide. Graduating from the University of Adelaide with first class honours in experimental physics for his project on Leipzig applicators, his current research is in the field of HDR surface brachytherapy.