Optimizing Photogrammetry Parameters for 3D Reconstruction of Fresh Brain Specimens

Photogrammetry has become an essential tool in medical and research fields for generating high-fidelity 3D models from 2D images. However, optimizing the imaging and processing parameters remains a challenge, particularly for fresh brain specimens, where repeated imaging is not feasible. This study systematically evaluates the impact of various Metashape alignment settings on 3D reconstruction outcomes, analyzing 12,600 configuration combinations and 63,000 alignments. Our results suggest that the optimal imaging setup consists of four photo sets: two captured at anatomical position (level with the brain and 30 cm above it at a 30° camera tilt) and two identical sets with the basal side facing upwards. The brain should be rotated 3° between shots, generating 120 images per set. Initial processing should be performed without masks using medium-precision alignment in Metashape. If alignment fails, we recommend generating one mask per image, delineating the brain's borders, and applying masks to key points. Notably, higher image density only improves alignment reliability when masking is selected to detected features and may only increase processing time. We also observed variability in alignment results under identical conditions, suggesting an inherent stochastic component in Metashape. Consequently, unsuccessful alignments should be repeated before modifying imaging parameters. To our knowledge, this is the first study to systematically define an optimal imaging and processing parameters for 3D photogrammetry of fresh brains using a turntable. Future research should focus on determining the minimum number of images required to ensure high-quality reconstructions.