Birds of the ancient Nile: Species identification in Egyptian animal mummies using multi-resolution computed tomography and deep learning image segmentation
Abstract
Using multi-resolution CT techniques, this project examined the visibility of identifiable skeletal traits in mummified avian bundles from ancient Egypt and the specificity with which avian taxa can be identified with digital 3D scans. Six mummified birds were scanned and processed with a deep learning segmentation algorithm. Three raptors were successfully identified as Falco tinnunculus, a species associated with the Egyptian goddess Isis. Analyses revealed that low-resolution (~110-80 μm) micro-CTs are sufficient for visualizing the bird skeleton (specifically the accessory pygostyle bones and distal wing sesamoid bones), while high-resolution (~30-20 μm) is necessary only for minute cranial details (the scleral ossicle). Clinical CT scans did not provide sufficient resolution to visualize skeletal traits; long bone measurements were found to be accurate with these scans. This research demonstrates the effectiveness of 3D imaging techniques for studying mummified birds and contributes to the growing body of research on ancient Egyptian animal mummies.