Combining Digital Camera Metadata and Artificial Intelligence to Sort Postmortem Gross Photographs: A Novel Approach in Image Classification in Forensic Pathology.

Reviewing postmortem photographs is routine in forensic pathology. Although standard photographic techniques are used, sorting postmortem photographs can be overly time-consuming, especially when reviewing large quantities of photographs. Artificial intelligence may assist in identifying body parts/locations but is challenged in determining spatial relationships such as sidedness, angle, and distance/depth. Metadata stored in modern digital cameras provides physical data that would assist in determining this information, which can be critical in the legal setting. This proof-of-concept study used photograph metadata and artificial intelligence to develop a tool to sort postmortem photographs using living volunteer subjects. This "tool" automatically sorts through general, close-up images (hands, eyes, face, and hands) and mock injury (with a scale) photographs. The study compared using photograph metadata and artificial intelligence in different proportions (hybrid photograph metadata and artificial intelligence vs. artificial intelligence alone). Results demonstrated that using both metadata and artificial intelligence for different tasks was more accurate and efficient than using artificial intelligence alone.