Donated bodies used to generate high-precision anatomic data for medical AI systems

A June 2026 study in Clinical Anatomy highlights the transition of donated bodies into high-precision foundational data for medical AI systems. Using advanced imaging like CT and cryosectioning, researchers can reconstruct organs like the heart at sub-0.2 mm resolutions. However, significant challenges remain, including physiological differences between living and postmortem tissues, low AI anatomical recognition accuracy (such as ChatGPT-4o's 22.26% accuracy), and unresolved ethical questions regarding donor consent and data governance.

Donated bodies for AI

• ▪A study published in Clinical Anatomy in June 2026 outlined the potential use of donated bodies to provide high-precision anatomic data for medical AI systems.
• ▪Spanish Anatomical Society Vice President Damián Sánchez-Quintana stated that medical research is shifting from viewing donated bodies as objects of study to using them as foundational data for AI.
• ▪Anatomic data from donated bodies can support interactive three-dimensional atlases, augmented reality applications, and the development of anatomic digital twins.

High-resolution anatomic imaging techniques

• ▪Using micro-computed tomography and cryosections at 0.1 mm, researchers can scan and segment organs from cadavers in multiple spatial planes without movement or respiratory artifacts.
• ▪Current imaging technologies can reconstruct a donated human heart at a resolution of less than 0.2 mm, recording its texture and histology in three dimensions.
• ▪Medical researchers use technologies such as CT, photogrammetry, laser scanning, and cryosectioning to create detailed three-dimensional models from donated human bodies.

Postmortem versus living anatomy

• ▪Spanish Anatomical Society Vice President Damián Sánchez-Quintana stated that digital twins of donated organs are anatomically perfect but physiologically dead, lacking living tone, pressure, and movement.
• ▪Postmortem tissues can change position and appearance due to the lack of blood flow, muscle contraction, sympathetic tone, and formalin fixation in donated bodies.

AI anatomic recognition accuracy

• ▪A study by Jagiellonian University Medical College researchers found that ChatGPT-4o had an overall accuracy of 22.26% in identifying 265 marked anatomic structures on cadaveric specimens.
• ▪The Jagiellonian University Medical College study showed ChatGPT-4o achieved 64.71% accuracy for osteological structures but only 8.82% accuracy for isolated thoracic organs.

Donor consent requirements

• ▪A study published in Clinical Anatomy in June 2026 noted that clear rules governing donor consent and commercial use are needed as medical AI technologies develop.
• ▪Spanish Anatomical Society Vice President Damián Sánchez-Quintana stated that bioethics committees must address moral dilemmas regarding donor consent for digitization, data sharing, and commercial use of donor data.

Data governance frameworks

• ▪A study published in Clinical Anatomy in June 2026 identified questions surrounding data governance and the high cost of detailed annotation as limitations of using digitized cadaver data.
• ▪Spanish Anatomical Society Vice President Damián Sánchez-Quintana stated that questions remain regarding how anatomic data obtained from donated bodies should be stored, shared, and used.

Debatable claims

• ▪Companies should be prohibited from commercializing AI models trained on digitized donated bodies
• ▪Medical institutions must obtain explicit donor consent before digitizing cadavers for AI