Generative AI-Assisted Medical Illustrations in Pediatric Neurosurgery : Workflows, Ethical Safeguards, and Journal...

Area of Science:

• Medical illustration
• Pediatric neurosurgery
• Artificial intelligence

Background:

• Generative AI offers potential for medical visuals, especially in pediatric neurosurgery where patient data is restricted.
• Ethical concerns include clinical accuracy, intellectual property, and pediatric patient protection, requiring careful oversight.
• Existing policies from major journals like ICMJE and JKNS guide AI usage.

Purpose of the Study:

• To present a human-in-the-loop workflow integrating generative AI and vector editing for accurate, ethical medical illustrations.
• To demonstrate practical applications in pediatric neurosurgery conditions.
• To ensure AI serves as a constrained assistant, maintaining patient privacy and scientific integrity.

Main Methods:

• Reviewed AI usage policies of major medical journals (ICMJE, JKNS).
• Developed a workflow combining clinician sketches, AI refinement, and vector-based digital editing.
• Created illustrative examples for sacral dimple, Crouzon syndrome, and Down syndrome.

Main Results:

• Vector-based editing allows correction of AI "hallucinations" and ensures anatomical precision.
• Journals permit AI for conceptual figures with transparent disclosure, but not as authors.
• The proposed workflow ensures AI-generated images are non-identifiable and respectful of patient dignity.

Conclusions:

• Generative AI can enhance pediatric neurosurgery visualization when ethical and technical safeguards are in place.
• Human-directed validation and adherence to journal policies are crucial for scientific integrity and patient privacy.
• AI should be a "constrained assistant," not an autonomous creator, in medical illustration.