Criação de um protótipo objeto simulador para o controle de qualidade de algoritmos de segmentação óssea para o uso em softwares de pós processamento tomográfico

Abstract

Artificial intelligence is revolutionizing medicine, particularly in radiology, thanks to increased computational power and theoretical depth. However, the reliability of AI algorithms in medical imaging raises ethical and quality concerns. This study developed a 3D simulator to evaluate CT bone segmentation, highlighting the lack of explainability of available AI models. The results highlight the dependence of density on shape in bone segmentation, emphasizing the need for quality control before widespread clinical use.