This paper introduces a prototype of a medical knowledge management system (MKMST) for visual expertise in medical diagnostic reasoning and medical education. The system proposed here, developed following medical expert (radiologists) recommendations, is based on medical image tagging and annotation. Medical images stored in picture archiving and communications systems (PACS) are accessed by medical experts to diagnose pathologies by means of a tagging system and the generation of a taxonomy. The system stores the annotations in a relational database management system and interacts via DICOM standard and HL7 with PACS and the health records. The knowledge base consists of the translation from tacit knowledge to explicit knowledge.