The electrocardiogram, chest x-ray, and skin lesion interpretation are a diagnostic process that applies image analysis. Knowledge and sufficient clinical experience are necessary to achieve expertise in these fields. However, recent advances in medical informatics, particularly in deep learning, are challenging this diagnostic process and physicians' performance. Only a fraction of clinical diagnostic support based on artificial intelligence (AI) has been validated in a clinical environment, limiting its use at the patient's bedside. Gradual AI integration into medical practice will require that the physicians remain able to assess the strengths and limitations of these new algorithms.