Text this: Visual-textual integration in LLMs for medical diagnosis: A preliminary quantitative analysis