From text to image: challenges in integrating vision into ChatGPT for medical image interpretation | Synapse