Vision-Language AI for Medical Image Understanding

MedAgent is a multimodal AI assistant designed to bridge the gap between complex medical imaging and human understanding by transforming raw visual data (X-ray CT scan, medical image) into structured, explainable clinical reports. At its core, the system is powered by the Gemini 2.5 Flash Vision-Language Model (VLM), which allows it to jointly reason over both images and natural language text. Developed using Python, Gradio, and Google Colab, MedAgent provides an accessible browser-based interface where users can upload various medical scan such as X-rays, MRIs, and CT scans and pose natural language questions about them. The system's architecture follows a structured agentic loop that includes an AI critique and self-review process to ensure high-quality output. Each analysis results in a comprehensive report containing: Structured Observations: Precise clinical descriptions of visible features. Medical Findings: Interpretations aligned with standard radiology conventions. Plain-Language Explanations: Accessible summaries tailored for students and non-experts. Uncertainty Notes: Transparent flagging of ambiguous or low-confidence areas. While MedAgent is explicitly not a diagnostic tool and does not replace a licensed physician, it serves as a powerful resource for medical education, interactive research analysis, and preliminary triage support. By combining advanced multimodal reasoning with transparent uncertainty flagging, MedAgent enables users to explore imaging cases with guided, AI-assisted insights.