AutoScanAI

AutoScanAI is an AI-powered system designed for radiological research and educational applications. It uses U-Net deep learning architecture to automatically detect and segment tumors in MRI scans. A key focus is interpretability, achieved through explainable AI (Grad-CAM heatmaps), which visually shows users which regions the model focused on. The final visual demo runs in a Streamlit web app, simulating a radiology assistant tool.

Key Features:

  • U-Net Segmentation: Automatically generates pixel-level tumor masks on MRI scans.

  • Explainable AI (XAI): Uses Grad-CAM overlays to show the model's focus, enhancing trust and transparency.

  • Streamlit Dashboard: Provides an interactive prototype for users to upload scans and explore predictions.

  • Software-only Visuals: Every stage produces visual artifacts: scan slices, segmentation masks, and heatmaps.

Conclusion: AutoScanAI highlights the role of AI in assisting medical professionals and enhancing early diagnosis through practical, reproducible, and visually compelling deep learning methods.

AI Lab Journal

AI Lab Journal

AI Lab Journal

AutoScanAI: Latest Insights

Follow the story of our R&D. Get behind-the-scenes updates, technical data, and team insights.

Follow the Progress

Follow Build in Public

For latest updates on this project, including weekly behind-the-scenes media and code snippets follow us.

Follow the Progress

Follow Build in Public

For latest updates on this project, including weekly behind-the-scenes media and code snippets follow us.

Follow the Progress

Follow Build in Public

For latest updates on this project, including weekly behind-the-scenes media and code snippets follow us.