Owais Ahmad@Owais1122

GemmaSight is an explainable AI-powered pathology assistant designed to support the early detection of cancer from digital pathology slide images. Traditional cancer diagnosis often requires specialized laboratories, expert pathologists, and significant processing time, making timely diagnosis difficult, especially in remote and underserved regions. GemmaSight addresses these challenges by providing fast, transparent, and AI-assisted preliminary screening. The system first processes a pathology slide image using Google's Pathology Foundation Model, which extracts rich visual embeddings that capture microscopic tissue features. These embeddings are then analyzed by a custom-trained classifier to predict whether the tissue is Cancer or Non-Cancer. Our classifier achieved approximately 94% accuracy with an AUC score of 0.98, demonstrating strong predictive performance. To improve trust and interpretability, GemmaSight integrates a FAISS-based similarity retrieval system. Instead of providing only a prediction, the system retrieves the three most similar pathology cases from its embedding database, allowing clinicians to compare the current sample with previously analyzed cases. Additionally, an occlusion heatmap highlights the image regions that most influenced the model's prediction, providing visual evidence of the AI's decision-making process. Finally, the pathology image, prediction, retrieved similar cases, and explainability outputs are passed to the MedGemma Vision-Language Model, which generates a concise, high-level medical report in approximately 30 seconds. GemmaSight is designed to assist—not replace—medical professionals. By combining accurate classification, explainable AI, similarity retrieval, and automated report generation into a single workflow, it enables faster, more transparent, and more accessible AI-assisted pathology, helping clinicians make informed decisions and improving access to quality healthcare in resource-limited settings.

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.