TC-SSA: Token Compression via Semantic Slot Aggregation for Gigapixel Pathology Reasoning

Researchers propose TC-SSA, a token compression framework that enables large vision-language models to process gigapixel pathology images by reducing visual tokens to 1.7% of original size while maintaining diagnostic accuracy. The method achieves 78.34% overall accuracy on SlideBench and demonstrates strong performance across multiple cancer classification tasks.

• →TC-SSA compresses gigapixel pathology images to just 1.7% of original token count while preserving diagnostic information.
• →The framework uses semantic slot aggregation with gated routing to overcome computational bottlenecks in medical imaging AI.
• →Model achieved 78.34% accuracy on SlideBench and over 95% AUC on multiple cancer classification datasets.
• →Solution addresses critical scalability issues for applying large vision-language models to computational pathology.
• →Method outperforms traditional sampling approaches under comparable computational budgets.