Medical hierarchical image classification via dual-geometry image–text learning

Key Points

• The aim is to enhance the accuracy of medical image classification using a dual-geometry approach.
• Implemented a novel dual-geometry image-text learning framework for classification.
• Conducted experiments across three datasets to evaluate performance improvements.
• Utilized various backbone models for comparison and integration.
• Achieved an average accuracy improvement of 7% across all three datasets at the fine-grained level.
• Consistent gains were noted when integrating with other backbone models.