Recent advances have utilized large language models for tabular reasoning, but real-world tables often appear as images.A new framework called TabUlar Reasoning with Bridged infOrmation (Turbo) has been introduced for multimodal tabular reasoning.Turbo leverages structured information available during training to enhance multimodal large language models.Experimental results show that Turbo achieves state-of-the-art performance in tabular reasoning with limited data.