Doctor Sun: A Bilingual Multimodal Large Language Model for Biomedical AI

TL;DR

This work introduces Doctor Sun, a large multimodal generative model specialized in medicine, developed to encode, integrate, and interpret diverse biomedical data modalities such as text and images.

Abstract

Large multimodal models (LMMs) have demonstrated significant potential in providing innovative solutions for various biomedical tasks, including pathology analysis, radiology report generation, and biomedical assistance. However, the existing multimodal biomedical AI is typically based on foundation LLMs, thus hindering the understanding of intricate medical concepts with limited medical training data. Moreover, recent LLaVA-induced medical LMMs struggle to effectively capture the intricate relationship between the texts and the images. Therefore, we introduce Doctor Sun, a large multimodal generative model specialized in medicine, developed to encode, integrate, and interpret diverse biomedical data modalities such as text and images. In particular, Doctor Sun integrates a pre-trained vision encoder with a medical LLM and conducts two-stage training on various medical datasets, focusing on feature alignment and instruction tuning. Moreover, we release SunMed-VL, a wide-range bilingual medical multimodal dataset, along with all associated models, code, and resources, to freely support the advancement of biomedical multimodal research.

Figures (3)

• Figure 1: The main architecture and training process of Doctor Sun.
• Figure 2: (a) The performance of Doctor Sun under different linguistic backbones. (b) Closed-set performance comparative analysis between DocS-M and related work.
• Figure 3: Different data ratios (k) between specific abilities and general abilities on three benchmarks.