Systematic review of multimodal approaches and evaluation metrics in Chinese medical image captioning
Abstract
This review examines Chinese medical image captioning (MIC), with emphasis on multimodal modeling and evaluation metrics for chest X-ray (CXR) report generation. Following the PRISMA 2020 guideline, this review searched IEEE Xplore, ACM Digital Library, ScienceDirect, Web of Science, PubMed, and arXiv for studies published between 2015 and 2025; 60 of 120 assessed full-text records were included. The study was organized around three themes: linguistic and data foundations, multimodal modeling trajectories, and evaluation methods for clinically meaningful assessment. The reviewed literature shows a movement from CNN-RNN encoder-decoder models toward Transformer-based architectures, cross-modal attention, retrieval-augmented generation, knowledge-guided modeling, and vision-language pre-training. In Chinese MIC, however, this modeling shift is constrained by limited public image-report resources, heterogeneous dataset access, and the language-specific effects of segmentation, compact terminology, negation, uncertainty, and clinical modifiers. General Chinese captioning and cross-lingual resources provide useful starting points, but they do not directly resolve the requirements of Chinese clinical reporting. Evaluation remains a distinct limitation in the reviewed literature. Common metrics such as BLEU, ROUGE, METEOR, CIDEr, and SPICE support comparison across studies, but they mainly reflect surface overlap and may fail to capture clinically meaningful errors. Recent work has…
How strong is the evidence?
Evidence Hub reads this as a systematic review. Its evidence level is 1 of 5 (Highest · pooled evidence). Pooled evidence from several studies sits at the top of the evidence pyramid.
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.