# Can Medical Vision-Language Pre-training Succeed with Purely Synthetic Data?

arXiv: https://arxiv.org/abs/2410.13523 · HTML: https://mlsanity.com/paper/2410.13523/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 26 of its 76 references that are in the map.
- First posted on arXiv 17 Oct 2024.

## Placed in

- Multimodal LLMs / vision-language models / [Medical vision-language models](https://mlsanity.com/mllm/medical-vision-language-models/)
- Synthetic data and self-improvement for LLMs / [Medical synthetic data and agents](https://mlsanity.com/synth/medical-synthetic-data-and-agents/)

## Builds on (most cited first, 26 of 26)

- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites](https://mlsanity.com/paper/2404.16821/) (2024)
- [Contrastive Learning of Medical Visual Representations from Paired Images and Text](https://mlsanity.com/paper/2010.00747/) (2020)
- [MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographs](https://mlsanity.com/paper/1901.07042/) (2019)
- [Long-CLIP: Unlocking the Long-Text Capability of CLIP](https://mlsanity.com/paper/2403.15378/) (2024)
- [Demystifying CLIP Data](https://mlsanity.com/paper/2309.16671/) (2023)
- [Multi-Granularity Cross-modal Alignment for Generalized Medical Visual Representation Learning](https://mlsanity.com/paper/2210.06044/) (2022)
- [Exploring scalable medical image encoders beyond text supervision](https://mlsanity.com/paper/2401.10815/) (2024)
- [Scaling Laws of Synthetic Images for Model Training ... for Now](https://mlsanity.com/paper/2312.04567/) (2023)
- [Learning Vision from Models Rivals Learning Vision from Data](https://mlsanity.com/paper/2312.17742/) (2023)
- [RaTEScore: A Metric for Radiology Report Generation](https://mlsanity.com/paper/2406.16845/) (2024)
- [SynthCLIP: Are We Ready for a Fully Synthetic CLIP Training?](https://mlsanity.com/paper/2402.01832/) (2024)
- [LLM-CXR: Instruction-Finetuned LLM for CXR Image Understanding and Generation](https://mlsanity.com/paper/2305.11490/) (2023)
- [Training on Thin Air: Improve Image Classification with Generated Data](https://mlsanity.com/paper/2305.15316/) (2023)
- [M-FLAG: Medical Vision-Language Pre-training with Frozen Language Models and Latent Space Geometry Optimization](https://mlsanity.com/paper/2307.08347/) (2023)
- [MedVersa: A Generalist Foundation Model for Medical Image Interpretation](https://mlsanity.com/paper/2405.07988/) (2024)
- [Decomposing Disease Descriptions for Enhanced Pathology Detection: A Multi-Aspect Vision-Language Pre-training Framework](https://mlsanity.com/paper/2403.07636/) (2024)
- [IMITATE: Clinical Prior Guided Hierarchical Vision-Language Pre-training](https://mlsanity.com/paper/2310.07355/) (2023)
- [MLIP: Enhancing Medical Visual Representation with Divergence Encoder and Knowledge-guided Contrastive Learning](https://mlsanity.com/paper/2402.02045/) (2024)
- [Diversify, Don't Fine-Tune: Scaling Up Visual Recognition Training with Synthetic Images](https://mlsanity.com/paper/2312.02253/) (2023)
- [Synth2: Boosting Visual-Language Models with Synthetic Captions and Image Embeddings](https://mlsanity.com/paper/2403.07750/) (2024)
- [Utilizing Synthetic Data for Medical Vision-Language Pre-training: Bypassing the Need for Real Images](https://mlsanity.com/paper/2310.07027/) (2023)
- [FreeTumor: Advance Tumor Segmentation via Large-Scale Tumor Synthesis](https://mlsanity.com/paper/2406.01264/) (2024)
- [T3D: Advancing 3D Medical Vision-Language Pre-training by Learning Multi-View Visual Consistency](https://mlsanity.com/paper/2312.01529/) (2023)
- [G2D: From Global to Dense Radiography Representation Learning via Vision-Language Pre-training](https://mlsanity.com/paper/2312.01522/) (2023)
- [PairAug: What Can Augmented Image-Text Pairs Do for Radiology?](https://mlsanity.com/paper/2404.04960/) (2024)

## Built on by (most cited first, 7 of 7)

- [GAIA: A Global, Multi-modal, Multi-scale Vision-Language Dataset for Remote Sensing Image Analysis](https://mlsanity.com/paper/2502.09598/) (2025)
- [Logic Unseen: Revealing the Logical Blindspots of Vision-Language Models](https://mlsanity.com/paper/2508.11317/) (2025)
- [Enhancing Test-Time Scaling of Large Language Models with Hierarchical Retrieval-Augmented MCTS](https://mlsanity.com/paper/2507.05557/) (2025)
- [MedVista3D: Vision-Language Modeling for Reducing Diagnostic Errors in 3D CT Disease Detection, Understanding and Reporting](https://mlsanity.com/paper/2509.03800/) (2025)
- [Artificial Intelligence for Biomedical Video Generation](https://mlsanity.com/paper/2411.07619/) (2024)
- [Dual form Complementary Masking for Domain-Adaptive Image Segmentation](https://mlsanity.com/paper/2507.12008/) (2025)
- [A Utility-preserving De-identification Pipeline for Cross-hospital Radiology Data Sharing](https://mlsanity.com/paper/2604.07128/) (2026)

## Abstract (arXiv)

Medical Vision-Language Pre-training (MedVLP) has made significant progress in enabling zero-shot tasks for medical image understanding. However, training MedVLP models typically requires large-scale datasets with paired, high-quality image-text data, which are scarce in the medical domain. Recent advancements in Large Language Models (LLMs) and diffusion models have made it possible to generate large-scale synthetic image-text pairs. This raises the question: "Can MedVLP succeed using purely synthetic data?" To address this, we use off-the-shelf generative models to create synthetic radiology reports and paired Chest X-ray (CXR) images, and propose an automated pipeline to build a diverse, high-quality synthetic dataset, enabling a rigorous study that isolates model and training settings, focusing entirely from the data perspective. Our results show that MedVLP models trained exclusively on synthetic data outperform those trained on real data by 3.8% in averaged AUC on zero-shot classification. Moreover, using a combination of synthetic and real data leads to a further improvement of 9.07%. Additionally, MedVLP models trained on synthetic or mixed data consistently outperform those trained on real data in zero-shot grounding, as well as in fine-tuned classification and segmentation tasks. Our analysis suggests MedVLP trained on well-designed synthetic data can outperform models trained on real datasets, which may be limited by low-quality samples and long-tailed distributions.