# Clinically Structured Surrogate Rewards for Post-SFT Medical Image Captioning

arXiv: https://arxiv.org/abs/2608.18654 · HTML: https://mlsanity.com/paper/2608.18654/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 20 references that are in the map.
- First posted on arXiv 19 Aug 2026.

## Builds on (most cited first, 13 of 13)

- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [BioBERT: a pre-trained biomedical language representation model for biomedical text mining](https://mlsanity.com/paper/1901.08746/) (2019)
- [BLEURT: Learning Robust Metrics for Text Generation](https://mlsanity.com/paper/2004.04696/) (2020)
- [Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks](https://mlsanity.com/paper/1506.03099/) (2015)
- [Self-critical Sequence Training for Image Captioning](https://mlsanity.com/paper/1612.00563/) (2016)
- [RadGraph: Extracting Clinical Entities and Relations from Radiology Reports](https://mlsanity.com/paper/2106.14463/) (2021)
- [AlignScore: Evaluating Factual Consistency with a Unified Alignment Function](https://mlsanity.com/paper/2305.16739/) (2023)
- [Self-Alignment Pretraining for Biomedical Entity Representations](https://mlsanity.com/paper/2010.11784/) (2020)
- [Multi-domain Clinical Natural Language Processing with MedCAT: the Medical Concept Annotation Toolkit](https://mlsanity.com/paper/2010.01165/) (2020)
- [ROCOv2: Radiology Objects in COntext Version 2, an Updated Multimodal Image Dataset](https://mlsanity.com/paper/2405.10004/) (2024)
- [Radiology Report Generation via Multi-objective Preference Optimization](https://mlsanity.com/paper/2412.08901/) (2024)
- [SC-Captioner: Improving Image Captioning with Self-Correction by Reinforcement Learning](https://mlsanity.com/paper/2508.06125/) (2025)
- [ClaimDiff-RL: Fine-Grained Caption Reinforcement Learning through Visual Claim Comparison](https://mlsanity.com/paper/2605.20278/) (2026)

## Abstract (arXiv)

Medical image captioning requires translating heterogeneous visual evidence into concise clinical descriptions, where errors in findings, assertion states, or anatomical relations can alter clinical meaning despite surface-level fluency. Sequence-level policy optimization can directly optimize complete captions, but common rewards rely on global text similarity, direct image-caption compatibility, or unordered concept overlap, leaving visual neighborhoods and clinical-claim structure implicit. We propose a clinically structured surrogate reward framework for post-SFT medical image captioning. The framework combines biomedical semantic and short-range lexical fidelity with two structured rewards: distributional image-neighborhood alignment, which matches the medical-image-bank distributions induced by reference and generated captions, and clinical graph consistency, which applies maximum-weight one-to-one matching to entities, assertion states, and typed relations. The four rewards are independently normalized within each rollout group, combined with fixed relative weights, and optimized with GDPO. Across organizer-evaluated hidden test sets for the Standard and Synthetical ImageCLEFmedical Caption tracks and three vision-language backbones, the method improves Overall, Relevance, and Factuality over matched SFT baselines in all six backbone-track combinations, with average relative gains of 3.4%, 2.1%, and 5.8%, respectively. Ablations and paired diagnostics indicate that the structured rewards provide complementary signals, reducing image-neighborhood divergence and improving entity-assertion-relation consistency.