# From Sampled Evidence to Radiology Reports: Objectives, Constraints, and Realization

arXiv: https://arxiv.org/abs/2610.17318 · HTML: https://mlsanity.com/paper/2610.17318/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 24 of its 41 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Builds on (most cited first, 24 of 24)

- [CLIPScore: A Reference-free Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2104.08718/) (2021)
- [CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison](https://mlsanity.com/paper/1901.07031/) (2019)
- [CheXbert: Combining Automatic Labelers and Expert Annotations for Accurate Radiology Report Labeling Using BERT](https://mlsanity.com/paper/2004.09167/) (2020)
- [RadGraph: Extracting Clinical Entities and Relations from Radiology Reports](https://mlsanity.com/paper/2106.14463/) (2021)
- [Improving Factual Completeness and Consistency of Image-to-Text Radiology Report Generation](https://mlsanity.com/paper/2010.10042/) (2020)
- [GREEN: Generative Radiology Report Evaluation and Error Notation](https://mlsanity.com/paper/2405.03595/) (2024)
- [Improving the Factual Correctness of Radiology Report Generation with Semantic Rewards](https://mlsanity.com/paper/2210.12186/) (2022)
- [Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translation](https://mlsanity.com/paper/2108.04718/) (2021)
- [Epsilon Sampling Rocks: Investigating Sampling Strategies for Minimum Bayes Risk Decoding for Machine Translation](https://mlsanity.com/paper/2305.09860/) (2023)
- [It's MBR All the Way Down: Modern Generation Techniques Through the Lens of Minimum Bayes Risk](https://mlsanity.com/paper/2310.01387/) (2023)
- [RadEval: A framework for radiology text evaluation](https://mlsanity.com/paper/2509.18030/) (2025)
- [Optimizing Non-decomposable Performance Measures: A Tale of Two Classes](https://mlsanity.com/paper/1505.06812/) (2015)
- [On the Bayes-optimality of F-measure maximizers](https://mlsanity.com/paper/1310.4849/) (2013)
- [Mitigating Metric Bias in Minimum Bayes Risk Decoding](https://mlsanity.com/paper/2411.03524/) (2024)
- [Model-Based Minimum Bayes Risk Decoding for Text Generation](https://mlsanity.com/paper/2311.05263/) (2023)
- [Uncertainty-Aware Decoding with Minimum Bayes Risk](https://mlsanity.com/paper/2503.05318/) (2025)
- [X-ray Made Simple: Lay Radiology Report Generation and Robust Evaluation](https://mlsanity.com/paper/2406.17911/) (2024)
- [Diversity Explains Inference Scaling Laws: Through a Case Study of Minimum Bayes Risk Decoding](https://mlsanity.com/paper/2410.15021/) (2024)
- [Semantic Consistency-Based Uncertainty Quantification for Factuality in Radiology Report Generation](https://mlsanity.com/paper/2412.04606/) (2024)
- [Reranking Laws for Language Generation: A Communication-Theoretic Perspective](https://mlsanity.com/paper/2409.07131/) (2024)
- [RA-RRG: Multimodal Retrieval-Augmented Radiology Report Generation with Key Phrase Extraction](https://mlsanity.com/paper/2504.07415/) (2025)
- [EditGRPO: Reinforcement Learning with Post-Rollout Edits for Clinically Accurate Chest X-Ray Report Generation](https://mlsanity.com/paper/2509.22812/) (2025)
- [SDR: Set-Distance Rewards for Radiology Report Generation](https://mlsanity.com/paper/2606.00440/) (2026)
- [Structure-Conditional Minimum Bayes Risk Decoding](https://mlsanity.com/paper/2510.20700/) (2025)

## Abstract (arXiv)

Sampling multiple reports creates an inference problem: how should sampled evidence be converted into a final decision, and does a better structured decision yield a better report? We study these questions through a framework that separates the decision signal, scoring objective, and feasible decision set, followed by text realization. On MIMIC-CXR and IU-Xray, controlled comparisons across 13 generator configurations per benchmark under grouped development/test evaluation reveal three findings. First, changing the objective under fixed evidence and rule capacity changes predictions and their held-out metric trade-offs. Second, relaxing candidate constraints has objective-dependent effects under shared signals and frozen scores. Diagnostics connect the IU macro-oriented assertion increase to near-zero label thresholds and show that limited candidate coverage coexists with selection headroom. Third, a controlled realization study with SFT-Qwen3-VL-8B shows that decision-level gains can be attenuated, amplified, or reversed. Change decomposition reveals that realization can remove incorrect assertions while losing correct findings, so target fidelity and reference accuracy need not improve together. Finding gains also do not consistently improve reference-based report quality. These results motivate separate evaluation of decision rules, candidate constraints, and realized reports.