Search papers, labs, and topics across Lattice.
This paper introduces Clinical Consensus Selection (CCS), a decoder-agnostic inference-time framework for radiology report generation that samples multiple candidate reports and selects the one with the highest clinical consensus. CCS combines text-based utilities with a radiology-adapted utility computed by an image-report-trained multimodal embedder to measure candidate agreement beyond surface-level textual similarity. Experiments across three datasets and multiple radiology MLLMs demonstrate that CCS consistently improves inference-time performance over single-path decoding and generic Best-of-N baselines, particularly on clinical metrics.
Radiology report generation models often hide clinically superior reports within their candidate pools, and a simple consensus-based selection method can unlock them.
Radiology report generation (RRG) is commonly formulated as a single-path generation task, where a multimodal large language model (MLLM) produces one decoded report as the final output. While recent progress has largely been driven by scaling training data, model capacity, and retrieval mechanisms, improving report quality at inference time remains underexplored. In this work, we observe that fixed radiology MLLMs often generate clinically stronger reports elsewhere in their candidate pool than the one selected by default decoding, suggesting that inference-time decision making remains an overlooked bottleneck. To address this, we propose Clinical Consensus Selection (CCS), a decoder-agnostic inference-time selection framework that samples multiple candidate reports and selects the one with the highest clinical consensus across the rollout pool. CCS unifies text-based utilities with a radiology-adapted utility computed by an image--report-trained multimodal embedder, which measures candidate agreement beyond surface-level textual similarity. Across three datasets and multiple radiology MLLMs, CCS consistently improves inference-time performance over single-path decoding and generic Best-of-N baselines, with particularly clear gains on clinical metrics. Further analysis shows that image-grounded utility forms a selection axis distinct from textual consensus and that substantial headroom remains for improving RRG at inference time.
