AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Learning with Counterfactual Explanations for Radiology Report Generation

Mingjie Li, Haokun Lin, Liang Qiu, Xiaodan Liang, Ling Chen, Abdulmotaleb Elsaddik, Xiaojun Chang

ECCV 2024 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded visual concept learning

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add pre-trained PubMedBERT and GPT-2 Medium as text encoder and decoder (ablation)BLIP base · IU-XrayCIDEr: 0.363 -> 0.510Improved
add counterfactual contrastive loss L_CF (ablation)LLM-based model · IU-XrayCIDEr: 0.510 -> 0.678Improved
add counterfactual loss on top of LLM and prompt (ablation)LLM + prompt · IU-XrayCIDEr: 0.698 -> 0.731Improved
add learnable factual and counterfactual prompt (ablation)LLM-based model · IU-XrayCIDEr: 0.510 -> 0.698Improved
add learnable prompt (ablation)LLM-based model · MIMIC-CXR clinical efficacyF1-score: 0.317 -> 0.366Improved
add L_CF to full model for clinical metrics (ablation)LLM + prompt · MIMIC-CXR clinical efficacyF1-score: 0.366 -> 0.405Improved
use BLEU-based hard negative sampling instead of random sampling (ablation)CoFE · IU-XrayCIDEr: 0.653 -> 0.678Improved
use BLEU-based negative sampling in the full model (ablation)CoFE · IU-XrayCIDEr: 0.706 -> 0.731Improved
CoFE compared with prior methods on MIMIC-CXR NLGCoFE · MIMIC-CXRCIDEr: CIDEr 0.453 vs RGRG 0.495; ROUGE-L 0.304Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy