AI Paper Atlas

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

BiOTPrompt: Bidirectional Optimal Transport Guided Prompting for Disease Evolution-aware Radiology Report Generation

Tengfei Liu, Yijian Fan, Boyue Wang, Yongli Hu, Mingjie Li, Jinghua Li, Junbin Gao, Xiaojun Chang, Zhihui Li, Baocai Yin

CVPR 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Medical vision-language models

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 bidirectional OT (vs unidirectional alternative) (ablation)BiOTPrompt · Longitudinal-MIMIC testCE F1: 0.388 -> 0.417Improved
add Dual-Flow Asymmetry Analyzer (ablation)BiOTPrompt · Longitudinal-MIMIC testCE F1: 0.382 -> 0.417Improved
add vision-language consistency constraint (ablation)BiOTPrompt · Longitudinal-MIMIC testCE F1: 0.376 -> 0.417Improved
add classification losses (Lvis-cls + Ltext-cls) (ablation)BiOTPrompt · Longitudinal-MIMIC testCE F1: 0.384 -> 0.417Improved
add KL divergence loss (ablation)BiOTPrompt · Longitudinal-MIMIC testCE F1: 0.399 -> 0.417Improved
add KL divergence loss (ablation)BiOTPrompt · Longitudinal-MIMIC testROUGE-L: 0.284 -> 0.285Improved
use fixed threshold instead of Top-K (ablation)asymmetric region selection · Longitudinal-MIMIC testCIDEr: 0 230 -> 0.236Improved
use soft sigmoid threshold instead of fixed threshold (ablation)asymmetric region selection · Longitudinal-MIMIC testCIDEr: 0.236 -> 0.203Hurt
use learnable scalar threshold instead of fixed threshold (ablation)asymmetric region selection · Longitudinal-MIMIC testCIDEr: 0.236 -> 0.228Hurt
vary asymmetry threshold delta from 0.04 to 0.09 (ablation)Dual-Flow Asymmetry Analyzer · Longitudinal-MIMIC testF1: F1 peaks around 0.05 (0.417); ROUGE-L 0.273 to 0.285Mixed
use prior images via BiOT prompts instead of R2GenGPT concatenationR2GenGPT · Longitudinal-MIMIC testCE F1: 0.316 -> 0.417Improved

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