MedSteer: 通过无训练激活引导进行反事实内窥镜合成
PolypSteer: Counterfactual Endoscopic Synthesis via Training-Free Activation Steering
AI总结：
MedSteer通过无训练激活引导生成反事实内窥镜图像，提升医学影像数据增强效果
摘要：
生成扩散模型越来越多地用于医学影像数据增强，但文本提示无法生成因果训练数据。重新提示会重置整个生成轨迹，改变解剖结构、纹理和背景。基于反向编辑的方法引入了重建误差，导致结构漂移。我们提出了MedSteer，一种无训练激活引导的内窥镜合成框架。MedSteer在扩散变换器的交叉注意力层中为每个对比提示对识别一个病理向量。在推理过程中，它引导图像激活沿此向量，从头生成反事实对，其中唯一的差异是引导的概念。其他结构通过构造得以保留。我们在Kvasir v3和HyperKvasir上进行了三个实验。在跨三个临床概念对的反事实生成中，MedSteer实现了翻转率分别为0.800、0.925和0.950，优于最佳反向基线在概念翻转率和结构保留方面的表现。在染色解缠绕中，MedSteer实现了75%的染色去除，优于20%（PnP）和10%（h-Edit）。在下游息肉检测中，使用MedSteer反事实对增强后，ViT AUC达到0.9755，优于数量匹配的重新提示方法的0.9083，证实了反事实结构驱动增益。代码在链接https://github.com/phamtrongthang123/medsteer
英文摘要：
Generative diffusion models are increasingly used for medical imaging data augmentation, but text prompting cannot produce causal training data. Re-prompting rerolls the entire generation trajectory, altering anatomy, texture, and background. Inversion-based editing methods introduce reconstruction error that causes structural drift. We propose PolypSteer, a training-free activation-steering framework for endoscopic synthesis. PolypSteer identifies a pathology vector for each contrastive prompt pair in the cross-attention layers of a diffusion transformer. At inference time, it steers image activations along this vector, generating counterfactual pairs from scratch where the only difference is the steered concept. All other structure is preserved by construction. We evaluate PolypSteer across three experiments on Kvasir v3 and HyperKvasir. On counterfactual generation across three clinical concept pairs, PolypSteer achieves flip rates of 0.800, 0.925, and 0.950, outperforming the best inversion-based baseline in both concept flip rate and structural preservation. On dye disentanglement, PolypSteer achieves 75% dye removal against 20% (PnP) and 10% (h-Edit). On downstream polyp detection, augmenting with PolypSteer counterfactual pairs achieves ViT AUC of 0.9755 versus 0.9083 for quantity-matched re-prompting, confirming that counterfactual structure drives the gain. Code is at link https://github.com/UARK-AICV/PolypSteer
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
