用双模型蒸馏加速医学图像生成，提升精度且保护隐私
Adaptively Distilled ControlNet: Accelerated Training and Superior Sampling for Medical Image Synthesis
- 通过教师-学生模型蒸馏，仅用掩码实现精准图像合成
- 在KiTS19和Polyps数据集上分割指标提升2.4%~3.5%
- 训练快、采样只用学生模型，适合医疗数据隐私场景
医学图像标注受限于隐私问题和人工标注成本，严重制约分割模型性能与泛化能力。尽管掩码可控的扩散模型在图像合成方面表现优异，但难以实现病灶掩码的精确对齐。本文提出自适应蒸馏控制网（Adaptively Distilled ControlNet），一种无需任务依赖的框架，通过双模型蒸馏加速训练与优化。训练时，教师模型基于掩码-图像对，在参数空间中通过预测噪声对齐来规范仅含掩码的学生模型，并引入基于病灶-背景比例的自适应正则化。采样阶段仅使用学生模型，实现隐私保护的医学图像生成。在两个不同医学数据集上的全面评估表明，该方法达到当前最优性能：TransUNet在KiTS19上mDice/mIoU分别提升2.4%/4.2%，SANet在Polyps上分别提升2.6%/3.5%，充分验证其有效性和优越性。代码已开源。
原文摘要 · Abstract (English)
Medical image annotation is constrained by privacy concerns and labor-intensive labeling, significantly limiting the performance and generalization of segmentation models. While mask-controllable diffusion models excel in synthesis, they struggle with precise lesion-mask alignment. We propose \textbf{Adaptively Distilled ControlNet}, a task-agnostic framework that accelerates training and optimization through dual-model distillation. Specifically, during training, a teacher model, conditioned on mask-image pairs, regularizes a mask-only student model via predicted noise alignment in parameter space, further enhanced by adaptive regularization based on lesion-background ratios. During sampling, only the student model is used, enabling privacy-preserving medical image generation. Comprehensive evaluations on two distinct medical datasets demonstrate state-of-the-art performance: TransUNet improves mDice/mIoU by 2.4%/4.2% on KiTS19, while SANet achieves 2.6%/3.5% gains on Polyps, highlighting its effectiveness and superiority. Code is available at GitHub.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。