文献收获_2024.06.11

Pancreatic Tumor Segmentation as Anomaly Detection in CT Images Using Denoising Diffusion Models
来源：https://arxiv.org/pdf/2406.02653

• DDM做image translation实现胰腺分割
  下面的是金标准
  
• Anomaly map
  差值出来的
  
• Segmentation map
  算dice系数，用于下游分类
  
• 评价表格
  算出来的dice系数很小
  
  
  1）Threshold是将map图像二值化，目的是计算dice，threshold越小，越少的点被卡掉，对应的肿瘤区域就越大，dice系数就越高。
  2）Given the classifier's imperfect ability to distinguish healthy and non-healthy regions, misclassified cases were excluded from the Dice score calculation.

TSID-Net: a two-stage single image dehazing framework with style transfer and contrastive knowledge transfer
来源：https://link.springer.com/article/10.1007/s00371-024-03511-2

• 利用 source image 和 target image 之间丰富的先验知识，提升转换后图像的真实性
  ❗我们的课题当中也适用，result-driven 可以理解为——图像对比是变了/病灶是去了，但解剖结构也变形了
  Second, most existing dehazing methods are primarily result-driven, which disregards the intermediate process of dehazing, and the rich prior knowledge present in clear and hazy images is not fully utilized, resulting in significant deviations between the dehazed results and the ground truth.