A Novel Focal Tversky loss function with improved Attention U-Net for lesion segmentation

Code

• github.com/nabsabraham/focal-tversky-unet
  OfficialIn papertf★ 0
• github.com/EvgenyDyshlyuk/Oil_Seep_Detection
  pytorch★ 4
• github.com/Jo-dsa/SemanticSeg
  pytorch★ 0
• github.com/woans0104/sk_project
  none★ 0
• github.com/SahinTiryaki/Brain-tumor-segmentation-Vgg19UNet
  tf★ 0
• github.com/woans0104/project_review
  none★ 0

Abstract

We propose a generalized focal loss function based on the Tversky index to address the issue of data imbalance in medical image segmentation. Compared to the commonly used Dice loss, our loss function achieves a better trade off between precision and recall when training on small structures such as lesions. To evaluate our loss function, we improve the attention U-Net model by incorporating an image pyramid to preserve contextual features. We experiment on the BUS 2017 dataset and ISIC 2018 dataset where lesions occupy 4.84% and 21.4% of the images area and improve segmentation accuracy when compared to the standard U-Net by 25.7% and 3.6%, respectively.

Tasks

Reproductions