Dice Loss

Dice loss is a widely used loss function in image segmentation, particularly in medical imaging, aiming to optimize the Dice similarity coefficient, a metric measuring the overlap between predicted and ground truth segmentations. Current research focuses on improving Dice loss's robustness to various challenges, including class imbalance, noisy labels, and the segmentation of objects with varying sizes and complex boundaries, often incorporating it within U-Net or similar architectures. These advancements enhance the accuracy and reliability of automated segmentation in diverse applications, from medical image analysis (e.g., vascular and organ segmentation) to 3D object detection in autonomous driving, leading to improved diagnostic capabilities and technological advancements.

Papers

October 10, 2024

Adaptive Real-Time Multi-Loss Function Optimization Using Dynamic Memory Fusion Framework: A Case Study on Breast Cancer Segmentation
Amin Golnari, Mostafa Diba
Case Study Loss Function Breast Tumor Segmentation Dynamic Memory Dice Loss Multi Loss Fusion Framework

July 1, 2024

Centerline Boundary Dice Loss for Vascular Segmentation
Pengcheng Shi, Jiesi Hu, Yanwu Yang, Zilve Gao, Wei Liu, Ting Ma
Vascular Segmentation Improved Box Dice Dice Loss

June 7, 2024

Contour-weighted loss for class-imbalanced image segmentation
Zhhengyong Huang, Yao Sui
Cross Entropy Abdominal Organ Segmentation Dice Loss Class Imbalanced

March 29, 2024

SeaBird: Segmentation in Bird's View with Dice Loss Improves Monocular 3D Detection of Large Objects
Abhinav Kumar, Yuliang Guo, Xinyu Huang, Liu Ren, Xiaoming Liu
Segmentation Based Approach 3D Detection Bird Specie Monocular 3D Detection Bird'S Eye View Map Segmentation Monocular 3D Detector Learning Depth Dice Loss Large Object

March 21, 2024

Depth-aware Panoptic Segmentation
Tuan Nguyen, Max Mehltretter, Franz Rottensteiner
Convolutional Neural Network Instance Segmentation Panoptic Segmentation Dice Loss Uncertainty Aware Panoptic

April 9, 2023

On the dice loss gradient and the ways to mimic it
Hoel Kervadec, Marleen de Bruijne
Gradient Descent Natural Gradient New Way Supervised Semantic Segmentation Dice Loss

April 3, 2023

Noisy Image Segmentation With Soft-Dice
Marcus Nordström, Henrik Hult, Atsuto Maki, Fredrik Löfman
Medical Image Segmentation Automatic Thresholding Improved Box Dice Dice Loss Soft Segmentation

March 28, 2023

Dice Semimetric Losses: Optimizing the Dice Score with Soft Labels
Zifu Wang, Teodora Popordanoska, Jeroen Bertels, Robin Lemmens, Matthew B. Blaschko
Soft Label Label Smoothing Segmentation Pipeline Improved Box Dice Dice Loss

November 8, 2022

Learning Spatio-Temporal Model of Disease Progression with NeuralODEs from Longitudinal Volumetric Data
Dmitrii Lachinov, Arunava Chakravarty, Christoph Grechenig, Ursula Schmidt-Erfurth, Hrvoje Bogunovic
Deep Learning LeArning Abstract Spatio Temporal Longitudinal Image Atrophy Pattern Dice Loss

September 25, 2022

Contour Dice loss for structures with Fuzzy and Complex Boundaries in Fetal MRI
Bella Specktor Fadida, Bossmat Yehuda, Daphna Link Sourani, Liat Ben Sira, Dafna Ben Bashat, Leo Joskowicz
Fetal Brain Simple Structure Fetal MRI Dice Loss Placenta Segmentation Dice Similarity Coefficient Uncertain Boundary

September 22, 2022

A CT-Based Airway Segmentation Using U$^2$-net Trained by the Dice Loss Function
Kunpeng Wang, Yuexi Dong, Yunpu Zeng, Zhichun Ye, Yangzhe Wang
Airway Segmentation Airway Tree Dice Loss

September 13, 2022

On the Optimal Combination of Cross-Entropy and Soft Dice Losses for Lesion Segmentation with Out-of-Distribution Robustness
Adrian Galdran, Gustavo Carneiro, Miguel Ángel González Ballester
Loss Function Cross Entropy Loss Lesion Segmentation Item Combination Distribution Robustness Dice Loss

July 19, 2022

The Dice loss in the context of missing or empty labels: Introducing $\Phi$ and $\epsilon$
Sofie Tilborghs, Jeroen Bertels, David Robben, Dirk Vandermeulen, Frederik Maes
Loss Function Medical Image Segmentation Context Information Dice Loss Multiclass Segmentation