Corpus ID: 218900753

Segmentation Loss Odyssey

@article{Ma2020SegmentationLO,
  title={Segmentation Loss Odyssey},
  author={Jun Ma},
  journal={ArXiv},
  year={2020},
  volume={abs/2005.13449}
}
  • Jun Ma
  • Published 2020
  • Computer Science, Engineering
  • ArXiv
Loss functions are one of the crucial ingredients in deep learning-based medical image segmentation methods. Many loss functions have been proposed in existing literature, but are studied separately or only investigated with few other losses. In this paper, we present a systematic taxonomy to sort existing loss functions into four meaningful categories. This helps to reveal links and fundamental similarities between them. Moreover, we explore the relationship between the traditional region… Expand
AUTO SEG-LOSS: SEARCHING METRIC SURROGATES
Designing proper loss functions is essential in training deep networks. Especially in the field of semantic segmentation, various evaluation metrics have been proposed for diverse scenarios. DespiteExpand
Modality specific U-Net variants for biomedical image segmentation: A survey
TLDR
The U- net framework is described, followed by the comprehensive analysis of the U-Net variants for different medical imaging or modalities such as magnetic resonance imaging, X-ray, computerized tomography/computerized axial tomography, ultrasound, positron emission tomographic, etc. Expand
Assessing the Impact of the Loss Function, Architecture and Image Type for Deep Learning-Based Wildfire Segmentation
Wildfires stand as one of the most relevant natural disasters worldwide, particularly more so due to the effect of global warming and its impact on various societal and environmental levels. In thisExpand
Prior Mask R-CNN Based on Graph Cuts Loss and Size Input for Precipitation Measurement
TLDR
This article proposes a novel instance segmentation network named prior mask R-CNN by fusing prior knowledge for automatic precipitation detection and designed an effective measurement extraction module to get specific measurements, such as the length of precipitations, from the final prediction results of the network. Expand
Investigation on the Effect of the Feature Extraction Backbone for Small Object Segmentation using Fully Convolutional Neural Network in Traffic Signs Application
TLDR
This paper investigates different backbone networks with grouped labels dataset in shape and colour and recommends VGGNet FCN with CE combined with DL for imbalanced datasets and VGG net as a backbone with Cross Entropy (CE) combined with Dice Loss (DL) for balanced dataset. Expand
Cutting-edge 3D Medical Image Segmentation Methods in 2020: Are Happy Families All Alike?
  • J. Ma
  • Computer Science, Engineering
  • ArXiv
  • 2021
TLDR
A comprehensive review of the top methods in ten 3D medical image segmentation challenges during 2020, covering a variety of tasks and datasets and identifies the ”happy-families” practices in the cutting-edge segmentation methods, which are useful for developing powerful segmentation approaches. Expand
Improving Wildfire Severity Classification of Deep Learning U-Nets from Satellite Images
TLDR
Experiments show that the BCE loss function coupled with a double-step U-Net architecture provides better results than current state-of-the-art solutions on a public labeled dataset of European wildfires. Expand
Loss Ensembles for Extremely Imbalanced Segmentation
  • J. Ma
  • Computer Science, Engineering
  • ArXiv
  • 2021
This short paper briefly presents our methodology details of automatic intracranial aneurysms segmentation from brain MR scans. We use ensembles of multiple models trained from different lossExpand
Double-Step deep learning framework to improve wildfire severity classification
TLDR
This work proposes to extend the state-of-the-art approach to automatically detect wildfires in satellite acquisitions and to associate a damage index from a defined scale by designing a configurable Double-Step Framework, which allows inspecting the prediction quality with different lossfunctions and convolutional neural networks used as backbones. Expand
On Power Jaccard Losses for Semantic Segmentation
In this work, a new generalized loss function is proposed called power Jaccard to perform semantic segmentation tasks. It is compared with classical loss functions in different scenarios, includingExpand
...
1
2
...

References

SHOWING 1-10 OF 18 REFERENCES
Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentations
TLDR
This work investigates the behavior of these loss functions and their sensitivity to learning rate tuning in the presence of different rates of label imbalance across 2D and 3D segmentation tasks and proposes to use the class re-balancing properties of the Generalized Dice overlap as a robust and accurate deep-learning loss function for unbalanced tasks. Expand
A Novel Focal Tversky Loss Function With Improved Attention U-Net for Lesion Segmentation
  • Nabila Abraham, N. Khan
  • Computer Science
  • 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)
  • 2019
TLDR
A generalized focal loss function based on the Tversky index is proposed to address the issue of data imbalance in medical image segmentation and improves the attention U-Net model by incorporating an image pyramid to preserve contextual features. Expand
Tversky Loss Function for Image Segmentation Using 3D Fully Convolutional Deep Networks
TLDR
A generalized loss function based on the Tversky index is proposed to address the issue of data imbalance and achieve much better trade-off between precision and recall in training 3D fully convolutional deep neural networks. Expand
3D Segmentation with Exponential Logarithmic Loss for Highly Unbalanced Object Sizes
TLDR
A fast converging and computationally efficient network architecture for accurate segmentation and an exponential logarithmic loss which balances the labels not only by their relative sizes but also by their segmentation difficulties is proposed. Expand
Major Vessel Segmentation on X-ray Coronary Angiography using Deep Networks with a Novel Penalty Loss Function
TLDR
A novel loss function pGD was introduced by adding a term for penalizing false negative and false positive to generalized dice coefficient and DenseNet121 with pGD showed improved segmentation performance compared to GD. Expand
Optimizing Intersection-Over-Union in Deep Neural Networks for Image Segmentation
TLDR
This paper proposes an approach for directly optimizing this intersection-over-union (IoU) measure in deep neural networks and demonstrates that this approach outperforms DNNs trained with standard softmax loss. Expand
Distance Map Loss Penalty Term for Semantic Segmentation
TLDR
A novel distance map derived loss penalty term for semantic segmentation is proposed, using distance maps, derived from ground truth masks, to create a penalty term, guiding the network's focus towards hard-to-segment boundary regions. Expand
Asymmetric Loss Functions and Deep Densely-Connected Networks for Highly-Imbalanced Medical Image Segmentation: Application to Multiple Sclerosis Lesion Detection
TLDR
This paper developed a 3D fully convolutional densely connected network (FC-DenseNet) with large overlapping image patches as input and an asymmetric similarity loss layer based on Tversky index, which led to the lowest surface distance and the best lesion true positive rate. Expand
V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
TLDR
This work proposes an approach to 3D image segmentation based on a volumetric, fully convolutional, neural network, trained end-to-end on MRI volumes depicting prostate, and learns to predict segmentation for the whole volume at once. Expand
U-Net: Convolutional Networks for Biomedical Image Segmentation
TLDR
It is shown that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Expand
...
1
2
...