Human-Machine Collaboration for Medical Image Segmentation

@article{Last2020HumanMachineCF,
  title={Human-Machine Collaboration for Medical Image Segmentation},
  author={F. Last and Tassilo Klein and Mahdyar Ravanbakhsh and Moin Nabi and K. Batmanghelich and Volker Tresp},
  journal={ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2020},
  pages={1040-1044}
}
  • F. Last, T. Klein, Volker Tresp
  • Published 1 May 2020
  • Computer Science
  • ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Image segmentation is a ubiquitous step in almost any medical image study. Deep learning-based approaches achieve state-of-the-art in the majority of image segmentation benchmarks. However, end-to-end training of such models requires sufficient annotation. In this paper, we propose a method based on conditional Generative Adversarial Network (cGAN) to address segmentation in semi-supervised setup and in a human-in-the-loop fashion. More specifically, we use the generator in the GAN to… 

Figures and Tables from this paper

Disentanglement based Active Learning
TLDR
DAL is a new active learning technique based on query synthesis which leverages the concept of disentanglement and automatically labels majority of the datapoints, thus drastically reducing the human labelling budget in active learning.
A Survey of Human-in-the-loop for Machine Learning
TLDR
This survey intends to provide a high-level summarization for human-in-the-loop and to motivate inter-ested readers to consider approaches for designing effective human- in- the-loop solutions.

References

SHOWING 1-10 OF 33 REFERENCES
Uncertainty-Driven Semantic Segmentation through Human-Machine Collaborative Learning
TLDR
This work proposes a method based on conditional Generative Adversarial Network (cGAN), which addresses segmentation in a semi-supervised setup and in a human-in-the-loop fashion, which uses the discriminator to identify unreliable slices for which expert annotation is required and the generator in the GAN to synthesize segmentations on unlabeled data for which the model is confident.
Interactive Medical Image Segmentation Using Deep Learning With Image-Specific Fine Tuning
TLDR
A novel deep learning-based interactive segmentation framework by incorporating CNNs into a bounding box and scribble-based segmentation pipeline and proposing a weighted loss function considering network and interaction-based uncertainty for the fine tuning is proposed.
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.
Semi Supervised Semantic Segmentation Using Generative Adversarial Network
TLDR
A semi-supervised framework is proposed – based on Generative Adversarial Networks (GANs) – which consists of a generator network to provide extra training examples to a multi-class classifier, acting as discriminator in the GAN framework, that assigns sample a label y from the K possible classes or marks it as a fake sample (extra class).
A survey on deep learning in medical image analysis
Segmentation Guided Image-to-Image Translation with Adversarial Networks
TLDR
A novel Segmentation Guided Generative Adversarial Networks (SGGAN) is proposed, which leverages semantic segmentation to further boost the generation performance and provide spatial mapping in image-to-image translation.
Automatic 3D Cardiovascular MR Segmentation with Densely-Connected Volumetric ConvNets
TLDR
A novel densely-connected volumetric convolutional neural network to automatically segment the cardiac and vascular structures from 3D cardiac MR images, referred as DenseVoxNet, and it is shown that this network can achieve better performance than other 3D ConvNets but with fewer parameters.
Total Variation Random Forest: Fully Automatic MRI Segmentation in Congenital Heart Diseases
TLDR
A fully automatic supervised segmentation technique for segmenting the great vessel and blood pool of pediatric cardiac MRIs of children with Congenital Heart Defects is proposed and promising results on the HVSMR16 training dataset are obtained in a leave-one-out cross-validation.
Automated Cardiovascular Segmentation in Patients with Congenital Heart Disease from 3D CMR Scans: Combining Multi-atlases and Level-Sets
TLDR
An automatic method that enables segmentation of the whole heart and the great vessels from 3D MRI scans is presented and achieves an average Dice index of 0.89 for the blood pool and 0.75 for the ventricular myocardial tissue.
Semantic Segmentation using Adversarial Networks
TLDR
An adversarial training approach to train semantic segmentation models that can detect and correct higher-order inconsistencies between ground truth segmentation maps and the ones produced by the segmentation net.
...
1
2
3
4
...