• Corpus ID: 236447359

ENHANCE (ENriching Health data by ANnotations of Crowd and Experts): A case study for skin lesion classification

@article{Raumanns2021ENHANCEH,
  title={ENHANCE (ENriching Health data by ANnotations of Crowd and Experts): A case study for skin lesion classification},
  author={Ralf Raumanns and Gerard Schouten and Max Joosten and Josien P. W. Pluim and V. Cheplygina},
  journal={ArXiv},
  year={2021},
  volume={abs/2107.12734}
}
We present ENHANCE, an open dataset with multiple annotations to complement the existing ISIC and PH2 skin lesion classification datasets. This dataset contains annotations of visual ABC (asymmetry, border, colour) features from non-expert annotation sources: undergraduate students, crowd workers from Amazon MTurk and classic image processing algorithms. In this paper we first analyse the correlations between the annotations and the diagnostic label of the lesion, as well as study the agreement… 

References

SHOWING 1-10 OF 37 REFERENCES
Multi-task Ensembles with Crowdsourced Features Improve Skin Lesion Diagnosis
TLDR
It is shown that multi-task models with individual crowdsourced features have limited effect on the model, but when combined in an ensembles, leads to improved generalisation.
Crowd disagreement of medical images is informative
TLDR
This work predicts whether a skin lesion from the ISIC 2017 dataset is a melanoma or not, based on crowd annotations of visual characteristics of that lesion, using the mean annotations and other distribution moments, illustrating disagreement.
The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions
TLDR
The HAM10000 (“Human Against Machine with 10000 training images”) dataset is released, which consists of 10015 dermatoscopic images which are released as a training set for academic machine learning purposes and are publicly available through the ISIC archive.
Center-Focusing Multi-task CNN with Injected Features for Classification of Glioma Nuclear Images
TLDR
This work proposes three methods that improve the performance of a previously-developed semi-supervised CNN and injects (concatenate) pre-extracted VGG features into an intermediate layer of the authors' Semi-Supervised CNN so that during training, the CNN can learn a set of additional features.
BreastNet: Entropy-Regularized Transferable Multi-task Learning for Classification with Limited Breast Data
TLDR
A framework to automatically separate malignant from benign breast lesions using limited breast ultrasound data is described, and detailed analysis of the choice of regularizing parameter and visual evidence that introduction of confusion leads to increase in feature generalization are provided.
Skin lesion analysis toward melanoma detection: A challenge at the 2017 International symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC)
TLDR
The design, implementation, and results of the latest installment of the dermoscopic image analysis benchmark challenge are described, to support research and development of algorithms for automated diagnosis of melanoma, the most lethal skin cancer.
Risk of Training Diagnostic Algorithms on Data with Demographic Bias
TLDR
A survey of the MICCAI 2018 proceedings is conducted to investigate the common practice in medical image analysis applications and shows that it is possible to learn unbiased features by explicitly using demographic variables in an adversarial training setup, which leads to balanced scores per subgroups.
Integrate Domain Knowledge in Training CNN for Ultrasonography Breast Cancer Diagnosis
TLDR
The result showed that CNN classifier can not only give reasonable performance in predicting breast cancer, but also propose potential lesion regions which can be integrated into the breast ultrasound system in the future.
...
...