A Self supervised learning framework for imbalanced medical imaging datasets

Dimensions overall score 5.0

• The AMIMV method extends the MIMV method with a new augmentation strategy to construct asymmetric multi-image, multi-view pairs to address both data scarcity and dataset imbalance in medical image classification.method 95%
  Evidencepartial

  The MIMV method proposed by us in an earlier work is extended with a new augmentation strategy to construct asymmetric multi-image, multi-view (AMIMV) pairs to address both data scarcity and dataset imbalance in medical image classification.

  Implicationpartial

  Directly and explicitly stated in the abstract as the first contribution of the work.

  Verificationpartial

  partial

• The paper carries out a data analysis to evaluate the robustness of AMIMV under varying degrees of class imbalance in medical imaging.method 95%
  Evidencepartial

  We carry out a data analysis to evaluate the robustness of AMIMV under varying degrees of class imbalance in medical imaging.

  Implicationpartial

  Directly and explicitly stated in the abstract as the second contribution of the work.

  Verificationpartial

  partial

• The paper evaluates eight representative SSL methods in 11 medical imaging datasets (MedMNIST) under long-tailed distributions and limited supervision.method 95%
  Evidencepartial

  We evaluate eight representative SSL methods in 11 medical imaging datasets (MedMNIST) under long-tailed distributions and limited supervision.

  Implicationpartial

  Directly and explicitly stated in the abstract as the third contribution of the work.

  Verificationpartial

  partial

• Experimental results on MedMNIST show an improvement of 4.25% on retinaMNIST, 1.88% on tissueMNIST, and 3.1% on DermaMNIST.result 90%
  Evidencepartial

  Our experimental results on the MedMNIST dataset show an improvement of 4.25% on retinaMNIST, 1.88% on tissueMNIST, and 3.1% on DermaMNIST.

  Implicationpartial

  Directly stated in the abstract with specific numeric results, though the baseline for comparison is not explicitly defined in the provided text.

  Verificationpartial

  partial

• The issue of investigating the robustness of SSL to imbalanced data has rarely been addressed in the domain of medical image classification.limitation 85%
  Evidencepartial

  Self supervised learning (SSL) methods have been proposed to deal with the first problem to a certain extent, but the issue of investigating the robustness of SSL to imbalanced data has rarely been addressed in the domain of medical image classification.

  Implicationpartial

  Directly stated in the abstract as a motivation for the work, though it is a claim about the state of the field rather than a finding of the paper itself.

  Verificationpartial

  partial

• Self-supervised learning methods have been proposed to deal with the problem of non-availability of large quantities of labeled training data in medical imaging.technical 80%
  Evidencepartial

  Self supervised learning (SSL) methods have been proposed to deal with the first problem to a certain extent

  Implicationpartial

  Directly stated in the abstract as background context, though it is a general claim about existing methods rather than a novel finding of this paper.

  Verificationpartial

  partial

• Two common problems in medical imaging analysis are non-availability of large labeled datasets and dealing with imbalanced data where rare classes have limited data.technical 95%
  Evidencepartial

  Two problems often plague medical imaging analysis: 1) Non-availability of large quantities of labeled training data, and 2) Dealing with imbalanced data, i.e., abundant data are available for frequent classes, whereas data are highly limited for the rare class.

  Implicationpartial

  Directly and explicitly stated as the opening premise of the abstract.

  Verificationpartial

  partial

Competitive landscape

A self-supervised learning framework that improves medical image classification accuracy on imbalanced and scarce datasets.