Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

One Shot GANs for Long Tail Problem in Skin Lesion Dataset using novel content space assessment metric

Kunal Deo, Deval Mehta, Kshitij Jadhav

TL;DR

The One Shot GANs model was employed to augment the tail class of HAM10000 dataset by generating additional samples to enhance accuracy, and a novel metric tailored to suit One Shot GANs was utilized.

Abstract

Long tail problems frequently arise in the medical field, particularly due to the scarcity of medical data for rare conditions. This scarcity often leads to models overfitting on such limited samples. Consequently, when training models on datasets with heavily skewed classes, where the number of samples varies significantly, a problem emerges. Training on such imbalanced datasets can result in selective detection, where a model accurately identifies images belonging to the majority classes but disregards those from minority classes. This causes the model to lack generalizability, preventing its use on newer data. This poses a significant challenge in developing image detection and diagnosis models for medical image datasets. To address this challenge, the One Shot GANs model was employed to augment the tail class of HAM10000 dataset by generating additional samples. Furthermore, to enhance accuracy, a novel metric tailored to suit One Shot GANs was utilized.

One Shot GANs for Long Tail Problem in Skin Lesion Dataset using novel content space assessment metric

TL;DR

The One Shot GANs model was employed to augment the tail class of HAM10000 dataset by generating additional samples to enhance accuracy, and a novel metric tailored to suit One Shot GANs was utilized.

Abstract

Long tail problems frequently arise in the medical field, particularly due to the scarcity of medical data for rare conditions. This scarcity often leads to models overfitting on such limited samples. Consequently, when training models on datasets with heavily skewed classes, where the number of samples varies significantly, a problem emerges. Training on such imbalanced datasets can result in selective detection, where a model accurately identifies images belonging to the majority classes but disregards those from minority classes. This causes the model to lack generalizability, preventing its use on newer data. This poses a significant challenge in developing image detection and diagnosis models for medical image datasets. To address this challenge, the One Shot GANs model was employed to augment the tail class of HAM10000 dataset by generating additional samples. Furthermore, to enhance accuracy, a novel metric tailored to suit One Shot GANs was utilized.
Paper Structure (35 sections, 13 equations, 6 figures, 9 tables)

This paper contains 35 sections, 13 equations, 6 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Our Contribution
  3. Preliminaries
  4. Subset Selection of Images
  5. Wasserstein GAN (WGANs)
  6. One-Shot GANs
  7. Fréchet Inception Distance
  8. Content-Space assessment
  9. Content assessment metric
  10. Pre-processing for Spatial metric
  11. Channel Selection
  12. Otsu's method
  13. Selection of images
  14. Classifier utilized
  15. Methodology and Results
  16. ...and 20 more sections

Figures (6)

  • Figure 1: Flowchart for the process of the proposed solution.
  • Figure 2: One-Shot GAN. The two-branch discriminator judges the content distribution separately from the scene layout realism and thus enables the generator to produce images with varying content and global layoutsOne_Shot_GANs.
  • Figure 3: Flowchart describing the process Content-space assessment
  • Figure 4: (a) The original image selected for Morphological operation (b) Closing operation is performed (c) Image after Closing and Erosion (d) Image after Closing, Erosion and Interpolation
  • Figure 5: (a) images obtained after morphological operations (b) Red channel analysis of the Tumor (c) Green channel analysis of the Tumor (d) Blue channel analysis of the Tumor.
  • ...and 1 more figures