Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Effect sizes as a statistical feature-selector-based learning to detect breast cancer

Nicolas Masino, Antonio Quintero-Rincon

TL;DR

In this work, an algorithm and experimental results demonstrate the feasibility of developing a statistical feature-selector-based learning tool capable of reducing the data dimensionality using parametric effect size measures from features extracted from cell nuclei images.

Abstract

Breast cancer detection is still an open research field, despite a tremendous effort devoted to work in this area. Effect size is a statistical concept that measures the strength of the relationship between two variables on a numeric scale. Feature selection is widely used to reduce the dimensionality of data by selecting only a subset of predictor variables to improve a learning model. In this work, an algorithm and experimental results demonstrate the feasibility of developing a statistical feature-selector-based learning tool capable of reducing the data dimensionality using parametric effect size measures from features extracted from cell nuclei images. The SVM classifier with a linear kernel as a learning tool achieved an accuracy of over 90%. These excellent results suggest that the effect size is within the standards of the feature-selector methods

Effect sizes as a statistical feature-selector-based learning to detect breast cancer

TL;DR

In this work, an algorithm and experimental results demonstrate the feasibility of developing a statistical feature-selector-based learning tool capable of reducing the data dimensionality using parametric effect size measures from features extracted from cell nuclei images.

Abstract

Breast cancer detection is still an open research field, despite a tremendous effort devoted to work in this area. Effect size is a statistical concept that measures the strength of the relationship between two variables on a numeric scale. Feature selection is widely used to reduce the dimensionality of data by selecting only a subset of predictor variables to improve a learning model. In this work, an algorithm and experimental results demonstrate the feasibility of developing a statistical feature-selector-based learning tool capable of reducing the data dimensionality using parametric effect size measures from features extracted from cell nuclei images. The SVM classifier with a linear kernel as a learning tool achieved an accuracy of over 90%. These excellent results suggest that the effect size is within the standards of the feature-selector methods

Paper Structure

This paper contains 13 sections, 17 equations, 3 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Methodology
  3. Dataset
  4. Software
  5. Effect size
  6. Cohen's d standardized effect size
  7. Cohen's D
  8. $U$ measures
  9. Feature-selector-based decision rule
  10. Feature-selector-based learning
  11. Confidence intervals (CIs)
  12. Results and Discussion
  13. Conclusions

Figures (3)

  • Figure 1: Linear-SVM illustration
  • Figure 2: t-SNE scatter plot examples of some initial features for malignant cancerous samples (red circles), and benign non-cancerous samples (blue circles). (a) Area worst vs. Concave points worst y (b) Perimeter worst vs. Radius worst (c) Texture worst vs. Concave point mean. y (d) Area see vs. Texture mean.
  • Figure 3: Decision rule threshold results are illustrated by a red line for the features under study. (a) Cohen's d: $0.8$. y (b) Cohen's D: $0.8$. (c) Cohen's $U_1$: $0.5$. (d) Cohen's $U_2$: $0.7$.(e) Cohen's $U_3$: $0.8$.