Exploring the similarity of medical imaging classification problems

TL;DR

This paper investigates how to quantify and visualize the similarity between medical imaging classification problems using meta-features, enabling better meta-learning strategies.

Contribution

It introduces a method to characterize datasets by classifier performance ranks and visualizes their similarities, achieving 89.3% accuracy in dataset classification.

Findings

Meaningful dataset clusters emerge in 2D similarity space

Meta-feature based similarity can classify datasets by origin with high accuracy

Meta-learning has potential to improve medical image analysis

Abstract

Supervised learning is ubiquitous in medical image analysis. In this paper we consider the problem of meta-learning -- predicting which methods will perform well in an unseen classification problem, given previous experience with other classification problems. We investigate the first step of such an approach: how to quantify the similarity of different classification problems. We characterize datasets sampled from six classification problems by performance ranks of simple classifiers, and define the similarity by the inverse of Euclidean distance in this meta-feature space. We visualize the similarities in a 2D space, where meaningful clusters start to emerge, and show that the proposed representation can be used to classify datasets according to their origin with 89.3\% accuracy. These findings, together with the observations of recent trends in machine learning, suggest that meta-learning could be a valuable tool for the medical imaging community.