Learning multiple non-mutually-exclusive tasks for improved classification of inherently ordered labels

TL;DR

This paper introduces a multi-task learning approach for medical image classification that simultaneously addresses multiple thresholds of malignancy risk, enhancing diagnostic accuracy by leveraging shared information across related tasks.

Contribution

It proposes a novel method to reformulate threshold-based classification as multiple non-mutually-exclusive tasks, enabling effective Multi-Task Learning in medical diagnosis.

Findings

Improved classification performance over single-threshold models

Enhanced information extraction from existing data

Better handling of ordered label structures

Abstract

Medical image classification involves thresholding of labels that represent malignancy risk levels. Usually, a task defines a single threshold, and when developing computer-aided diagnosis tools, a single network is trained per such threshold, e.g. as screening out healthy (very low risk) patients to leave possibly sick ones for further analysis (low threshold), or trying to find malignant cases among those marked as non-risk by the radiologist ("second reading", high threshold). We propose a way to rephrase the classification problem in a manner that yields several problems (corresponding to different thresholds) to be solved simultaneously. This allows the use of Multiple Task Learning (MTL) methods, significantly improving the performance of the original classifier, by facilitating effective extraction of information from existing data.