Learning multiple non-mutually-exclusive tasks for improved classification of inherently ordered labels
This addresses incremental improvements in computer-aided diagnosis tools for medical imaging by enhancing classification accuracy for ordered risk labels.
The paper tackles the problem of medical image classification with ordered malignancy risk labels by rephrasing it into multiple tasks corresponding to different thresholds, enabling the use of Multiple Task Learning (MTL) to significantly improve classifier performance through better data extraction.
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.