Continual Active Learning for Efficient Adaptation of Machine Learning Models to Changing Image Acquisition
This addresses the issue of model reliability in clinical imaging environments where acquisition changes are common, though it is incremental as it builds on existing continual and active learning techniques.
The paper tackles the problem of deep learning models losing accuracy due to changes in medical imaging hardware and protocols by proposing a continual active learning method that adapts to new data streams and selects optimal examples for labeling. Results show it outperforms naive active learning on brain age estimation with T1-weighted MRI from three scanners, requiring less manual labeling.
Imaging in clinical routine is subject to changing scanner protocols, hardware, or policies in a typically heterogeneous set of acquisition hardware. Accuracy and reliability of deep learning models suffer from those changes as data and targets become inconsistent with their initial static training set. Continual learning can adapt to a continuous data stream of a changing imaging environment. Here, we propose a method for continual active learning on a data stream of medical images. It recognizes shifts or additions of new imaging sources - domains -, adapts training accordingly, and selects optimal examples for labelling. Model training has to cope with a limited labelling budget, resembling typical real world scenarios. We demonstrate our method on T1-weighted magnetic resonance images from three different scanners with the task of brain age estimation. Results demonstrate that the proposed method outperforms naive active learning while requiring less manual labelling.