Deep learning-based unlearning of dataset bias for MRI harmonisation and confound removal

TLDR

Large multi‑site MRI datasets increase statistical power, but scanner‑related differences in acquisition protocols and hardware introduce nonbiological variance that can mask biologically relevant signals. The authors propose a deep‑learning training scheme, inspired by domain adaptation, that iteratively learns scanner‑invariant features while preserving performance on the primary task, and can also remove other known confounds. They demonstrate the framework on regression, classification, and segmentation tasks using two network architectures, employing an iterative update approach to balance scanner invariance and task performance. The framework successfully harmonises multi‑site data, adapts to biased datasets and limited labels, removes additional confounds, and is broadly applicable to diverse neuroimaging studies.

Abstract

Increasingly large MRI neuroimaging datasets are becoming available, including many highly multi-site multi-scanner datasets. Combining the data from the different scanners is vital for increased statistical power; however, this leads to an increase in variance due to nonbiological factors such as the differences in acquisition protocols and hardware, which can mask signals of interest. We propose a deep learning based training scheme, inspired by domain adaptation techniques, which uses an iterative update approach to aim to create scanner-invariant features while simultaneously maintaining performance on the main task of interest, thus reducing the influence of scanner on network predictions. We demonstrate the framework for regression, classification and segmentation tasks with two different network architectures. We show that not only can the framework harmonise many-site datasets but it can also adapt to many data scenarios, including biased datasets and limited training labels. Finally, we show that the framework can be extended for the removal of other known confounds in addition to scanner. The overall framework is therefore flexible and should be applicable to a wide range of neuroimaging studies.