Nonparametric Density Flows for MRI Intensity Normalisation
This work addresses the challenge of aggregating multi-centre medical imaging data for machine learning applications, which is crucial for improving analysis accuracy in healthcare, though it is incremental as it builds on existing intensity normalization techniques.
The paper tackles the problem of inconsistent tissue intensities in multi-site MRI data by introducing a novel intensity normalization scheme based on density matching, which uses Dirichlet process Gaussian mixtures and mass-conserving flow to transform histograms, resulting in substantially more compatible tissue intensity statistics compared to a baseline affine transformation and smoother transformations than state-of-the-art methods.
With the adoption of powerful machine learning methods in medical image analysis, it is becoming increasingly desirable to aggregate data that is acquired across multiple sites. However, the underlying assumption of many analysis techniques that corresponding tissues have consistent intensities in all images is often violated in multi-centre databases. We introduce a novel intensity normalisation scheme based on density matching, wherein the histograms are modelled as Dirichlet process Gaussian mixtures. The source mixture model is transformed to minimise its $L^2$ divergence towards a target model, then the voxel intensities are transported through a mass-conserving flow to maintain agreement with the moving density. In a multi-centre study with brain MRI data, we show that the proposed technique produces excellent correspondence between the matched densities and histograms. We further demonstrate that our method makes tissue intensity statistics substantially more compatible between images than a baseline affine transformation and is comparable to state-of-the-art while providing considerably smoother transformations. Finally, we validate that nonlinear intensity normalisation is a step toward effective imaging data harmonisation.