The presence of confounding effects (or biases) is one of the most critical challenges in using deep learning to advance discovery in medical imaging studies. Confounders affect the relationship between input data (e.g., brain MRIs) and output variables (e.g., diagnosis). Improper modeling of those relationships often results in spurious and biased associations. Traditional machine learning and statistical models minimize the impact of confounders by, for example, matching data sets, stratifying data, or residualizing imaging measurements. Alternative strategies are needed for state-of-the-art deep learning models that use end-to-end training to automatically extract informative features from large set of images. In this article, we introduce an end-to-end approach for deriving features invariant to confounding factors while accounting for intrinsic correlations between the confounder(s) and prediction outcome. The method does so by exploiting concepts from traditional statistical methods and recent fair machine learning schemes. We evaluate the method on predicting the diagnosis of HIV solely from Magnetic Resonance Images (MRIs), identifying morphological sex differences in adolescence from those of the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA), and determining the bone age from X-ray images of children. The results show that our method can accurately predict while reducing biases associated with confounders. The code is available at https://github.com/qingyuzhao/br-net.

No abstract

While unsupervised variational autoencoders (VAE) have become a powerful tool in neuroimage analysis, their application to supervised learning is under-explored. We aim to close this gap by proposing a unified probabilistic model for learning the latent space of imaging data and performing supervised regression. Based on recent advances in learning disentangled representations, the novel generative process explicitly models the conditional distribution of latent representations with respect to the regression target variable. Performing a variational inference procedure on this model leads to joint regularization between the VAE and a neural-network regressor. In predicting the age of 245 subjects from their structural Magnetic Resonance (MR) images, our model is more accurate than state-of-the-art methods when applied to either region-of-interest (ROI) measurements or raw 3D volume images. More importantly, unlike simple feed-forward neural-networks, disentanglement of age in latent representations allows for intuitive interpretation of the structural developmental patterns of the human brain.

The nigra substantia nigra pars compacta (SNc) and substantia pars reticulata (SNr) form two major basal ganglia components with different functional roles. SNc dopaminergic (DA) neurones are vulnerable to cell death in Parkinson's disease, and NMDA receptor activation is a potential contributing mechanism. We have investigated the sensitivity of whole-cell and synaptic NMDA responses to intracellular ATP and GTP application in the SNc and SNr from rats on postnatal day (P) 7 and P28. Both NMDA current density (pA/pF) and desensitization to prolonged or repeated NMDA application were greater in the SNr than in the SNc. When ATP levels were not supplemented, responses to prolonged NMDA administration desensitized in P7 SNc DA neurones but not at P28. At P28, SNr neurones desensitized more than SNc neurones, with or without added ATP. Responses to brief NMDA applications and synaptic NMDA currents were not sensitive to inclusion of ATP in the pipette solution. To investigate these differences between the SNc and SNr, NR2 subunit-selective antagonists were tested. NMDA currents were inhibited by ifenprodil (10 μm) and UBP141 (4 μm), but not by Zn2+ (100 nm), in both the SNr and SNc, suggesting that SNc and SNr neurones express similar receptor subunits; NR2B and NR2D, but not NR2A. The different NMDA response properties in the SNc and SNr may be caused by differences in receptor modulation and/or trafficking. The vulnerability of SNc DA neurones to cell death is not correlated with NMDA current density or receptor subtypes, but could in part be related to inadequate NMDA receptor desensitization.

No abstract