Promises and pitfalls of deep neural networks in neuroimaging-based psychiatric research
This work identifies key problems for researchers and clinicians in psychiatry using neuroimaging, but it is incremental as it synthesizes existing knowledge without new empirical results.
The paper reviews the application of deep neural networks in neuroimaging-based psychiatric research, highlighting their potential for improved diagnostics and treatment, while addressing challenges like small, biased datasets and algorithmic issues.
By promising more accurate diagnostics and individual treatment recommendations, deep neural networks and in particular convolutional neural networks have advanced to a powerful tool in medical imaging. Here, we first give an introduction into methodological key concepts and resulting methodological promises including representation and transfer learning, as well as modelling domain-specific priors. After reviewing recent applications within neuroimaging-based psychiatric research, such as the diagnosis of psychiatric diseases, delineation of disease subtypes, normative modeling, and the development of neuroimaging biomarkers, we discuss current challenges. This includes for example the difficulty of training models on small, heterogeneous and biased data sets, the lack of validity of clinical labels, algorithmic bias, and the influence of confounding variables.