Research Spotlight: Professor Steven Williams, King’s College London
Leveraging behavioral and imaging measures in psychiatry
While many other disease areas regularly require imaging techniques to diagnose conditions and inform treatment plans, psychiatry relies heavily on subjective measures.1,2 Professor Steve Williams recognizes the potential of objective behavioral measures and imaging in psychiatry.3 His current research aims to harness the capabilities of imaging to identify targets for new therapies and determine key behavioral markers in psychiatric conditions, such as major depressive disorder, to support clinicians in their treatment of patients.3
Toward a scalable approach to imaging in psychiatry
One imaging technique that is frequently enlisted in psychiatry research is functional magnetic resonance imaging (fMRI), which quantifies metabolism and blood flow.4,5 Although fMRI is not without its challenges, promising results that could shape how clinicians diagnose and treat patients have emerged from fMRI research.4,5 Despite these findings, there is an absence of widely accepted imaging biomarkers, and some may argue that we will never find an imaging marker that could be considered a “major depressive disorder test,” for example.4,5 Prof. Williams’ approach instead involves identifying subtypes of major depressive disorder through imaging and correlating certain connectivity patterns with specific symptoms while attempting to identify simple behavioral measures to classify and group patients accordingly.4 This could eventually provide a more scalable and realistic approach rather than looking for a catch-all major depressive disorder test.
Find out how Prof. Williams and his team hope to use fMRI and a series of cognitive tasks to inform new therapies for major depressive disorder and schizophrenia.
So, I’m Professor Steve Williams. I’m the Head of Department here at the Institute of Psychiatry, Psychology & Neuroscience, King’s College London.
Text onscreen - Cognitive assessment and neuroimaging
We’ve developed a battery of cognitive tasks that will work across all devices. It taps into all the major cognitive domains, and each of those tasks only lasts a couple of minutes, so we’ve gamified them to make them engaging and enjoyable for everyone to do. So, how are we going to use these tasks is to find which ones are robust across devices. The ones that are flaky on a smartphone but robust on a laptop we might drop, but the ones that are consistently performing the same across devices is where we’ll start. Then we’re bringing those into the laboratory, and we’ll be testing in our patient cohorts and picking out those tasks that are sensitive to the pathology of depression or schizophrenia. If that’s worked well, the subset of maybe four or five tasks that we then know are robust but also sensitive to pathology would be taken inside the scanner, and we learn all about the brain circuits that are impaired in those different conditions. With the tasks that we’ve settled on that are robust to the environment, whether it’s at home or the laboratory, and those that are sensitive to pathology in, for example, depression or schizophrenia, will be taken forward to the scanner, and we’ll use functional MRI to actually look inside the brain at activity during those tasks. The circuits that we then visualize using this technology will inform the next generation of therapies.
Further reading
- Lee J, et al. Personalized diagnosis and treatment for neuroimaging in depressive disorders. J Pers Med 2022;12:1403.
A review analyzing structural and functional MRI studies that incorporate machine learning algorithms. This article focuses on the application of machine learning techniques to major depressive disorder and indicates that despite the current challenges, these techniques could become useful to clinical practice. - Chen J, et al. Neurobiological divergence of the positive and negative schizophrenia subtypes identified on a new factor structure of psychopathology using non-negative factorization: an international machine learning study. Biol Psychiatry 2020;87:282–293.
This article addresses the heterogeneity often seen in schizophrenia. The researchers used machine learning to identify a four-factor structure representing all four symptom domains of schizophrenia as the most reliable and widely applicable representation of psychopathology.
Cite this article as Research Spotlight: Professor Steven Williams, King’s College London. Connecting Psychiatry. Published February 2025. Accessed [month day, year]. [URL]
-
Wilcox CE, et al. Neuroimage Clin 2020;26:102084.
-
Hussain S, et al. Biomed Res Int 2022;2022:5164970.
-
King’s College London. Professor Steve Williams PhD. Available at: https://www.kcl.ac.uk/people/steve-williams-ioppn. Last accessed: June 2024.
-
Pilmeyer J, et al. J Neuroimaging 2022;32:582–595.
-
Zhuo C, et al. Transl Psychiatry 2019;9:335.
SC-US-77229
SC-CRP-16647
[September] 2024
Related content
Research Spotlight: Professor Leanne Williams, Stanford Center for Precision Health and Wellness
Sign up now for new features
The newsletter feature is currently available only to users in the United States
*Required field