Neurological MRI-based biomarkers for treatment navigation in depression
My name is Jesper Pilmeyer (28 years old) and I am a PhD candidate in the Signal Processing Systems group at the Department of Electrical Engineering (Biomedical Diagnostics research group). I finished my bachelor’s and master’s in Medical Engineering at TU/e at the Department of Biomedical Engineering. I graduated with a specialization in medical image analysis.
In 2019, I started my PhD, after which we initiated a clinical study called Neurotrend in collaboration with Philips and the epilepsy centre Kempenhaeghe. Neurotrend is one of the first Eindhoven Engine OpenCall projects.
Predicting the clinical outcome
This study is aimed at predicting the clinical outcome (i.e., the course/development of a disease) of people with depression based on MRI scans. More specifically, we obtain structural, functional (activity) and vascular MRI scans of the brain of subjects with and without depression at the beginning of the study and after a year. During the one-year period, we monitor their depression symptoms and cognitive ability. In this way, we can predict how the depression will develop over time based on the first scans but also evaluate brain changes over a year and correlate this to symptom changes. The clinical study was ethically approved in 2021 and its data acquisition is almost finished at the time of writing.
Preliminary results
From the preliminary results, we can conclude that brain activity patterns and interaction between brain networks is time-varying and that including this neurodynamic nature in a model improves the prediction of depression symptom severity changes over time compared to more standard/static approaches (brain activity/synchronicity over the whole functional MRI scan). Moreover, we demonstrate that a relatively novel MRI acquisition method, called multi-echo multiband imaging, increases the functional MRI signal quality and improves, amongst other things, the temporal resolution. This is beneficial as it allows us to more reliably model network interactions. Another interesting finding was the fact that brain volume and tissue properties of several limbic structures, which are known to be involved in emotion processing, also have predictive value for clinical outcome in depression. A smaller amygdala (associated with fear processing) volume correlated significantly with a higher number of lifetime depressive episodes.
Improving the models and interpreting clinical meaning
In the last period of the PhD, I will focus on improving the models and interpreting the clinical meaning of these results, which will further help in understanding the aberrant brain mechanisms in subjects with depression. We hope to show other researchers the direction in which we think future MRI studies related to psychiatric disorders should head. Taking into account the complex, dynamically interactive brain while implementing the aforementioned MRI acquisitions could lead to more replicative results, especially if carried out in studies with a larger sample size. Even though we will not yet be able to apply these models in the clinic to support (still subjective) clinical decision-making, we are contributing significantly to existing depression-related MRI research. We have demonstrated the potential of state-of-the-art analyses and acquisitions in combination with a multi-modal MRI-based longitudinal study for depression diagnosis/prognosis purposes.