Theses
The Translational Psychiatry team supervises research theses on various levels ranging from the Bachelor’s to the PhD level. The theses in our lab are always co-supervised by prof. Stefan Borgwardt or prof. Christina Andreou and one of our postdoctoral researchers.
While many of our research projects are of interdisciplinary nature, this nature means that they consist of multiple different aspects and can be broken down into researcherable chunks for thesis projects of all sizes and complexity.
We outline the available thesis projects and respective supervisors below. Please read the description carefully and get in touch with us, if interested. You are invited to use common administrative routes, such as the official allocation process for Psychology students at the University of Lübeck or send us an unsolicited application including your favourite project and a brief CV. A detailed scope of your project and specific tasks will be determined together with your supervisors. Theses can be written in English or German.
We are looking forward to working with you on your next research project.
Video analysis
Facial cues and expressions constitute a component of bodily responses that provide useful information about one’s mental health state. They can be modelled consistently in terms of fundamental facial muscle movements, called facial Action Units (AU) according to the Facial Action Coding System (FACS). Types of Action Units, such as blinks and eyes closed, potentially support the evaluation of one’s mental health state, e.g., evaluation of the treatment response.
We recorded the facial expressions of patients with psychosis during videogame-based training which have shown promising results in the context of treating psychosis.
The scope of the thesis is the evaluation of the facial expressions during the treatment and the association with the clinical symptoms and severity.
Skills: Experience with clinical interviews, willingness to learn Python.
Please note this is research project in a collaboration with prof. Heinz Handels (Institute for medical informatics).
Supervising researchers: Alexandra Korda & Stefan Borgwardt
Modality-specific sensory sweet spots for hallucination-like percepts in psychosis
Hallucinations and hallucinatory experiences occur in the general population as well as in patients with psychosis. In simple behavioural tasks, a perceptual decision on whether a sound contains relevant content (e.g., a human voice) coupled with a subsequent confidence rating can be used to assess the frequency and stimulus properties of operationalised hallucinations. Building on a preceding project, which investigated perceptual hallucination thresholds, we port this paradigm to a special room for surround sound presentation. More specifically, this project asks whether auditory hallucination thresholds vary in space (i.e., depending on where the sound is played from around the participants head)? A laboratory study at McGill University will be used to gather behavioural reports of young adults and then analysed with linear mixed-effects models either in R or Jamovi. Some code is available. This project expects the thesis to be written in English.
Supervising researchers: Léon Franzen, Christina Andreou & Stefan Borgwardt
Speech analysis
Language is the main source of information in psychiatry and represents “big data” at the level of the individual. Language and behavior are amenable to computational natural language processing (NLP) analytics, which may help operationalize the mental status examination. Models that bring together multiple speech features can distinguish speakers with psychiatric disorders from healthy controls with high accuracy.
We recorded and stored the speech of patients with psychosis during structured interviews. The scope of the project is the identification of language risk factors for the prediction of the clinical symptoms and severity in patients with first episode psychosis and clinical high risk state.
Skills: Experience with clinical interviews, willingness to learn Python.
Please note this is research project in a collaboration with the Discourse for Psychosis Consortium.
Supervising researchers: Alexandra Korda & Stefan Borgwardt
Personality dimensions of perceptual decision biases and metacognitive confidence
This research project aims to investigate potential links between personality dimensions of the Big 5 personality inventory and perceptual decisions characteristics (i.e., perceptual sensitivity vs strategic bias) and metacognitive confidence ratings (Nonline ~190 and Nlab ~60). Perceptual detection tasks resemble the numerous decisions all of us make every day. This may be detecting a human voice or face in noisy surroundings. While we have learned much about the mechanisms of audio-visual perceptual decisions in recent decades, we do not know whether and how personality dimensions affect strategic decision bias. A series of exploratory research questions explores whether 1) potential links exist and 2) how decision characteristics vary with personality dimensions? One main aspect of interest is the relationship of neuroticism to perceptual decision bias and confidence. As analyses, we expect the use of signal detection theory and linear mixed-effects modelling either in R or Jamovi. One dataset should be chosen. Comparisons of the laboratory and online data are optional but likely require some coding skills.
Supervising researchers: Léon Franzen, Christina Andreou, Jonas Obleser & Stefan Borgwardt
Brain MRI data analysis
Brain MR images are available from the ADNI database, including Cognitively Normal, Mild Cognitive Impairment and Alzheimer’s Disease. Statistical features, e.g., entropy, will be extracted to capture the microscopic alterations in tissue characteristics of the brain. These features named, radiomics texture feature, enable the quantification of the grey levels and brain patterns on MRI, via voxel inter-relations and spectral properties of the images. The scope is to identify neuroanatomical biomarkers able to discriminate between the three groups using qualitative and/or quantitative analysis.
Skills: Willingness to learn MATLAB.
Please note that this is a collaborative research project with prof. Nico Bunzeck (Institute for Psychology)
Supervising researchers: Alexandra Korda, Nico Bunzeck & Stefan Borgwardt
The effect of state vs trait prior expectations on perceptual decisions in psychosis
Recent theoretical advances postulate that perception is the product of sensory stimulation and prior expectations. Prior expectations can take the form of implicit or explicit expectations that an individual brings to a task (priors as traits). But these trait-level expectations may also be modulated throughout the course of an experiment, which would represent prior expectations as (modulated) states. In this thesis project, we ask how long does it take human participants to detect changes in stimulus occurrence probabilities in a challenging auditory or visual perceptual decision task. In a behavioural task conducted online, statistical probabilities of the occurrence of a human voice/face in a noisy stimulus will be modulated block-wise every ~30 trials. The continuum hypothesis of psychosis lends itself as a theoretical model for this exploration along the continuum ranging from healthy to clinical populations. Analyses would include the examination of choice probabilities throughout these blocks of trials and their relation to trait markers of prior expectations obtained prior to testing through validated questionnaires. Computational modelling of these changes in choice probabilities using MATLAB or Python is optional.
Supervising researchers: Léon Franzen, Christina Andreou & Stefan Borgwardt
Routes of metacognitive evidence accumulation
This thesis project asks whether metacognitive confidence differs if queried in combination with or separately to the perceptual decision. Statistically confidence is defined as the probability that a chosen hypothesis/option is correct (Sanders et al., 2016), whereby it is contingent on a perceptual decision in the first place. Here, building on an existing perceptual decision-making task, we are interested in comparing two evidence accumulation routes within the same participants. Both routes can result in a confident decision, however, their characteristics may vary. To this end, this project compares one-step and two-step decision reports (routes) in which response and confidence ratings are given either at once or sequentially. The routes’ effect on response times, evidence accumulation, and metacognitive efficiency is of main interest. Since first data on audio-visual two-step decisions exists, your task would be to complement this dataset with an auditory or visual behavioural online study using an alternating block design with two consecutive blocks per route. Recruiting and testing patients with psychosis via the Centre of Integrative Psychiatry (ZIP) would be optional. A collaboration with the ZIP psychosis track is in place.
Supervising researchers: Léon Franzen, Christina Andreou & Stefan Borgwardt
Sensory sweet spots for hallucination-like percepts in psychosis – in space
Hallucinations and hallucinatory experiences occur in the general population as well as in patients with psychosis. In simple behavioural tasks, a perceptual decision on whether a sound contains relevant content (e.g., a human voice) coupled with a subsequent confidence rating can be used to assess the frequency and stimulus properties of operationalised hallucinations. Building on a preceding project, which investigated perceptual hallucination thresholds, we port this paradigm to a special room for surround sound presentation. More specifically, this project asks whether auditory hallucination thresholds vary in space (i.e., depending on where the sound is played from around the participants head)? A laboratory study at McGill University will be used to gather behavioural reports of young adults and then analysed with linear mixed-effects models either in R or Jamovi. Some code is available. This project expects the thesis to be written in English.
Supervising researchers: Léon Franzen, Christina Andreou & Stefan Borgwardt
Tailoring screening for mental health disorders to specific groups of adolescents
This project provides an opportunity to get to know the current screening procedures for mental health issues, specifically psychosis, used in the field of early detection and intervention within the Baltic Early Treatment Service (BEATS). While this field traditionally relies on commencing screening upon young adult patients being referred to BEATS and showing first symptoms, we aim to shift screening forward in time to the vulnerable period of adolescence. Further, this new screening approach should be tailored to populations who have a substantially increased risk for developing mental health disorders in adolescence. Your thesis would revolve around parts of the conceptual design of this new screening approach and a first evaluation of an initial batch of behavioural and questionnaire data. The focus will be on finding links between these multimodal data sources using correlative and linear modelling approaches. Strong interest in clinical questions would be beneficial.
Supervising researchers: Léon Franzen, Christina Andreou & Stefan Borgwardt
Substance-induced changes in ongoing brain activity at rest
This project aims to investigate whether substance-induced changes in degree centrality (DC; see below) can provide insight into DC alterations observed in patients with psychosis. Specifically, we will examine fMRI resting-state data from healthy volunteers after administration of d-amphetamine, LSD, and MDMA, on separate occasions. These substances trigger differential changes in the dopaminergic and serotonergic neurotransmitter systems that affect brain connectivity. Similarly, endogenous changes in these neurotransmitter systems occur in patients with psychosis. The outcome measure of interest is DC, which quantifies the number of “connections” a given region (or node) has with other regions (or the entire brain). The first goal is to determine if the substance-induced changes in DC differ depending on the different neuromodulatory influences of the substances (i.e., d-amphetamine-dopaminergic, MDMA-serotonergic). This can be achieved with voxel-wise ANOVA analysis in SPM, a Matlab-based toolbox for brain imaging data analysis. The second goal is to test whether these pharmacologically-induced changes are clinically relevant for psychosis. Therefore, DC will also be examined in a group of patients with schizophrenia and a group of healthy controls. Subsequent analyses will use artificial intelligence to investigate whether pathophysiological (i.e., psychosis-related) changes in DC can be validated externally to substance-induced changes in the DC of healthy volunteers.
Supervising researchers: Mihai Avram & Stefan Borgwardt