Grant: Master's Award: Canada Graduate Scholarships
Université de Montréal
$27,000.00
Modeling how psychotic symptoms distort visual information processing in the human brain
FY 2024-25
Summary
Despite years of research, the neural mechanisms supporting psychotic symptoms like hallucinations and delusions remain poorly understood. Neuroimaging studies have had limited success in revealing the neurobiological basis of these symptoms due to their heterogeneous manifestations. Combining intensive neuroimaging with encoding models based on deep learning (DL) might offer a promising avenue for training within-subject models that adequately capture distortions in perception of reality.This objective can be accomplished utilizing data from the PRISME database. PRISME includes 22 participants with a common history of psychosis who underwent 10-12 fMRI sessions each over 12 months while viewing objects from the THINGS image database.This master project examines how delusion severity impacts DL models' ability to predict neural activity in individuals with Schizophrenia spectrum disorders. DL algorithms generate semantic representations of visual stimuli based on objective image descriptions, independent of participant interpretations. Delusions may disrupt the alignment between these representations and brain activity, potentially indicating altered semantic processing in the brain. We hypothesize that delusion severity will correlate negatively with predictive accuracy in brain regions implicated in delusional symptoms. These findings could inform the development of more refined DL models that account for the distinct neural processes associated with delusions.Preliminary analyses show an inverse relation between predicted neural activity and delusion severity in regions such as the superior parietal lobule, precuneus, and central sulcus. This suggests that distortion of activity in these regions may partly explain how delusions alter reports to reality.
Department
Program
Master's Award: Canada Graduate Scholarships
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database
As published
Id
529988
Name
Ujevco Arina
Region
Québec
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Modeling how psychotic symptoms distort visual information processing in the human brain
Programtype2
Studentship Programs
Orgnameinp2
Université de Montréal
Cihramount2
$27,000
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Artificial Intelligence; Contrastive Language-Image Pre-Training (Clip); Deep Learning; Encoding Models; Functional Magnetic Resonance Imaging; Neuroimaging; Psychotic Symptoms; Schizophrenia Spectrum Disorders; Semantic Processing; Subjective Experience
Abstract
Despite years of research, the neural mechanisms supporting psychotic symptoms like hallucinations and delusions remain poorly understood. Neuroimaging studies have had limited success in revealing the neurobiological basis of these symptoms due to their heterogeneous manifestations. Combining intensive neuroimaging with encoding models based on deep learning (DL) might offer a promising avenue for training within-subject models that adequately capture distortions in perception of reality.This objective can be accomplished utilizing data from the PRISME database. PRISME includes 22 participants with a common history of psychosis who underwent 10-12 fMRI sessions each over 12 months while viewing objects from the THINGS image database.This master project examines how delusion severity impacts DL models' ability to predict neural activity in individuals with Schizophrenia spectrum disorders. DL algorithms generate semantic representations of visual stimuli based on objective image descriptions, independent of participant interpretations. Delusions may disrupt the alignment between these representations and brain activity, potentially indicating altered semantic processing in the brain. We hypothesize that delusion severity will correlate negatively with predictive accuracy in brain regions implicated in delusional symptoms. These findings could inform the development of more refined DL models that account for the distinct neural processes associated with delusions.Preliminary analyses show an inverse relation between predicted neural activity and delusion severity in regions such as the superior parietal lobule, precuneus, and central sulcus. This suggests that distortion of activity in these regions may partly explain how delusions alter reports to reality.
Competitiondate
202412
Approvedterm2
1 yr 0 mth
Pinamesdelim
Ujevco, Arina
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $27,000~ $0~