Tesi doctoral de Marta Reales Moreno: “Fractal analysis in psychiatric neuroimaging: The frontal lobe as a key region in clinical progression”. Dirigida pel Dr. Stefan J Borgwardt i pel Dr. David M. Smith
Background:
The World Health Organization
(WHO) emphasizes the need to improve mental health care. Clinical models for
treatment and prediction, though sensitive, often show low specificity (47%) –
raising concerns about overdiagnosis and unnecessary interventions. Psychiatric
disorders remain primarily clinical, and although structural brain
abnormalities are assumed to exist, neuroimaging is mainlyusedtoexcludeorganiccausesratherthantoguidetherapeuticdecisions.WithinMRIresearch, fractalmetricshaveshownpromisebyprovidingcomplementaryinformationtoconventionalmeasures, whichsofarhavefailedtoyieldclinicallyapplicablebiomarkers.Accordingly,themainobjectivewas toreviewtheuseoffractalanalysisinpsychiatricneuroimaging,andsubsequentlytoassessitspotential
forpredictingpsychosisriskandunderstandingthestructuralsubstratesunderlyingpsychoticsymptoms and disorder progression.
Methods:First,asystematicreviewidentified39MRIstudiesemployingfractalanalysisinpsychiatry, classified
according to DSM-V categories. Second, Fractal Dimension (FD) was extracted
from structural MRI data of 194 participants – First Episode Psychosis (FEP),
Clinical High Risk (CHR), and Healthy Controls (HC) – to train neural network
classifiers, which were validated on CHR individuals who transitioned to
psychosis. Finally, the brain regions contributing to the classification wereidentified,andpartialcorrelationsandanatomicalanalyseswereusedtoexaminetheassociations
between FD values and symptom severity scores.
Results:Whole-brainFDanalysesmayunderestimatelocalabnormalitiesduetotheinclusionofahigh proportionoftissue.ClassificationperformedbestforFEPvs.HCandFEPvs.CHR-NonTransition– with accuracy, specificity, and sensitivity above 70%.
External validation confirmed generalizability, achieving accuracies of 87.5%
and 93.8%, respectively. Key predictive regions allowed the identification of
sensorimotor and visual clusters associated with both psychosis transition and
motor hyperactivity (q<0.001), indicating that lower structural irregularity
is related to greater clinical severity.
Conclusions:
Frontal lobe alterations
emerge as a consistent feature across psychiatric conditions. In early
psychosis, dysfunction of sensorimotor regions and visual, parietal, and
cerebellar networks underlies initial motor hyperactivity, while mild frontal
and parietal alterations reflect distributed network dysfunction that
consolidates the FEP. Given their functional impact and well-characterized
clinical stages, psychotic disorders provide an ideal framework to extend this
approach to other psychiatric conditions in future research.
Lectura de la tesi: 28/04/2026 (informació extreta de l’Agenda activitats de la web Escola de Doctorat).