Fully Automated Habenula Segmentation Provides Robust and Reliable Volume Estimation Across Large Magnetic Resonance Imaging Datasets, Suggesting Intriguing Developmental Trajectories in Psychiatric Disease

Jürgen Germann; Flavia Venetucci Gouveia; Raquel C R Martinez; Marcus Vinicius Zanetti; Fábio Luís de Souza Duran; Tiffany M Chaim-Avancini; Mauricio H Serpa; M Mallar Chakravarty; Gabriel A Devenyi

doi:10.1016/j.bpsc.2020.01.004

Fully Automated Habenula Segmentation Provides Robust and Reliable Volume Estimation Across Large Magnetic Resonance Imaging Datasets, Suggesting Intriguing Developmental Trajectories in Psychiatric Disease

Biol Psychiatry Cogn Neurosci Neuroimaging. 2020 Sep;5(9):923-929. doi: 10.1016/j.bpsc.2020.01.004. Epub 2020 Jan 31.

Authors

Jürgen Germann¹, Flavia Venetucci Gouveia², Raquel C R Martinez³, Marcus Vinicius Zanetti⁴, Fábio Luís de Souza Duran⁵, Tiffany M Chaim-Avancini⁵, Mauricio H Serpa⁵, M Mallar Chakravarty⁶, Gabriel A Devenyi⁷

Affiliations

¹ University Health Network, Toronto, Ontario, Canada; Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Montreal, Quebec, Canada. Electronic address: germannj@gmail.com.
² Sunnybrook Research Institute, Toronto, Ontario, Canada.
³ Division of Neuroscience, Hospital Sírio-Libanês, São Paulo, Brazil; Laboratory of Psychopathology and Psychiaric Therapeutics (LIM-23), Institute of Psychiatry, University of São Paulo Medical School, São Paulo, Brazil.
⁴ Division of Neuroscience, Hospital Sírio-Libanês, São Paulo, Brazil; Laboratory of Psychiatric Neuroimaging (LIM-21), Institute of Psychiatry, University of São Paulo Medical School, São Paulo, Brazil.
⁵ Laboratory of Psychiatric Neuroimaging (LIM-21), Institute of Psychiatry, University of São Paulo Medical School, São Paulo, Brazil.
⁶ Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Montreal, Quebec, Canada; Department of Biological and Biomedical Engineering, McGill University, Montreal, Quebec, Canada; Department of Psychiatry, McGill University, Montreal, Quebec, Canada.
⁷ Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Montreal, Quebec, Canada; Department of Psychiatry, McGill University, Montreal, Quebec, Canada.

PMID: 32222276
DOI: 10.1016/j.bpsc.2020.01.004

Abstract

Studies of habenula (Hb) function and structure provided evidence of its involvement in psychiatric disorders, including schizophrenia and bipolar disorder. Previous studies using magnetic resonance imaging (manual/semiautomated segmentation) have reported conflicting results. Aiming to improve Hb segmentation reliability and the study of large datasets, we describe a fully automated protocol that was validated against manual segmentations and applied to 3 datasets (childhood/adolescence and adult bipolar disorder and schizophrenia). It achieved reliable Hb segmentation, providing robust volume estimations across a large age range and varying image acquisition parameters. Applying it to clinically relevant datasets, we found smaller Hb volumes in the adult bipolar disorder dataset and larger volumes in the adult schizophrenia dataset compared with healthy control subjects. There are indications that Hb volume in both groups shows deviating developmental trajectories early in life. This technique sets a precedent for future studies, as it allows for fast and reliable Hb segmentation and will be publicly available.

Keywords: Automatic segmentation; Bipolar disorder; Habenula; MAGeTbrain; Schizophrenia; Volume.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Adolescent
Adult
Child
Habenula* / diagnostic imaging
Habenula* / physiopathology
Humans
Magnetic Resonance Imaging
Mental Disorders* / diagnostic imaging
Mental Disorders* / physiopathology
Reproducibility of Results

Grants and funding

WT_/Wellcome Trust/United Kingdom