Tractometry-Driven Subtype Identification in Schizophrenia Using Contrastive Learning
Poster No:
456
Submission Type:
Abstract Submission
Authors:
Xia Wu1, Congying Chu1, Lingzhong Fan1
Institutions:
1Brainnetome Center, Institute of Automation, Chinese Academy of Sciences, Beijing, China
First Author:
Co-Author(s):
Congying Chu
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences
Beijing, China
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences
Beijing, China
Lingzhong Fan
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences
Beijing, China
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences
Beijing, China
Introduction:
Schizophrenia (SZ) is a severe neuropsychiatric disorder marked by substantial heterogeneity in clinical symptoms, cognitive impairments, and neuropathological features (Alnaes, 2019; Okhuijsen-Pfeifer, 2020). This heterogeneity manifests not only in symptom diversity but also in treatment responses and long-term outcomes. The variability in clinical presentations suggests the existence of distinct subtypes with unique neurobiological correlates and clinical trajectories (Jablensky, 2006). As a "disconnection" syndrome, SZ is characterized by disruptions in interregional connectivity, with white matter fiber bundles playing a pivotal role (Pettersson-Yeo, 2011). Tractometry, the quantitative analysis of these bundles, is essential for elucidating the dysconnectivity hypothesis of SZ. Moreover, extracting disease-related features at the individual level is essential for a more refined understanding of the disease's pathophysiology and for developing targeted interventions (Chamberland, 2021). Contrastive learning provides a method to isolate disease-specific features from background noise, enhancing the exploration of heterogeneity and enabling subtype stratification (Abid, 2019; Aglinskas, 2022). This approach is applied to white matter data in this study to identify SZ subtypes and investigate their clinical and structural characteristics, thereby refining our understanding of the disorder's complexity and heterogeneity.
Methods:
The multi-center large-scale dataset used in this study was collected by Brainnetome Center from seven centers, including 608 SZ patients (age: 27.64 ± 7.35 years) and 561 healthy controls (HC; age: 28.51 ± 7.22 years). TractSeg and RadTract were used for fiber tracking and tractometry analysis using the T1w and diffusion magnetic resonance imaging (MRI) data of all participants. A contrastive variational autoencoder (CVAE) was designed to disentangle SZ specific-features from background information (shared-features) like scanner, site, and other factors (Fig 1A). The specific-features were further used for subtype stratification and the optimal number of subtypes was determined based on 26 metrics. Subtypes were compared for the Positive and Negative Syndrome Scale (PANSS) symptom dimensions (Chen, 2019) and white matter differences, with strict Bonferroni correction applied to ensure robust results. Significant findings were visualized to localize white matter sections and corresponding brain regions linked to schizophrenia's heterogeneity.
Results:
Contrastive learning effectively disentangled SZ specific information and HC shared variability. Subtype clustering identified two optimal subtypes (Fig 1B), evenly distributed across centers and genders (Fig 1C). After partitioning the data into two subtypes, background information, including scanner type, site, and gender, remained effectively disentangled from specific features, being incorporated into shared features. Furthermore, we observed that SZ onset was related to age in both subtypes (Fig 1D). PANSS comparisons revealed significant subtype differences in the affective symptom dimension (t = 3.12, p < 0.01; Fig 1E). White matter analyses identified significant differences in association, commissural, and projection fibers, particularly in the superior longitudinal fasciculus, cerebellar peduncles, thalamic and striatal projection fibers (Fig 2A). Most differences were localized in the left frontal-temporal region, with corresponding but smaller differences in the right hemisphere and bilateral cerebellar peduncles (Fig 2C).
Conclusions:
Contrastive learning disentangled disease-specific and background-related information, enabling meaningful subtype stratification. The identified subtypes exhibited significant differences in affective symptoms and white matter structures, and the localized differences in tract sections revealed distinct disconnection patterns. These findings offered insights into schizophrenia's neurobiological heterogeneity and its clinical relevance.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 2
Diffusion MRI Modeling and Analysis
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
White Matter Anatomy, Fiber Pathways and Connectivity
Novel Imaging Acquisition Methods:
Diffusion MRI
Keywords:
Machine Learning
MRI
Psychiatric Disorders
Schizophrenia
Tractography
White Matter
Other - Disconnection, Contrastive Learning, Heterogeneity
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
Aglinskas, A., Hartshorne, J. K., & Anzellotti, S. (2022). Contrastive machine learning reveals the structure of neuroanatomical variation within autism. Science, 376(6597), 1070-1074.
Alnaes, D. (2019). Brain Heterogeneity in Schizophrenia and Its Association With Polygenic Risk. JAMA Psychiatry, 76(7), 739-748.
Chamberland, M. (2021). Detecting microstructural deviations in individuals with deep diffusion MRI tractometry. Nature Computational Science, 1, 598-606.
Chen, J. (2020). Neurobiological Divergence of the Positive and Negative Schizophrenia Subtypes Identified on a New Factor Structure of Psychopathology Using Non-negative Factorization: An International Machine Learning Study. Biological Psychiatry, 87(3), 282-293.
Jablensky, A. (2006). Subtyping schizophrenia: implications for genetic research. Molecular Psychiatry, 11(9), 815-836.
Okhuijsen-Pfeifer, C. (2020). Demographic and clinical features as predictors of clozapine response in patients with schizophrenia spectrum disorders: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews, 111, 246-252.
Pettersson-Yeo, W. (2011). Dysconnectivity in schizophrenia: where are we now? Neuroscience & Biobehavioral Reviews, 35(5), 1110-1124.