Structural Brain Abnormalities in Schizophrenia Based on Contrast Learning
Poster No:
457
Submission Type:
Abstract Submission
Authors:
Lin Du1,2, Yuqing Sun1,2, Chaoyue Ding3, Xiaohan Tian1,2, biying peng1,2, Wenkun Lei1,2, Junxing Xian1,2, Jing Lou1,2, Yingjie Peng4, yuxuan hong1,2, Ruoxin Yang1,2, Meng Wang1,2, Xinghui Zhao1,2, Xinyi Dong1,2, Bing Liu1,2
Institutions:
1State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university, Beijing, China, 2IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, China, 3Brainnetome Center and National Laboratory of Pattern Recognition, Institute of Automation, Chinese, Beijing, China, 4Institute of Biophysics, Chinese Academy of Sciences, Beijing, China
First Author:
Lin Du
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Co-Author(s):
Yuqing Sun
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Chaoyue Ding
Brainnetome Center and National Laboratory of Pattern Recognition, Institute of Automation, Chinese
Beijing, China
Brainnetome Center and National Laboratory of Pattern Recognition, Institute of Automation, Chinese
Beijing, China
Xiaohan Tian
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
biying peng
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Wenkun Lei
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Junxing Xian
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Jing Lou
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
yuxuan hong
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Ruoxin Yang
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Meng Wang
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Xinghui Zhao
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Xinyi Dong
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Bing Liu
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing normal university|IDG/McGovern Institute for Brain Research, Beijing Normal University
Beijing, China|Beijing, China
Introduction:
Schizophrenia is a severe mental disorder affecting about 1% of the global population, impairing cognition, emotion, and behavior, and complicating diagnosis and treatment. Despite extensive research, its complex causes-spanning neural circuitry, genetics, and environmental factors-are still not well understood. Neuroimaging has revealed brain abnormalities, but high variability makes identifying schizophrenia-specific features difficult. Recent advances in deep learning provide new opportunities to tackle this challenge.
To improve understanding of schizophrenia-specific neuroanatomical alterations, we developed DECODE-SZ (Dual Encoder Contrastive Decoding for Schizophrenia), which integrates contrastive learning, 3D convolutional neural networks, and variational autoencoders. The model isolates schizophrenia-specific features by comparing brain scans of healthy controls and patients, with reduced computational demands and improved performance over prior models. Importantly, validation across 8 independent sites was conducted through cross-site testing: the model was trained on data from all other sites and validated on the target site, ensuring each site's evaluation served as a rigorous test of cross-site generalizability. This robust approach highlights DECODE-SZ's capacity for consistent performance, offering a promising tool for advancing diagnostics and innovating neuroimaging research.
To improve understanding of schizophrenia-specific neuroanatomical alterations, we developed DECODE-SZ (Dual Encoder Contrastive Decoding for Schizophrenia), which integrates contrastive learning, 3D convolutional neural networks, and variational autoencoders. The model isolates schizophrenia-specific features by comparing brain scans of healthy controls and patients, with reduced computational demands and improved performance over prior models. Importantly, validation across 8 independent sites was conducted through cross-site testing: the model was trained on data from all other sites and validated on the target site, ensuring each site's evaluation served as a rigorous test of cross-site generalizability. This robust approach highlights DECODE-SZ's capacity for consistent performance, offering a promising tool for advancing diagnostics and innovating neuroimaging research.
Methods:
In this study, we proposed a novel DECODE-SZ (Dual Encoder Contrastive Decoding for Schizophrenia) model, which combines contrastive learning, 3D convolutional neural network (3D CNN) architecture (see Fig.1), and variational autoencoder (VAE) structures to achieve exceptional generalizability while preserving individual specificity. The 2022 model by Aglinskas et al. was the first to introduce contrastive learning models to the study of disease-specific brain variation, achieving promising results in autism research. Our DECODE-SZ model further optimizes this approach and, for the first time, applies it in schizophrenia, validated across 8 sites. Notably, the DECODE-SZ model reduces total parameters by 73.44% compared to the Aglinskas model, while maintaining or improving performance. It sustains high feature extraction capability, significantly lowering computational demands and reducing the risk of overfitting.
Results:
In multi-site tests across 8 independent sites, the DECODE-SZ model demonstrated exceptional generalizability. Each site's validation involved training the model on data from other sites and testing it on the target site, ensuring rigorous cross-site generalization. It effectively distinguished schizophrenia-specific variations from common variations, with the schizophrenia-specific encoding vector (S) showing significantly higher correlations with the disease-related phenotype (PANSS_Total) compared to common variations (C) (|τ|>0.025, p<0.01). Further analysis revealed consistent schizophrenia-specific neuroanatomical variations across 8 sites (see Fig.2), highlighting the model's robustness and accuracy.
Conclusions:
To our knowledge, this is the first study to explore schizophrenia-specific neuroanatomical variations using a contrastive learning strategy combined with a dual-encoder architecture within a VAE framework. We developed the DECODE-SZ model to separate schizophrenia-specific neuroanatomical variations from common variations. Validation was conducted on 8 entirely independent sMRI datasets, yielding consistent results across all sites. In each case, the model successfully extracted schizophrenia-specific neuroanatomical features that showed stronger associations with clinical symptom scores (PANSS), while also identifying common structural variations shared by schizophrenia patients and healthy controls that were more closely related to demographic variables (age and sex). Finally, we explored schizophrenia-specific consistent variation regions across the 8 sites and computed the consistency variation heatmaps.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Modeling and Analysis Methods:
Segmentation and Parcellation
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Anatomy and Functional Systems 2
Keywords:
Machine Learning
Schizophrenia
STRUCTURAL MRI
Other - Contrastive learning
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.
Zhang, J., Rao, V. M., Tian, Y., Yang, Y., Acosta, N., Wan, Z., ... & Guo, J. (2023). Detecting schizophrenia with 3D structural brain MRI using deep learning. Scientific Reports, 13(1), 14433.
Aglinskas, A., Hartshorne, J. K., & Anzellotti, S. (2022). Contrastive machine learning reveals the structure of neuroanatomical variation within autism. Science, 376(6597), 1070-1074.
Moreno-De-Luca, D., & Martin, C. L. (2021). All for one and one for all: heterogeneity of genetic etiologies in neurodevelopmental psychiatric disorders. Current opinion in genetics & development, 68, 71-78.
Mohsenvand, M. N., Izadi, M. R., & Maes, P. (2020, November). Contrastive representation learning for electroencephalogram classification. In Machine Learning for Health (pp. 238-253). PMLR.
Chen, T., Kornblith, S., Norouzi, M., & Hinton, G. (2020, November). A simple framework for contrastive learning of visual representations. In International conference on machine learning (pp. 1597-1607). PMLR.