DWI Connectome-based Sex Classification in Young using Supervised Variational Autoencoder

Aditya Kothapalli; Aaryan Gera; Sathvik Reddy; Ninad Aithal; Debanjali Bhattacharya; Neelam Sinha

cs.LGOct 12, 2026

DWI Connectome-based Sex Classification in Young using Supervised Variational Autoencoder

Authors: Aditya Kothapalli, Aaryan Gera, Sathvik Reddy, Ninad Aithal, Debanjali Bhattacharya, Neelam Sinha

Organizations: Dept. of Computer Science and Engineering, Amrita School of Computing, Amrita Vishwa Vidyapeetham, Bengaluru, India · Vision and Artificial Intelligence Lab, Indian Institute of Science (IISc), Bengaluru, India · Dept. of Artificial Intelligence, Amrita School of Artificial Intelligence, Amrita Vishwa Vidyapeetham, Bengaluru, India · Centre for Brain Research, Indian Institute of Science (IISc), Bengaluru, India

Abstract

During adolescence, significant changes in brain development are influenced by hormonal changes that impact both the body and the brain. This study presents sex classification in 787 young adults having age range of 15-25 years using Diffusion-weighted (DWI) structural connectome, focusing on 82 Desikan-Killiany atlas ROIs. A supervised variational autoencoder (SVAE) is used to extract latent space features from DWI connectome data. Although several studies explored differences in brain structure and function between sexes, the main contribution of this study lies in incorporating SVAE to learn a compact, regularized latent representation for the task of sex classification. Initial exploratory data analysis shows that these features can capture the differences in sex-specific latent space patterns. Subsequently, the learned latent space SVAE features are fed to different ML models for sex classification. Among the considered ML classifiers, the multi-layer perceptron yields the highest accuracy of 80% in classifying sex. This demonstrates the effectiveness of SVAE in learning compact, informative representations from DWI connectome for sex classification. The present work highlights the potential of using DWI brain connectomics and deep learning to differentiate sex, paving the way for future research on diverse populations that may include transgender individuals, to better understand sex-related variations in brain structure.

Figures & tables

Fig. 1 : Schematic overview of the proposed approach for sex classification

Fig. 2 : DWI Connectome for one representative subject. (A) The generated streamlines across different brain regions according to DK Atlas using MRTrix. (B) DWI connectome: The values in the DWI connectivity matrix represents the strengths of fibre connectivity across different brain regions. The diagonal and off diagonal values show intra-hemispheric and inter-hemispheric connectivity, respectively.

Fig. 3 : The detailed SVAE architecture used in this study.

Fig. 4 : Illustration of SVAE latent space features for one representative male (blue) and female (red) subject. The plot clearly shows the differences in SVAE latent space pattern between male and female.

Fig. 5 : The heat map of confusion matrix using MLP for sex classification

Fig. 6 : Receiver operating characteristics curves from 10-fold cross-validation using various ML classifiers.

Fig. 7 : Illustration of sex classification using t-SNE plot; demonstrating the distinct patterns of DWI brain structural connectivity between male and female subjects. This highlights the effectiveness of ML classifiers in distinguishing between the SVAE latent space features of males and females subjects, considered in this study.

Model	Precision	Recall	F1-score	Accuracy	AUC
Stacking (SVM,XGB/LR)	0.749 ( $\pm 0.038$ )	0.774 ( $\pm 0.055$ )	0.781 ( $\pm 0.035$ )	0.784 ( $\pm$ 0.035)	0.860 ( $\pm$ 0.032)
RBF SVM	0.747 ( $\pm 0.028$ )	0.769 ( $\pm 0.056$ )	0.779 ( $\pm 0.030$ )	0.781 ( $\pm$ 0.030)	0.861 ( $\pm$ 0.032)
Logistic Regression	0.741 ( $\pm 0.034$ )	0.777 ( $\pm 0.066$ )	0.778 ( $\pm 0.038$ )	0.780 ( $\pm$ 0.038)	0.849 ( $\pm$ 0.029)
Voting (SVM, XGB, MLP)	0.756 ( $\pm 0.043$ )	0.712 ( $\pm 0.058$ )	0.765 ( $\pm 0.040$ )	0.770 ( $\pm$ 0.039)	0.855 ( $\pm$ 0.033)
LDA	0.747 ( $\pm 0.046$ )	0.723 ( $\pm 0.055$ )	0.763 ( $\pm 0.041$ )	0.768 ( $\pm$ 0.041)	0.827 ( $\pm$ 0.025)
Gradient Boosting	0.744 ( $\pm 0.043$ )	0.709 ( $\pm 0.068$ )	0.758 ( $\pm 0.045$ )	0.762 ( $\pm$ 0.043)	0.829 ( $\pm$ 0.046)

TABLE I: Summary of sex classification performance measures using SVAE latent space features. Among all ML classifier MLP outperformes with 80% accuracy.

Classification

Classification maps observations to discrete labels, including familiar, unknown, hierarchical, or multiple categories.

See all 410 papers

Sex-based Network-Specific Differences in Connectomes: A Krakencoder-Based Analysis
Jun 15, 2026Vibhashree S H, Debanjali Bhattacharya, Vamshi Krishna Kancharla +1Functional Connectivity Neuroimaging
Centre for Brain Research, Indian Institute of Science, Bengaluru 560012, India · Dept. of Artificial Intelligence, Amrita School of Artificial Intelligence, Amrita Vishwa Vidyapeetham, Bengaluru 560035, India
Surface-volume self-supervised representation learning of brain MRI for genetic discovery
Oct 1, 2026Tian Xia, Nuo Chen, Zihao Zhu +4Self-Supervised Learning Medical Image Representation Learning
D. Bradley McWilliams School of Biomedical Informatics, The University of Texas Health Science Center at Houston, Houston, US. · Department of Biomedical Informatics and Data Science, Yale School of Medicine, New Haven, US. · Department of Electrical and Computer Engineering, Texas A&M University, College Station, US. +1
Biological Sex Determination in Cadavers Using Deep Learning Algorithms from Computed Tomography Images of Pelvis and Skull
Jun 21, 2026Giovanna Herculano Tormena, Davi Nascimento Araújo, Germano Coimbra Soares de Carvalho +10Medical Image Classification
São Carlos School of Engineering, University of São Paulo, São Carlos 13566-590, Brazil · 212Aristoclides Teixeira Institute of Forensic (IMLAT), Goiânia 74.425-030, Brazil · 1São Carlos School of Engineering, University of São Paulo, São Carlos 13566-590, Brazil

Jun 15, 2026J/K move · Enter open · S save

Sex-based Network-Specific Differences in Connectomes: A Krakencoder-Based Analysis

Vibhashree S H, Debanjali Bhattacharya, Vamshi Krishna Kancharla, Neelam Sinha

Centre for Brain Research, Indian Institute of Science, Bengaluru 560012, India · Dept. of Artificial Intelligence, Amrita School of Artificial Intelligence, Amrita Vishwa Vidyapeetham, Bengaluru 560035, India

Functional Connectivity Neuroimaging Brain Network Analysis

Open paper AskPDF arXiv

This study examines how deficiencies in one brain connectome modality propagate to the other, using the Krakencoder as a simulation framework. Structural and functional connectomes from 702 healthy participants in the Human Connectome Project were analyzed, with the impact of each of the Yeo-7 functional networks assessed separately. Seven scenarios were considered, each involving the removal of a single network while the remaining networks were preserved. The resulting perturbations in cross-modal predictions were quantified using three complementary metrics: KL divergence on eigenvalue spectra, Frobenius norm, and Wasserstein distance. In addition, the persistence of sex-specific information within the predicted connectomes was evaluated. Across all metrics and both prediction directions, the Default Mode Network produced the largest perturbations, whereas the Somatomotor network yielded the smallest. Sex differences in network-level perturbation signatures were subtle, with the best result being an accuracy of 66.09% from connectomes predicted under network-removal conditions. In contrast, connectomes predicted from intact inputs achieved substantially higher sex classification accuracy, reaching up to 84.76%. These findings confirm that full predicted connectomes retain considerably more sex-discriminative information than perturbation-derived signatures alone.

Abstract

Classification

Explore similar work

Sex-based Network-Specific Differences in Connectomes: A Krakencoder-Based Analysis

Surface-volume self-supervised representation learning of brain MRI for genetic discovery

Biological Sex Determination in Cadavers Using Deep Learning Algorithms from Computed Tomography Images of Pelvis and Skull