Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-Task Adversarial Variational Autoencoder for Estimating Biological Brain Age with Multimodal Neuroimaging

Muhammad Usman, Azka Rehman, Abdullah Shahid, Abd Ur Rehman, Sung-Min Gho, Aleum Lee, Tariq M. Khan, Imran Razzak

TL;DR

The Multitask Adversarial Variational Autoencoder, a custom deep learning framework designed to improve brain age predictions through multimodal MRI data integration, and positions M-AVAE as a powerful tool for metaverse-based healthcare applications in brain age estimation.

Abstract

Despite advances in deep learning for estimating brain age from structural MRI data, incorporating functional MRI data is challenging due to its complex structure and the noisy nature of functional connectivity measurements. To address this, we present the Multitask Adversarial Variational Autoencoder, a custom deep learning framework designed to improve brain age predictions through multimodal MRI data integration. This model separates latent variables into generic and unique codes, isolating shared and modality-specific features. By integrating multitask learning with sex classification as an additional task, the model captures sex-specific aging patterns. Evaluated on the OpenBHB dataset, a large multisite brain MRI collection, the model achieves a mean absolute error of 2.77 years, outperforming traditional methods. This success positions M-AVAE as a powerful tool for metaverse-based healthcare applications in brain age estimation.

Multi-Task Adversarial Variational Autoencoder for Estimating Biological Brain Age with Multimodal Neuroimaging

TL;DR

The Multitask Adversarial Variational Autoencoder, a custom deep learning framework designed to improve brain age predictions through multimodal MRI data integration, and positions M-AVAE as a powerful tool for metaverse-based healthcare applications in brain age estimation.

Abstract

Despite advances in deep learning for estimating brain age from structural MRI data, incorporating functional MRI data is challenging due to its complex structure and the noisy nature of functional connectivity measurements. To address this, we present the Multitask Adversarial Variational Autoencoder, a custom deep learning framework designed to improve brain age predictions through multimodal MRI data integration. This model separates latent variables into generic and unique codes, isolating shared and modality-specific features. By integrating multitask learning with sex classification as an additional task, the model captures sex-specific aging patterns. Evaluated on the OpenBHB dataset, a large multisite brain MRI collection, the model achieves a mean absolute error of 2.77 years, outperforming traditional methods. This success positions M-AVAE as a powerful tool for metaverse-based healthcare applications in brain age estimation.

Paper Structure

This paper contains 16 sections, 16 equations, 5 figures, 4 tables.

Table of Contents

  1. Encoder Branches and Latent Variables
  2. Cross Encoder Reconstruction
  3. Age and Sex Prediction Strategy
  4. Objective Function
  5. Adversarial Loss
  6. Variational Loss
  7. Generic-Unique Distance Ratio Loss
  8. Regression Loss
  9. Classification Loss
  10. Reconstruction Loss
  11. Full Objective
  12. Datasets
  13. Model Architecture and Training Strategy
  14. Comparison With the State-of-the-Art Methods
  15. Robustness Analysis
  16. ...and 1 more sections

Figures (5)

  • Figure 1: Architecture of our proposed Multitask Adversarial Variational Autoencoder (M-AVAE) for predicting brain age and sex estimation from multimodal MRI data which includes sMRI and fMRI scans.
  • Figure 2: Age and gender distribution across the training and validation sets of the OpenBHB dataset dufumier2022openbhb.
  • Figure 3: Details of the various components of our proposed M-AVAE architecture. Subfigures (a), (b), (c), and (d) represent the architectural details of the encoder, decoder, age regressor, and sex classifier networks, respectively. Note that the discriminator shares the same architecture as the sex classifier.
  • Figure 4: Scatter plots illustrating the predicted age versus chronological age for nine different machine learning models. Subfigures (a), (b), (c), (d), (e), (f), (g), (h), and (i) represent the results from Random Forest (RF), Support Vector Regression (SVR), Gaussian Process Regression (GPR), Partial Least Squares Regression (PLSR), Multiple Kernel Learning (MKL) wilson2019multiple, Incomplete Multi-Source Fusion (iMSF) yuan2012multi, Adversarial Autoencoder (AAE), Multitask Adversarial Autoencoder (M-AAE), and the proposed Multitask Adversarial Variational Autoencoder (M-AVAE), respectively. The yellow line indicates the mean value, while the red lines represent the confidence intervals.
  • Figure 5: The bar chart compares the mean absolute error (MAE) in years of various multimodal and unimodal regression methods for brain age prediction.