A ResNet mini architecture for brain age prediction

Xuan Zhang,Si-Yuan Duan,Si-Qi Wang,Yao-Wen Chen,Shi-Xin Lai,Ji-Sheng Zou,Yan Cheng,Ji-Tian Guan,Ren-Hua Wu,Xiao-Lei Zhang

DOI: https://doi.org/10.1038/s41598-024-61915-5

IF: 4.6

2024-05-17

Scientific Reports

Abstract:The brain presents age-related structural and functional changes in the human life, with different extends between subjects and groups. Brain age prediction can be used to evaluate the development and aging of human brain, as well as providing valuable information for neurodevelopment and disease diagnosis. Many contributions have been made for this purpose, resorting to different machine learning methods. To solve this task and reduce memory resource consumption, we develop a mini architecture of only 10 layers by modifying the deep residual neural network (ResNet), named ResNet mini architecture. To support the ResNet mini architecture in brain age prediction, the brain age dataset (OpenNeuro #ds000228) that consists of 155 study participants (three classes) and the Alzheimer MRI preprocessed dataset that consists of 6400 images (four classes) are employed. We compared the performance of the ResNet mini architecture with other popular networks using the two considered datasets. Experimental results show that the proposed architecture exhibits generality and robustness with high accuracy and less parameter number.

multidisciplinary sciences

What problem does this paper attempt to address?

The paper aims to address the problem of how to effectively predict brain age. With age, the human brain undergoes structural and functional changes that are associated with various health problems, especially neurodegenerative diseases. Neuroimaging techniques such as MRI can detect these changes and provide valuable information for assessing brain development and aging, as well as diagnosing neurological disorders. Existing deep learning methods have achieved some success in brain age prediction, but computational complexity and resource consumption remain challenging. To this end, the researchers propose a lightweight deep learning network architecture called ResNet Mini (ResMini) by modifying the deep residual neural network (ResNet) to reduce the number of layers, thus reducing the number of parameters and improving efficiency. The paper uses two datasets: the OpenNeuro brain age dataset with 155 participants and a preprocessed Alzheimer's disease MRI dataset with 6400 images to evaluate the performance of ResMini compared to other popular networks. The experimental results show that ResMini maintains high accuracy while having fewer parameters and shorter execution time, demonstrating good generalization ability and robustness. Compared to other networks such as ResNet18 and BHCnet, ResMini achieves similar or slightly improved accuracy while outperforming them in terms of parameter count and execution time. Additionally, ResMini maintains stable performance when dealing with imbalanced sample distributions. In conclusion, this paper presents a lightweight variant of ResNet aimed at addressing the computational complexity issue in deep learning for brain age prediction, achieving more efficient and accurate predictions.

A ResNet mini architecture for brain age prediction

Brain Age Prediction with 3D ResNet34 Model in Healthy Control, Mild Cognitive Impairment, and Alzheimer's Disease

A deep learning model for brain age prediction using minimally preprocessed T1w images as input

Transfer Learning Models on Brain Age Prediction

Brain Age Prediction Based on Resting-State Functional MRI Using Similarity Metric Convolutional Neural Network.

Predicting Brain Age of Healthy Adults Based on Structural MRI Parcellation Using Convolutional Neural Networks

Connectome-based Predictive Modelling of Ageing, Overall Cognitive Functioning and Memory Performance.

MRI-based brain age prediction model for children under 3 years old using deep residual network

Deep Neural Networks Learn General and Clinically Relevant Representations of the Ageing Brain

Estimating Brain Age with Global and Local Dependencies

Brain age prediction via cross-stratified ensemble learning

Accurate Brain Age Prediction with Lightweight Deep Neural Networks.

Parallel Pathway Convolutional Neural Network with Low-rank Fusion for Brain Age Prediction.

Brain Age Prediction Based on Resting-State Functional Connectivity Patterns Using Convolutional Neural Networks

Predicting age from resting-state scalp EEG signals with deep convolutional neural networks on TD-brain dataset

Brain Age Prediction: Deep Models Need a Hand to Generalize

Brain Age Estimation From MRI Using Cascade Networks with Ranking Loss

Age Prediction Using Resting-State Functional MRI

Predicting Age from White Matter Diffusivity with Residual Learning

Brain age prediction using the graph neural network based on resting-state functional MRI in Alzheimer's disease

OTFPF: Optimal Transport-Based Feature Pyramid Fusion Network for Brain Age Estimation with 3D Overlapped ConvNeXt