Brain age estimation based on 3D MRI images using 3D convolutional neural network
- PDF / 1,556,565 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 99 Downloads / 187 Views
Brain age estimation based on 3D MRI images using 3D convolutional neural network Nastaran Pardakhti 1 & Hedieh Sajedi 1 Received: 2 March 2019 / Revised: 9 January 2020 / Accepted: 27 May 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Brain Age Estimation (BAE) has become a popular challenge in the field of medical and computer sciences in recent years. In the medical sciences field, the investigation on the brain structure and its relationship with aging is considerable. In the computer sciences field, creating an efficient Machine Learning (ML) model of BAE would lead to have accurate regression models. In this paper, a 3D Convolutional Neural Network (3D-CNN) model is used to train a brain age estimation system. To reach a more accurate system, two other regression methods are also applied on the final feature vector generated by 3DCNN system. The system is applied on the samples of IXI dataset normalized by SPM14. Next, to ensure the model’s generalization, 47 healthy samples of ADNI1 dataset are used. Furthermore, some MRI images achieved from Alzheimer patients are feed to the proposed model and the effects of Alzheimer disease on brain aging are investigated. The best Mean Absolute Error (MAE) on evaluation dataset is about 5 years, with Root Mean Square Error (RMSE) = 13.5. The model generalization by a new healthy dataset was evaluated and the result is with the MAE value of about 6 years. Keywords Brain age estimation . Brain MRI . Chronological age . Deep learning . Image processing . Machine learning
1 Introduction Brain Age Estimation refers to the estimation of human age by brain neuroimaging images especially Magnetic resonance imaging (MRI) images. MRI is a very popular and applicable
* Hedieh Sajedi [email protected] Nastaran Pardakhti [email protected]
1
School of Mathematics, Statistics and Computer Science, College of Science, University of Tehran, Tehran, Iran
Multimedia Tools and Applications
Number of published papers
70
IEEE 60 50
ScienceDirect
40
PubMed
30
Scopus
20 10 0 A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
A = brain & CNN(s) B = brain & CNN(s) & MRI C = brain & Convolutional Neural Network D = brain & Convolutional Neural Network & MRI
Fig. 1 Comparison of the number of published papers about using CNNs in brain imaging tasks found by four search engines
technique of neuroimaging, which gives a rich information about an organ [15]. BAE based on MRI images has been a popular challenging issue in the recent years. Figure 1 illustrates the number of publications in this area until October 2019. Brain aging normally occurs during the lifetime and the aging process follows a specific pattern [24]. Almost all researches have proven that the brain volume generally decreases during the aging process [2]. Therefore, BAE models may be created using aging patterns of the brain. The studies about brain aging also have proven the effects of diseases such as Alzheimer on brain, which make the brain structure older [8]. However, some h
Data Loading...
