BAENet: A Deep Learning Framework with Enhanced Convolutional Neural Network for Brain Age Estimation

Abstract

The assessment of brain age has specific applications like finding whether the aging pattern looks like healthy brain or it is affected by some element. Since brain is very important organ that controls human functionality, it is useful to know its aging patterns. The development of artificial intelligence, such as deep learning, has made it feasible to analyze brain MRI pictures and identify problems in the brain. There are many existing research endeavours to estimate brain age of individuals. The deep learning model like convolutional neural network is widely used in medical image processing. However, from the literature, it was observed that there is need for architecture resilience for improving age estimation process. Towards this end we proposed a deep learning framework that helps in automatic estimation of brain age. We also proposed a novel deep learning model and named it as Brain Age Estimation Network (BAENet). This model is based on CNN. The proposed model Has increased learning capabilities in solving the problem of brain age estimation. The model has more number of layers for progressively learning the brain modality for leveraging performance. The Learning based Brain Age Estimation (LbBAE) algorithm is what we suggested. A benchmark data set known as UK Biobank is used for our empirical study. The outcomes of the experiment showed that the suggested model outperformed ResNet50 with least Mean Absolute Error (MAE) 2.14 and highest test accuracy 99.50%.