Your browser does not support JavaScript!



[2019/11/25] NYMU Uses AI to Develop Technology that Estimates the Age of a Brain

National Yang-Ming University Institute of Neuroscience Professor Lin


How to know the true age of a brain and the health status of this organ has always been a difficult challenge for the medical community. The team of NYMU Aging and Health Research Center has implemented the “AI Innovation” project supported by the Ministry of Science and Technology in cooperation with NVIDIA Taiwan with the aim of developing a brain age estimation model. This approach uses brain magnetic resonance imaging (MRI) technology and an artificial intelligence algorithm. Recently they are have been able to correctly estimated the brain ages of a number of subjects and their achievement stood out at the International Brain Age Analysis Competition and as a result won the first prize for Asia.


The human brain gradually deteriorates with age, and these changes are often accompanied by decreased cognitive functioning, behavioral changes, and various clinical pathologies. Traditionally a physician is only able to use their own experience to interpret brain neuroimaging. This is limited by what the human eye can see, and any hidden cellular changes within the image are often extremely difficult to detect. Professor Lin of NYMU Institute of Neuroscience, director Chen of NYMU Aging and Health Research Center, associate researcher Zhou from NYMU Brain Research Center, and Professor Shi’s team from NVIDIA Taiwan’s “AI Technology Center”, have used an artificial intelligence algorithm in combination with human brain imaging to develop an integrated deep learning convolutional neural network brain estimation model. Their model is able to precisely estimate an individual's brain age.




Professor Lin said that the fact that many populations are becoming aged is a major challenge to modern society. In the past, the individual aging process always has used an individual's chronological age (in Taiwan their ID card age) as a reference, but the genetic diversity of individuals, their various life experiences and the differential effects of environmental factors vary greatly and therefore all of these factors can have very different effect on the accumulating age of the brain. This makes a simple assessment of the health of an individual’s brain compared to their chronological age very difficult. The calculation of a brain's physiological age (brain age) does need to be estimated because it is able to provide a basis for a range of health indicators, and also is a means of predicting the probability of lesions affecting the brain.


Normally brain age is considered to be the same as the chronological age. However, if the brain age is greater than the chronological age, this means that the brain has undergone an accelerated aging process. Such overaged brains are prone to range of neurological and psychiatric risks and this may even be associated with an increase in mortality. Studies have shown that, for older individuals over the age of 72, the risk of death increases by 6.1% for each additional year of brain age compared with chronological age.




The “Integrated Deep Learning Convolutional Neural Network Brain Age Estimation Model”, which was developed by National Yang Ming University Institute of Neuroscience and the Huida NVIDIA team, allows a computer to accurately estimate brain age in a time. This model helped the cooperating research team to the win fourth place in the world and first place in Asia at this year’s Global Predictive Analytics Competition (PAC). Professor Lin hopes that this technology will not only be used for clinical diagnosis and as a predictive indicator for treatment , but he also believes that it will be applied n Taiwan and elsewhere in other areas, such as personalized health care, in health risk assessment and for efficacy evaluations when investigating new treatments. All of the above areas are very important to Taiwan, which currently in the process of becoming an advanced aging society.



 Professor Lin and his team (left) and NVIDIA TaiwanTechnology Center team of Professor Shi (right)