Summary: By using artificial neural networks to analyze neuroimaging data, researchers are able to accurately determine biological age.
Source: Max Planck Society
A person’s biological age can be accurately determined from brain images using the latest AI technology, called artificial neural networks.
Until now, however, it was unclear what characteristics these networks used to infer age. Researchers from the Max Planck Institute for Human Cognitive and Brain Sciences have now developed an algorithm that reveals that age estimation goes back to a whole range of brain characteristics, providing general information about health status of somebody.
The algorithm could thus help to detect tumors or Alzheimer’s disease more quickly and make it possible to draw conclusions on the neurological consequences of diseases such as diabetes.
Deep neural networks are an artificial intelligence technology that already enriches our daily lives on several levels: artificial networks, which are modeled on the functioning of real neurons, can understand and translate language, interpret texts and recognize objects and people in pictures. But they can also determine a person’s age based on an MRI of their brain.
It is true that it would be easier to know the age by asking the person. However, determining the age of the machine also gives you an idea of what a healthy brain normally looks like at different stages of life.
If the network estimates that the biological age of the brain based on the scan is higher than it actually is, this may indicate possible disease or injury.
Previous studies, for example, have shown that the brains of people with certain conditions, such as diabetes or severe cognitive impairment, appear to have more years under their belt than they actually do. In other words, the brains were in a biologically worse state than one would assume based on the age of these people.
Although artificial neural networks can accurately determine biological age, until now it was unclear what information from brain images their algorithms used to do so. Scientists in the field of AI research also call this the “black box problem”.
According to this, you insert an image of the brain into the model, the “black box”, let it process it, and in the end, you only get its answer. However, due to the complexity of the networks, it was previously unclear how this response is generated.
An algorithm to interpret AI results
Scientists at the Max Planck Institute for Human and Brain Cognitive Sciences in Leipzig therefore wanted to open the black box: what does the model look at to arrive at its result, brain age? To do this, they worked with the Fraunhofer Institute for Telecommunications in Berlin to develop a new interpretation algorithm that can be used to analyze network age estimates.
“This is the first time that we have applied the interpretation algorithm in a complex regression task,” says Simon M. Hofmann, PhD. candidate at the Max Planck Institute and first author of the underlying study, which has now appeared in the journal NeuroImage.
“We can now determine exactly which regions and features of the brain indicate higher or lower biological age.”
This showed that artificial neural networks use, among other things, white matter to make predictions. Accordingly, they look in particular at the number of small cracks and scars that run through the nerve tissue of the brain. They also analyze the width of the furrows in the cerebral cortex or the size of the cavities, the so-called ventricles.
Previous studies have shown that the older a person is, the larger their sulci and ventricles are on average. What is interesting is that the artificial neural networks arrived at these results by themselves, without having received this information. During their training phase, they only had the brain scans and the person’s actual years of life at their disposal.
“Of course, an increased age estimate can also be interpreted as an error in the model,” said Veronica Witte, head of the research group. “But we were able to show that these deviations are biologically significant.”
For example, researchers have confirmed that people with diabetes have an increased brain age. They were able to show that patients have more white matter lesions.
Future role in medical diagnosis
It is already clear that artificial neural networks will play an increasingly important role in medical diagnosis. Knowing what these algorithms are guided by will therefore become increasingly important: in the future, a brain scan could be automatically analyzed by different networks, each specializing in certain fields: one draws conclusions about Alzheimer’s disease, another on tumours, and yet another on possible mental disorders.
“The doctor then not only receives information that certain diseases may be present. It also sees which areas of the brain underlie diagnoses,” says Hofmann.
The corresponding features are marked directly in the MRI image by the algorithms in each case and can thus be detected more easily by healthcare professionals, who in turn can then draw immediate conclusions about the severity of a disease.
It would also be easier to detect diagnostic errors: if the analysis is based on biologically implausible areas, such as errors that occurred during the creation of the image, these can be immediately detected by the doctor. The research team’s interpretation algorithm can therefore also help improve the accuracy of the artificial neural networks themselves.
In a follow-up study, the researchers now want to explore in more detail why their models also examine features of the brain that have so far played little role in aging research – for example, neural networks also focus on the cerebellum. How aging processes progress there in healthy and sick people remains a mystery to scientists.
About this news about AI and biological age
Author: Press office
Source: Max Planck Society
Contact: Press Office – Max Planck Society
Image: Image is in public domain
Original research: Free access.
“Towards the interpretability of deep learning models for multimodal neuroimaging: finding structural changes in the aging brain” by Simon M. Hofmann et al. NeuroImage
Towards the interpretability of deep learning models for multimodal neuroimaging: finding structural changes in the aging brain
Brain age (BA) estimates based on deep learning are increasingly used as a neuroimaging biomarker for brain health; however, the underlying neural characteristics remained unclear.
We combined sets of convolutional neural networks with Layered Relevance Propagation (LRP) to detect which brain features contribute to BA.
Trained on magnetic resonance imaging (MRI) data from a population-based study (not = 2637, 18–82 years), our models accurately estimated age based on single and multiple modalities, region-restricted images, and whole brain (mean absolute errors of 3.37–3.86 year).
We find that BA estimates capture aging on both small and large scales, revealing gross enlargements of the ventricles and subarachnoid spaces, as well as white matter lesions and atrophies that appear throughout the brain. . The deviation from expected aging reflected cardiovascular risk factors, and accelerated aging was more pronounced in the frontal lobe.
By applying LRP, our study demonstrates how superior deep learning models detect brain aging in healthy and at-risk individuals throughout adulthood.
#Electronic #comparison #MRI #brain #images #accurately #determines #biological #age #Neuroscience #News