Deep learning technology can tell your doctor accurate brain age information in just a few seconds

We all know that human cognitive ability declines with age. Neuroscientists have long known that this decline is related to changes in the anatomy of the brain. Therefore, it is easy to think of the use of magnetic resonance imaging (MRI) of the brain to identify signs of aging, and even to determine "brain age."

In addition, differences between brain age and chronological age may also reveal the occurrence of conditions such as dementia.

However, the entire analysis process is lengthy because of the large amount of pre-processing of MRI data to identify the age characteristics of the brain. These pretreatments include the removal of non-brain tissue such as skulls from MRI images, as well as the classification of white matter, gray matter, and other tissues of the brain, as well as the removal of image artifacts using various data smoothing techniques.

Deep learning technology can tell your doctor accurate brain age information in just a few seconds

All of these data processing can eventually take more than 24 hours, which is extremely detrimental or even a major obstacle to the doctor's clinical diagnosis.

Recently, the Giovanni Montana research team from King's College London used the original image of the MRI scanner to train its deep-learning machine to automatically recognize brain age.

This deep learning technique takes only a few seconds, and even when the patient is examined in the scanner, the doctor can tell the doctor the exact brain age information, which will provide doctors with reliable and useful information for making the correct clinical diagnosis.

This method is a standard deep learning technique. Montana and colleagues used more than 2,000 healthy brain MRI scans between the ages of 18 and 90, and all samples did not have any neurological disease that could affect brain age. So the brain age of these samples is actual Age is consistent.

Deep learning technology can tell your doctor accurate brain age information in just a few seconds

Among them, each scanned image is obtained by standard T1-weighted MRI scan, which is consistent with the type of output of most modern MRI scanners. During the acquisition process, each scan will be labeled with the actual age of the tester.

The researchers used 80% of the total image set to train the convolutional neural network to determine the age of the brain to be tested by brain MRI scan images. In addition, the researchers used 200 images in the remaining image set to verify the "brain age" recognition process.

Finally, the researchers used the remaining 20% ​​of the images to test the deep learning neural network to determine its accuracy in measuring brain age.

At the same time, the team compared its proposed deep learning techniques with traditional brain age determination methods. Traditional "brain age" measurements require extensive image processing to identify white matter and gray matter regions of the brain, followed by statistical analysis called Gaussian process regression.

The result is very interesting.

For the analysis of given pre-processed data, both the deep learning method and the Gaussian process regression method can accurately determine the actual age of the patient, and the error of both is less than 5 years.

However, for a given original MRI data, the deep learning method shows a clear advantage, and the average error of the measurement results is only 4.66 years. In contrast, the standard Gaussian process regression method performs poorly and can only give a rough age with an average error of almost 12 years.

In addition, the analysis of deep learning methods takes only a few seconds compared to traditional standard methods requiring up to 24 hours of data pre-processing. The only data pre-processing required for deep learning methods is to ensure that the orientation of the original images is consistent and the pixel dimensions between the images are consistent.

This rapid measurement is of great significance to doctors.

The Montana team said: "With this deep learning software, when the patient is still in the scanner, he can provide clinicians with predictive data on brain age,"

In addition, the team compared images obtained with different scanners to show that the deep learning technique can be used for scans from different regions of the world.

The researchers also compared the brain ages of twins, indicating how brain age is associated with genetic factors. Interestingly, this correlation declines with age, suggesting that environmental factors become more pronounced over time, suggesting a direction for future research.

Overall, this is an exciting technological breakthrough that is expected to significantly affect clinicians' diagnostic methods. A large body of evidence suggests that conditions such as diabetes and schizophrenia and traumatic brain injury are associated with accelerated brain aging. Therefore, a rapid and accurate method for measuring brain aging characteristics may have an important impact on how clinicians treat these diseases in the future.

The Montana team said: "The brain age prediction represents an accurate, highly reliable and genetically effective phenotype that is expected to be used as a biomarker for brain aging."

7.92mm Pitch Series Connector

Dongguan SOLEPIN Electronics Co., Ltd , https://www.wentae.com