Researchers at Johns Hopkins University have developed a technology called artificial intelligence (AI) training to collect images of mouse brain cells in real time. AI tools, combined with advanced ultra-small microscopes, make it possible to pinpoint exactly where and when cells are activated during movement, learning, and memory. This information may someday help scientists understand how the brain functions and is affected by disease.
On March 22, researchers performed mice experiments in Nature Communications.
When a mouse''s head is withdrawn for imaging, its neurological function might not be true, according toXingde Li, a PhD in biomedical engineering at the Johns Hopkins University School of Medicine. Whenever we see exactly what is happening among individual brain cells and their connections, we must see precisely what happens in the animal, while the animal is freely moving around, eating, and socializing.
To obtain this enormously detailed data, a Lis team developed ultra-small microscopes that mice can wear on the top of their head. These microscopes are measured in a couple of millimeter in diameter, which make them susceptible to motion disturbances. These microscopes, according to scientists, would require more than 20 frames per second to prevent motion disturbances.
Li says there are two ways to increase frame rate. You may increase the scanning speed, and you may lower the number of points scanned.
Lis engineering team quickly discovered they had reached six frames per second, which maintained excellent image quality, but was far below the required rate. So, the team decided to merge through the second technique to reduce the number of points scanned. This technique would however cause the microscope to capture less-resolution data.
When it is impossible or tedious to create a computer program for a task, such as reliably recognizing a cluster of features as a human face, computer scientists prefer to be able to program themselves by processing large quantities of data.
A significant obstacle in the proposed AI approach was the absence of similar images of mouse brains to train the AI against. This experiment then began training the AI to identify brain blocks from images of fixed samples of mouse brain tissue. This step then enlisted the AI to identify brain cells with natural structural variation and a little bit of motion created by the movements of the mouse''s breathing and heartbeat.
According to Li, whenever we collect data from a moving mouse, we expect it to be similar enough for the AI network to recognize.
The researchers compared microscope scanning points with factors of 2, 4, 8, 16 and 32, and determined how accurately the AI would enhance the image and restore the image resolution.
The researchers claim that the AI could improve image quality by up to 26 frames per second.
The researchers analyzed how well the AI tool performed in tandem with a mini microscope attached to the head of a moving mouse. Using the combination AI and microscope, they were able to consciously observe activity spikes of individual brain cells activated by the mouse walking, rotating, and generally exploring its environment.
According to Li, we would never have seen this information at a high resolution and frame rate before. This would make it possible to obtain more information on how the brain is dynamically connected to action on a cell level.
According to the study, an AI program may be able to accurately interpret images up to 52 or even 104 frames per second if it improves.