Scientists have developed a new, compact imaging device that is set to transform how we study the brain.
The device marks a major step forward in transforming how neuroscientists study the brain. By enabling high-resolution, real-time imaging of brain activity in freely moving mice, the miniaturized microscope allows researchers to observe the relationship between neural activity and behavior with unprecedented detail.
This technological advance is expected to significantly deepen our understanding of brain function, providing critical insights into how perception, cognition, and behavior are intricately linked to underlying neural processes.
Ultimately, these insights could inform the development of new therapeutic strategies for brain disorders, potentially improving treatments and outcomes for human health.
Lensless camera reconstructs 3D objects using tiny lenslets
Building on his previous research, Weijian Yang, professor of electrical and computer engineering, and his team developed DeepInMiniscope, a lensless camera capable of producing three-dimensional images from a single exposure. Instead of using a single bulky lens, the camera employs a thin mask embedded with dozens of tiny lenslets, each capturing a unique perspective of the same object.
Advanced computational algorithms then combine these multiple perspectives to reconstruct a detailed 3D image, offering a compact and efficient approach to high-resolution imaging that could transform how researchers study complex structures.
Previous imaging systems performed well with large objects in low-scattering environments, such as robotic vision for assembling parts, but they struggled to capture the fine details of biological or biomedical samples. In living tissue, light scattering is common, signal contrast is often low, and reconstructing intricate features across a large volume presents a significant computational challenge.
DeepInMiniscope overcomes these limitations with a redesigned mask containing more than 100 miniaturized, high-resolution lenslets, paired with a novel neural network for image reconstruction. This approach allows the system to “see” through tissue, enabling detailed biomedical imaging without surgery or other invasive procedures.
Instantly capturing mouse neuronal activity
DeepInMiniscope uses a neural network that merges model-based iterative optimization with conventional deep learning. The resulting unrolled network consists of multiple stages, each acting like a mini-network that mimics one iteration of optimization.
For Yang’s microscope, the network instantly reconstructs high-resolution details across a large 3D volume, integrating data from all 100 lenslets into a single coherent image. Using this approach, Yang and his team have successfully recorded real-time neuronal activity in mice, capturing intricate brain processes with unprecedented clarity.
“Our algorithm combines interpretability, efficiency, scalability and precision. It requires only a minimal amount of training data, yet it can robustly and accurately process large-scale datasets at high speed,” explained Feng Tian, a postdoctoral researcher in Yang’s lab and first author of the study.
Next, Yang aims to shrink the device to just 2 square centimeters, which is about the size of a mouse’s hat, and make it cordless, enabling real-time imaging of brain activity in freely moving mice and advancing understanding of how the brain drives behavior.