Deep Learning Algorithms Help Map Developing Brains

first_img Robot Dog Astro Can Sit, Lie Down, and Save LivesDeepMind’s AlphaStar AI Crushes Pro StarCraft Players Thanks to machine learning, researchers are one step closer to understanding the structure and function of different brain regions.Scientists from the Brain Research Institute of the University of Zurich (UZH) and the Swiss Federal Institute of Technology (ETH) developed a fully automated brain registration method for segmenting gray matter in mice.“My lab aims to reveal how the mammalian brain develops its abilities to process and react to sensory stimuli,” study co-author Theofanis Karayannis told Tech Xplore. “Most of the work we do is on the experimental side, utilizing the mouse as a model system and techniques that range from molecular-genetic to functional and anatomical.”Though initially tested on animals (there’s a reason they’re called “lab rats”), the system—a fully automated convolutional neural network (CNN)-based method called DeNeRD (Detect Neurons in different brain Regions during Development)—could eventually be applied to humans.This study is part of a larger project, including work in which Karayannis and colleagues use deep-learning algorithms to track inhibitory neurons, gauging the development of brain capabilities at specific points in time.“The first step toward this goal is to accurately register the regions of interest in a mouse brain against a standard reference atlas, with minimum human supervision,” according to the research paper. “The second step is to scale this approach to different animal ages, so as to also allow insights into normal and pathological brain development and aging.”Using the computational skills of Ph.D. student Asim Iqbal, the team was able to test some image-registration-based methods gaining traction among neuroscientists.“We quickly realized that existing techniques are suboptimal for cases where the tissue sections are rotated,” Karayannis said. “Or when their geometry is compromised due to methodological issues, for example during brain tissue slicing.”So, they developed their own deep learning method—Segmenting Brain Regions (SeBRe)—to produce reliable results regardless of scale, rotation, and morphological issues, Tech Xplore reported.The group trained their neural network on brain sections of 14-day-old mice, then tested its ability to generate anatomical masks of previously unidentified sections of the brain in four-, 14-, 28-, and 56-day-old mice.“SeBRe outperformed all existing brain registration methods,” according to Tech Xplore.In the future, SeBRe could be used to track and quantify anatomical changes in the developing brain, as well as to identify other information, like which genes are expressed during the growth of a mouse brain.“Hopefully, this study is the beginning of a path that will ultimately allow us and the community to explore alterations in brain structure and function,” Karayannis said. “Not only across different developmental stages, but also in devastating brain disorders, helping to identify new biomarkers and form novel hypotheses about disease generation and progression.”Let us know what you like about Geek by taking our survey. Stay on targetlast_img

Leave a Reply

Your email address will not be published. Required fields are marked *