Dr. Thomas Macrina | AI for Whole Brain Circuit Mapping
Oct 13, 2023
auto_awesome
Dr. Thomas Macrina, CEO of Zetta AI, explores advancements in AI-analyzed neuroimaging and the recovery of petascale cortical circuits. The talk delves into AI intricacies, potential pathways for whole-brain connectomes, and challenges in mapping neural circuits. They discuss the impact of AI on proofreaders and clarify terminology. The podcast also explores the limitations of connectomes and the goal of accessibility for neuroscientists. Additionally, they discuss the hiring process for machine learning projects and the challenges in analyzing neural circuits. Future plans include working with brain size data sets and launching a brain connects program.
Advances in AI and 3D electron microscopy have enabled the reconstruction of larger neural circuits, paving the way for a better understanding of brain structure and function.
Continued advancements in reconstruction accuracy and automation hold promise for future connectomics research.
Connectomics has the potential to transform our understanding of the brain and unlock new insights into brain function and structure.
Deep dives
Neural circuits - Understanding the brain's function
Neural circuits, or the wiring diagrams of the brain, play a crucial role in understanding how the brain functions. By mapping neuron-to-neuron connectivity at the level of single synapses, researchers can gain insights into perception, memory, decision-making, and more. Studying neural circuits also helps in identifying miswiring that underlies brain disorders. Advances in AI and 3D electron microscopy have enabled the reconstruction of larger neural circuits, paving the way for a better understanding of brain structure and function.
The challenge of mapping neural circuits
Mapping neural circuits is a complex process that involves imaging the tissue using 3D electron microscopy. After imaging, computational steps such as volume assembly, cell segmentation, and connectivity analysis are performed. However, the large physical volumes and high resolution required for neural circuit mapping present significant challenges, as they generate massive amounts of data. Overcoming these challenges by improving imaging technology and AI algorithms is crucial for advancing the field of connectomics.
Progress in neural circuit reconstruction
Advancements in AI and deep learning have revolutionized the field of connectomics. Convolutional nets trained with supervised learning have been successfully used for semi-automatic reconstruction of neural circuits. This approach involves volume assembly, cell segmentation, and connectivity analysis. These techniques have been applied to large-scale reconstructions like the mouse cortex and the fruit fly brain, enabling the analysis of neural circuits with hundreds of thousands of neurons. Continued advancements in reconstruction accuracy and automation hold promise for future connectomics research.
The potential of connectomics and its implications
Connectomics has the potential to transform our understanding of the brain and unlock new insights into brain function and structure. By providing detailed maps of neural circuits, researchers can gain insights into how the brain perceives, processes memories, and executes decisions. Connectomics can aid in understanding brain disorders that stem from miswiring at the neuronal level. The ability to accurately reconstruct larger neural circuits and the integration of functional data could further enhance our understanding of the brain and pave the way for new discoveries in neuroscience.
The future of connectomics and the role of AI
The future of connectomics holds the promise of reconstructing and understanding more complex neural circuits, including the whole mouse brain. Despite the challenges ahead, such as improving segmentation accuracy and effectively analyzing large-scale connectomes, recent advancements in AI, deep learning frameworks, and GPU computing have propelled the field forward. Collaboration between neuroscientists, AI engineers, and imaging experts will be crucial in driving future advancements in connectomics and making this groundbreaking research more accessible to all researchers.
Dr. Thomas Macrina, CEO of Zetta AI and possessing a rich background in deploying deep learning to map neural circuits via 3D electron microscopy datasets, explores the rapid advancements in AI-analyzed neuroimaging, offering insights into the accelerated recovery of synaptic-level neural circuits.
Key Highlights
Substantial progress has been realized in the recovery of petascale cortical circuits, with an impending delivery of a whole brain connectome for fruit flies on the horizon.
The talk delves into the intricacies of AI that empower these developments, probing potential pathways to realize the exascale whole-brain connectome of a mouse.
Foresight Institute is a research organization and non-profit that supports the beneficial development of high-impact technologies. Since our founding in 1987 on a vision of guiding powerful technologies, we have continued to evolve into a many-armed organization that focuses on several fields of science and technology that are too ambitious for legacy institutions to support.
Allison Duettmann
The President and CEO of Foresight Institute, Allison Duettmann directs the Intelligent Cooperation, Molecular Machines, Biotech & Health Extension, Neurotech, and Space Programs, alongside Fellowships, Prizes, and Tech Trees. She has also been pivotal in co-initiating the Longevity Prize, pioneering initiatives like Existentialhope.com, and contributing to notable works like "Superintelligence: Coordination & Strategy" and "Gaming the Future".