Sandia National Labs FY20 LDRD Annual Report


High-throughput discovery of tools for brain protection. Protection of the brain from chemical, biological, and radiological threats is dependent on successful non-invasive delivery of medical countermeasures across the blood-brain barrier. Existing delivery methods of cargo to the brain lead to inefficient or non-specific penetration of the blood-brain-barrier, so the team focused on using functional and selective screening of brain-penetrating nanobodies under physiological conditions to identify novel noninvasive routes of delivery to the brain. Through this project, the team generated a large >10 10 library of nanobodies already being utilized in multiple Sandia studies, including COVID-19 research. Several of these nanobodies were identified as promising blood-brain-barrier shuttles and are being further investigated. If successfully developed, these blood-brain-barrier shuttles will assist in national security threat mitigation. (PI: Maxwell Stefan) Ultra-efficient sensing system through holistic design. Many metrics used to evaluate high- quality optical devices are focused on imaging, and therefore biased by what observers perceive to be “good” images. Many optical devices, however, are now being used as part of a larger system to perform machine learning and are therefore less constrained by traditional imaging performance. This project developed computational imaging techniques that design the optical component while simultaneously developing the machine learning algorithms. By leveraging the photonics, optical science, and machine learning capabilities at Sandia, the team developed generalizable techniques to create nearly arbitrary linear optical systems. These systems can directly measure significantly compressed data while still retaining the ability to classify scenes via a machine learning element. Creating task-specific optical devices can enable lower size, weight, and power-consuming sensors ultimately capable of performing a machine learning task. This technology could influence the growing domain of ubiquitous sensing and suggests that computational imaging can play a practical role in national security relevant problems. (PI: Gabriel Carlisle Birch)

(Left) Microscope image of a multiphoton lithography-generated refractive computational optical element, designed via a machine learning optimization process. (Top right) The predicted output for the computational sensing device from simulations. (Bottom right) The measured output of the computational sensing device.



Made with FlippingBook - professional solution for displaying marketing and sales documents online