Project Description

Many scientific research applications rely on the analysis of imagery produced using scientific instruments such as microscopes. This analysis benefits from neural networks because, unlike other computer vision approaches, this approach does not rely on hand-engineered features. Instead, neural networks learn the features needed for the image-processing task from the data. In this project, Patton’s team will use the Multinode Evolutionary Neural Networks for Deep Learning and the Evolutionary Optimization of
euromorphic Systems, which are artificial intelligence systems, to accelerate understanding of two different application areas: (1) nanoscale material fabrication using scanning transmission electron microscopy (STEM) imagery and (2) cancer research and treatment using digital pathology imagery. The results of this work could bring real-time, image-based feedback for STEM significantly closer to a reality and enable the linking of pathology images and cancer registry abstracts to create a unique, population-wide, molecular view of

Allocation History

Source Hours Start Date End Date