As a non-regulatory research organization, NIST cultivates trust in AI-related research by using rigorous methods and decades-long experience to develop tools, technical guidance, and best practice guides, that accurately measure and understand the capabilities and limitations of AI technologies and the underlying data that catalyzes them. NIST’s AI research efforts can be broken down into two categories: Foundational and Applied Research.
Applied Research to Deploy AI for NIST Research
This area focuses on advancing application of AI to NIST metrology problems and enabling NIST scientists to draw routinely on machine learning and AI tools to gain deeper insight into their research. Some NIST efforts in this area include:
- Subspace-informed Deep Learning Solutions for Nanoscale Microscopy
NIST is developing AI and machine learning methods to improve the fidelity of novel nanoscale microscopy techniques that are under research by NIST physicists. (Read more)
- AI-based Measurements For Imaging
NIST streamlines deriving measurements from terabyte-sized images and improves image-based measurement accuracy via web-based applications addressing image analyses and AI-based modeling. (Read more)
- Deep Learning for MRI Reconstruction and Analysis
The goal of the project is to initiate the development of the metrology and standards infrastructure to ensure that medical DL-based systems are (1) trained on validated physics-based data and (2) provide reliability, accuracy, and explainability. (Read more)
- Imaging & Artificial Intelligence (AI) to Assess Tissue Quality: We have collaborated with Peter Bajcsy (NIST/ITL) and Kapil Bharti (NIH/NEI) to develop an approach for non-invasively assessing quality of clinical-grade tissue-engineered human retinal pigment epithelium (RPE). The method uses quantitative brightfield absorbance imaging (QBAM) and AI. Absorbance imaging was implemented in Kapil Bharti’s Good Manufacturing Practice facility where we noninvasively collected brightfield images of the clinical-grade constructs. Constructs were imaged using absorbance imaging and traditional RPE potency test data were collected. The RPE potency tests were measurements of polarized secretion of vascular endothelial growth factor (VEGF) and measurements of trans-epithelial resistance (TER). Peter Bajcsy's team directed the image processing, image analysis and implementation of AI methods. AI algorithms were trained to predict potency (VEGF and TER) from the absorbance images. The trained AI algorithm made correct predictions for 35 of 36 test image data sets. This work demonstrates how a garden variety microscope, if used carefully, can make a precise, reproducible measurement of tissue quality.
Funding opportunities are available through grants and cooperative agreements via the Measurement Science and Engineering (MSE) Research Grant Program.
See also AI Research - Foundational