NanoXtract

NanoXtract extracts quantitative image descriptors from transmission electron microscopy (TEM) images of nanomaterials and uses them to support read-across modeling to predict zeta potential and inform interpretation of nanomaterial behavior and toxicity.

Key Features:

• Image Descriptor Extraction: Processes TEM images to generate 18 quantitative image descriptors that characterize nanomaterial morphology and features.
• Zeta Potential Prediction (Read-across): Uses the extracted image descriptors to develop read-across predictive models for zeta potential of nanomaterials.
• Nearest Neighbor Grouping: Employs a nearest neighbor approach to group nanomaterials by similarity based on TEM-derived features.
• Insights into Toxicity Drivers: Modeling identifies aspect ratio and core composition as significant drivers influencing nanomaterial toxicity.

Scientific Applications:

• Nanotoxicology: Supports assessment of nanomaterial interactions with cells and potential toxic effects by linking image descriptors to biological outcomes.
• Stability and Reactivity Prediction: Enables inference of colloidal stability and agglomeration tendencies through zeta potential prediction from image-derived descriptors.
• Safe-by-Design Guidance: Informs design of safer nanomaterials by identifying structural descriptors associated with toxicity risk.
• Regulatory Assessment Support: Provides quantitative descriptors and read-across predictions to aid safety and regulatory evaluations of nanomaterials.

Methodology:

Image processing of TEM images to extract 18 quantitative image descriptors; nearest neighbor grouping based on TEM-derived similarity features; development of a read-across predictive model using those descriptors to predict zeta potential.