Oxford Nanoimaging Launches ML-Driven Platform for Lipid Nanoparticle Characterization
Oxford Nanoimaging (ONI), founded in 2016 as a spin-out from the University of Oxford and headquartered in Oxford, UK, and San Diego, California, launched a characterization kit, which includes optimized reagents, imaging hardware, and ML-driven software designed for analyzing lipid nanoparticles (LNPs).
ONI’s LNP Profiler Kit with AutoLNP Software
According to ONI, traditional lipid nanoparticle (LNP) characterization methods, such as CryoEM and bulk assays, can be labor-intensive and often lack sufficient sensitivity for detailed, particle-level assessments. To address this, the company introduced its ONI Application Kit: LNP Profiler, a platform based on super-resolution microscopy, described as "the first and only verified solution within the SMLM and super-resolution space for particle analysis".
The kit combines optimized reagents, specialized imaging hardware, and machine-learning software (AutoLNP) to enable precise, single-particle analysis.
Contents of the ONI LNP Profiler Kit: Includes reagents and assay chips for multiplexed detection of surface markers, ligands, and cargo across 16 lanes, compatible with AutoLNP image analysis.
The platform is designed for quantitative characterization of key LNP parameters, including particle size, cargo encapsulation, payload distribution, and ligand profiling, at nanoscale resolution. The AutoLNP software automates image processing and data analysis to reduce manual effort, assay variability, and potential human errors.
In direct comments, the company explained that for the assay, they applied machine learning to overcome several analytical challenges.
- Firstly, their models were trained on real experimental data to clearly distinguish true particles from background noise. This included identifying variations in particle fragments, aggregates, and differing densities to accurately define particle boundaries and diameter for subsequent analysis steps.
- Secondly, when labeling internal cargo, ONI stated that they optimized signal-to-noise by isolating signals precisely aligned with the particle centroid, minimizing any potential skewing. In cases where artifacts such as solution contaminants, bubbles, or large aggregates appeared in a small subset of the data, the software dynamically excluded these anomalies to avoid distorting the analysis. They added that positivity calculations were similarly adjusted to account for background signals.
- Thirdly, for surface ligand labeling, the company indicated that the approach was similar—identifying the particle centroid allowed accurate colocalization of ligand signals with the actual particles. According to ONI, this ensured positivity measures were corrected for any background or non-specific signals.
- Finally, ONI noted that their ML models were iteratively trained and manually annotated multiple times until achieving optimal performance. The company emphasized that their approach allowed analyses to be performed approximately ten times faster compared to traditional stepwise methods (such as DBSCAN clustering followed by filtering). They also pointed out that the system was cloud-managed, removing the need for specialized on-site hardware, although they planned to deploy local software versions of these models soon.
Liliana Barbieri, Product Manager at ONI, stated that the company's aim is to provide biopharma teams with automated tools for efficient formulation optimization and improved data-driven decision-making in developing RNA therapeutics, gene editing approaches, and vaccines.
ONI previously introduced related specialized products, including the CODI collaborative analysis software in 2021, the EV Profiler kits for extracellular vesicle research, and the Aplo Flow automated fluidics platform in 2024. In 2022, the company raised $75 million in Series B funding led by ARCH Venture Partners and Casdin Capital and expanded its U.S. presence by opening headquarters in San Diego, California.
Topics: Tools & Methods
