Research Outcomes

Understanding nanoparticle morphology is essential for elucidating their physical behavior and functional performance. Such morphological characterization of nanoparticles is important for understanding structure–function relationships in colloidal and bionano systems. Here, we present a deep-learning-based framework that enables nanoparticle morphology classification from standard nanoparticle tracking analysis (NTA) data by integrating features derived from single-particle Brownian-motion trajectories and temporal fluctuations in scattered-light intensity. The combined feature representation consistently improved binary (two-category) classification performance across all particle-type combinations, reducing the low-accuracy cases observed in single-feature models and achieving accuracies exceeding 0.82 with 100-frame data. For three-class classification, the per-class correctness averaged approximately 80%. The integrated approach maintained stable performance even with limited particle counts or short data lengths (down to ∼20 frames), underscoring its robustness for practical scenarios such as biomedical diagnostics, rare-material analysis, and environmental nanoparticle monitoring. These capabilities highlight its practical relevance for applications such as extracellular vesicle characterization and nanomedicine quality control. By integrating complementary trajectory- and scattering-intensity-based descriptors, this framework offers a reliable and practical route to scalable, morphology-resolved nanoparticle analysis in liquid media using standard NTA measurements.

https://doi.org/10.1021/acsanm.6c01701