Delivering medical agents to diseased tissues has been challenging, leading researchers to study the in vivo transport process in the body for improving delivery. Many imaging techniques exist for mapping the distribution of medical agent-carrying nanoparticles in tissues, but they cannot capture the three-dimensional context of tissues with single nanoparticle resolution. Here, we developed 3DEM-NPD, a three-dimensional electron microscopy (3D EM) machine learning strategy to image and map single nanoparticle distributions (NPD) in tissues. 3DEM-NPD provides unbiased visualization and quantification of individual nanoparticles within organs. We applied this technique to quantify nanoparticle transport through tumor blood vessel endothelial cells. We measured the cell diameter, surface area, and volume and found that traditional 2D EM cannot accurately measure these features. We used machine learning to locate over 550,000 nanoparticles in less than 3 h with an accuracy of over 82%. The 3DEM-NPD method allowed us to establish a metric to quantify nanoparticle transport at the single nanoparticle level and to quantify the morphological features of ~2,800 vesicles. We find that on average there are only 2.4 nanoparticles per vesicle with a theoretical maximum of 158 nanoparticles per vesicle (~66x increase). These surprising results suggest the need to increase vesicle encapsulation efficiency for improved transport and they provide a benchmark for increasing nanoparticle transport and delivery. This technique may provide unique insights into the interactions between medical agents, drug carriers, emerging materials, and cells at the single-nanoparticle level throughout tissues.
Read full abstract