Atomic force microscopy (AFM) has become an indispensable tool for characterizing DNA nanostructures. However, conventional AFM analysis relies heavily on manual operation, which is inherently time‐consuming and prone to subjective bias. Herein, a transfer learning‐assisted AFM single‐particle analysis (AFMSPA) tool is presented, which enables automated high‐throughput AFM structural analysis of DNA origami nanostructures. AFMSPA leverages a U‐Net‐based neural network trained on experimentally acquired datasets to automatically detect and classify individual DNA origami in raw AFM images. It further enables the on‐demand extraction of morphological parameters by users. The tool currently supports fully automated quantitative analysis for four representative 2D DNA origami nanostructures, i.e., triangle, rectangle, hexagon, and diatom, each achieving a detection accuracy above 0.95. Moreover, by integrating a human‐in‐the‐loop transfer learning strategy, AFMSPA can be readily adapted to a broad range of DNA origami nanostructures, offering a scalable solution for the statistical analysis of AFM datasets in the field of self‐assembled nanomaterials.
Dai et al. (Thu,) studied this question.