Segmentation and Morphological Handedness Classification of Chiral Materials by Deep Learning

Handedness classification plays a crucial role in the synthesis and application of chiral nanomaterials, while currently, it usually relies on manual detection and identification. Artificial intelligence is increasingly being integrated into scientific discovery to achieve goals that might not have been possible using traditional methods alone. Here, we introduce a novel framework to automatically recognize and classify chiral nanoparticles based on their asymmetric morphology in scanning electron microscope images. By combining image segmentation models with convolutional neural networks, we create a workflow to achieve a high accuracy of classification on real SEM images with minimal labeling. The approach has been successfully applied to two chiral nanomaterials, demonstrating its robustness and potential for integration into high-throughput SEM analysis workflows and further studies of chiral materials.