Yoshiyuki Kubota

Meso2EM links mesoscale functional imaging to targeted electron microscopy by preserving neuron identity across scale changes using shared landmarks.

This study uses a mouse model of artificial hibernation to show that specific synaptic engram structures are preserved despite widespread spine and synapse elimination, suggesting their role in long-term memory retention.

This paper introduces a deep learning method to enhance the resolution and speed of point-scanning imaging systems by restoring undersampled images.

UNI-EM provides a user-friendly software environment for DNN-based segmentation of EM images, enabling non-experts to perform 2D and 3D segmentation without programming knowledge.