Reproduce “ZoomSEM: accelerating large-scale brain connectomics with intelligent SEM super-resolution” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jiateng Shou, Zhiwei Xiong, Ruikang Xu, Zhicheng Wang, Peiyao Shi, Te Shi, Haiqun Jin, Chunying Yin and 5 more

ZoomSEM reconstructs coarse serial electron microscopy scans into fine-resolution volumes using real paired scans and neighboring-section information. Its recurrent transformer combines aligned features across sections using affine-deformable multi-scale alignment. On mouse entorhinal-cortex data, it supports 4×, 8×, and 16× upscaling; from 32 nm toward 4 nm, authors report recovering fine structures missed by competitors. A modeled estimate gives 21.0× speedup, including acquisition and inference; overhead limits gains.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.