Fig. 3.

Overall workflow of the 3D reconstruction of C. elegans. (A) Enhancement of image quality by novel machine learning model-based super-resolution and denoising while the details of C. elegans in microscopy images are preserved. (B) Machine learning-based accurate segmentation of C. elegans. (C) Template matching algorithm-based accurate measurement of worm rotation speed. (D) Alignment of C. elegans body for precise 3D reconstruction thereafter. (E) 3D reconstruction of worms based on the customized shape-from-silhouette 3D reconstruction algorithms. (F) Generated 3D models of C. elegans.