The paper introduces D-LMBmap, an integrated open-source software suite designed for the automated 3D mapping of whole-brain neuronal connectivity using light-sheet fluorescence microscopy. This framework addresses critical bottlenecks in neuroimaging by offering advanced tools for automated axon segmentation, brain-style transfer, and multiscale 3D registration. By utilizing deep neural networks and innovative data augmentation, the system significantly reduces the manual labor typically required for annotating complex axonal projections and identifying anatomical regions. It also employs a multi-constraint registration strategy that ensures high accuracy when aligning experimental samples with standard brain atlases, even in damaged tissues or across different imaging modalities. Ultimately, this comprehensive pipeline enables high-throughput, mesoscale profiling of diverse neuron types, providing researchers with a more efficient method for quantifying and visualizing intricate brain-wide circuits.

References:

• Li Z, Shang Z, Liu J, et al. D-LMBmap: a fully automated deep-learning pipeline for whole-brain profiling of neural circuitry[J]. Nature Methods, 2023, 20(10): 1593-1604.