CREMI

这一挑战的目标是评估从串行切片电子显微镜数据中自动重建神经元和神经元连接的算法。不仅通过评估神经元分割的质量，还通过评估检测突触和识别突触伙伴的准确性来进行比较。该挑战是在来自成年黑腹果蝇大脑组织的三个大型且多样化的数据集上进行的，包括神经元分割基础事实和突触连接的注释。一个成功的解决方案将证明其效率和普遍性，并具有巨大的潜力，可以减少在电子显微镜体积中手动重建神经回路所花费的时间。

The goal of this challenge is to evaluate algorithms for automatically reconstructing neurons and their neuronal connections from serial-section electron microscopy (ssEM) data. Performance comparisons will be conducted based not only on the quality of neuronal segmentation, but also on the accuracy of synaptic detection and synaptic partner identification. This challenge utilizes three large, diverse datasets sourced from the brain tissue of adult *Drosophila melanogaster*, which are accompanied by ground-truth annotations for neuronal segmentation and synaptic connections. A successful solution will validate its efficiency and generalizability, while holding great potential to reduce the time spent on manual reconstruction of neural circuits within electron microscopy volumes.

搜集汇总

背景与挑战

背景概述

CREMI数据集是一个用于评估神经元自动重建算法的挑战性数据集，专注于从串行切片电子显微镜数据中进行神经元分割和突触连接检测。它包含来自成年黑腹果蝇大脑组织的三个大型多样化样本，提供神经元分割基础事实和突触注释，旨在推动神经科学中神经回路重建的自动化进程，减少手动工作量。