mousavikahaki / traceproofreading Goto Github PK

Trace proofreading is a machine learning based method for detection and correction of topologcal errors in automated neuron traces. This is essential for accurate mapping of structural connectivity.

Our understanding of synaptic connectivity in the brain relies on the ability to accurately trace sparsely labeled populations of neurons from 3D optical microscopy stacks of images. A variety of algorithms and software tools have been developed for automated neuron tracing, but the resulting traces often contain topological errors such as broken and incorrectly merged branches. Trace proofreading is an automated method that utilizes machine learning to correct such errors.

Traces of neurons obtained from 3D optical microscopy stacks of images unavoidably contain topological errors such as broken branches, which result from the non-uniform intensity of neurites, and fused branches, which arise from branch proximity and a relatively low resolution of light microscopy.

Trace Proofreading

Seyed Mostafa Mousavi Kahaki: [email protected]
Armen Stepanyants: [email protected]