Restoration Enhanced Spine and Neuron (RESPAN) Analysis is an efficient and accessible pipeline leveraging the latest advancements in image restoration, U-Net segmentation, and GPU processing. For ease of use this pipeline is made available as a standalone application for Windows. Code is also available to run from Python.

Developed in collaboration with the Polleux Lab (Zuckerman Institute, Columbia University).

• GUI based automatic segmentation of spines and dendrites
• 3D morphological and intensity measurements
• 3D spatial relationships
• tabular and image outputs, including validation images, to ensure accuracy
• Spine Arrays - extract each spine as a 3D volume for visualization and further machine learning training of spine morphological features and spine classification

• perform automated quantitative assessment of the performance of the trained U-Net model and SpinePipe pipeline
• validate and readily evaluate improvements to the pipeline models using ground-truth data

• simplified GUI-based tool to facilitate training of new nnUnet models

• spine neck analysis
• track spine morphology and signal intensity over time