Thresholdmann: A Web tool for interactively creating adaptive thresholds to segment MRI data.

Abstract

Brain extraction and segmentation are the first step for most neuroimaging analyses. Automatic methods work well in adult human brains, but produce unreliable results in non-human data, due to muscle tissue, skull, and luminosity gradients. Thresholdmann (https://neuroanatomy. github.io/thresholdmann) is an open source Web tool for the interactive application of space-varying thresholds to Nifti volumes. No download or installation are required and all processing is done on the user’s computer. Nifti volumes are dragged and dropped onto the Web app and become available for visual exploration in a stereotaxic viewer. A space-varying threshold is then created by setting control points, each with their own local threshold. Each point can be repositioned or removed, and each local threshold can be adjusted in real time using sliders or entering their values numerically. The threshold direction can be switched to allow segmentation of the structure of interest in different imaging modalities, such as T1 and T2 weighted contrasts. The opacity of the mask and the brightness and contrast of the MRI image can be adjusted via sliders. A 3D model of the thresholded mask can be computed to inspect the result in an interactive 3D render. Finally, the thresholded mask, the space varying threshold and the list of control points can be saved for later use in scripted workflows, able to reproduce the thresholded volume from the original data.