Multi-atlas segmentation of subcortical brain structures via the AutoSeg software pipeline

Abstract

Automated segmenting and labeling of individual brain anatomical regions, in MRI are challenging, due to the issue of individual structural variability. Although atlas-based segmentation has shown its potential for both tissue and structure segmentation, due to the inherent natural variability as well as disease-related changes in MR appearance, a single atlas image is often inappropriate to represent the full population of datasets processed in a given neuroimaging study. As an alternative for the case of single atlas segmentation, the use of multiple atlases alongside label fusion techniques has been introduced using a set of individual “atlases” that encompasses the expected variability in the studied population. In our study, we proposed a multi-atlas segmentation scheme with a novel graph-based atlas selection technique. We first paired and co-registered all atlases and the subject MR scans. A directed graph with edge weights based on intensity and shape similarity between all MR scans is then computed. The set of neighboring templates is selected via clustering of the graph. Finally, weighted majority voting is employed to create the final segmentation over the selected atlases. This multi-atlas segmentation scheme is used to extend a single-atlas-based segmentation toolkit entitled AutoSeg, which is an open-source, extensible C++ based software pipeline employing BatchMake for its pipeline scripting, developed at the Neuro Image Research and Analysis Laboratories of the University of North Carolina at Chapel Hill. AutoSeg performs N4 intensity inhomogeneity correction, rigid registration to a common template space, automated brain tissue classification based skull-stripping, and the multi-atlas segmentation. The multi-atlas-based AutoSeg has been evaluated on subcortical structure segmentation with a testing dataset of 20 adult brain MRI scans and 15 atlas MRI scans. The AutoSeg achieved mean Dice coefficients of 81.73% for the subcortical structures.

Multi-atlas segmentation of biomedical images: A survey

“…Yet, a growing number of methods recognize the value in the deformation fields themselves and propose to use information about the amount of deformation in the computations of the fusion weights. For example, Commowick and Malandain (2007) used the Euclidean norm of the deformation, Ramus et al (2010) used its Jacobian determinant, and Wang et al (2014b) used its harmonic energy.…”

“…These approaches introduce additional complexity to the system, but can outperform standard similarity measures in the atlas selection task. Another approach to increase the efficiency and accuracy of atlas selection utilizes clustering, where the atlases, possibly together with the novel image(s), are analyzed to identify clusters of similar cases using methods such as k-means (Nouranian et al, 2014), affinity propagation (Langerak et al, 2013) and Floyd’s algorithm (Wang et al, 2014b). Then, cluster representatives (or exemplars) are used for the initial search of the most relevant atlases.…”

Postnatal Zika virus infection is associated with persistent abnormalities in brain structure, function, and behavior in infant macaques

“…Datasets were processed using AutoSeg_3.3.2 segmentation package (87) to get the volumes of brain white mater (WM), gray matter (GM), cerebrospinal fluid (CSF), cortical (temporal visual area, temporal auditory area, prefrontal, frontal, parietal and occipital lobes) and subcortical (hippocampus, amygdala) brain areas. Image processing steps included: 1) averaging T1 and T2 images to improve signal-to-noise ratio, 2) intensity inhomogeneity correction, 3) rigid body registration of the subject MRI to the three or the six months UNC-Emory infant RM atlases (35), 4) tissue segmentation and skull-stripping, 5) registration of the atlas to the subject’s brain to generate cortical parcellations (affine followed by deformable SyN/ANTS registration), 6) manual editing of the amygdala and hippocampus was done on all scans using previously published neuroanatomical boundaries (31).…”

Multi-Atlas Library for Eliminating Normalization Failures in Non-Human Primates

“…The concept of a multi-atlas approach has recently gained traction for label mapping (Wu et al 2015; Wang et al 2014). Specifically, more accurate individual subject labeling can be achieved by mapping to multiple templates.…”

“…Existing NHP-specific tools also can demonstrate high failure rates, even for experienced users. Our own experience with several NHP-specific tools (Wang et al 2014; Fedorov et al 2011) has demonstrated failure rates as high as 50%.…”