A meta-algorithm for brain extraction in MRI

Laboratory of Neuro Imaging, Department of Neurology, David Geffen School of Medicine at UCLA, 710 Westwood Plaza, Los Angeles, CA 90095-1769, USA.
NeuroImage (Impact Factor: 6.36). 11/2004; 23(2):625-37. DOI: 10.1016/j.neuroimage.2004.06.019
Source: PubMed


Accurate identification of brain tissue and cerebrospinal fluid (CSF) in a whole-head MRI is a critical first step in many neuroimaging studies. Automating this procedure can eliminate intra- and interrater variance and greatly increase throughput for a labor-intensive step. Many available procedures perform differently across anatomy and under different acquisition protocols. We developed the Brain Extraction Meta-Algorithm (BEMA) to address these concerns. It executes many extraction algorithms and a registration procedure in parallel to combine the results in an intelligent fashion and obtain improved results over any of the individual algorithms. Using an atlas space, BEMA performs a voxelwise analysis of training data to determine the optimal Boolean combination of extraction algorithms to produce the most accurate result for a given voxel. This allows the provided extractors to be used differentially across anatomy, increasing both the accuracy and robustness of the procedure. We tested BEMA using modified forms of BrainSuite's Brain Surface Extractor (BSE), FSL's Brain Extraction Tool (BET), AFNI's 3dIntracranial, and FreeSurfer's MRI Watershed as well as FSL's FLIRT for the registration procedure. Training was performed on T1-weighted scans of 136 subjects from five separate data sets with different acquisition parameters on separate scanners. Testing was performed on 135 separate subjects from the same data sets. BEMA outperformed the individual algorithms, as well as interrater results from a subset of the scans, when compared for the mean Dice coefficient, a rating of the similarity of output masks to the manually defined gold standards.

24 Reads
  • Source
    • "In meta-algorithm (Rex et al., 2004; Shi et al., 2012), several existing brain extraction methods are combined to compensate the weaknesses of each method. However, the model should be specifically designed through meta-algorithm to gain optimum performance when new data are sufficiently different from previous training datasets (Rex et al., 2004). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain extraction is an important procedure in brain image analysis. Although numerous brain extraction methods have been presented, enhancing brain extraction methods remain challenging because brain MRI images exhibit complex characteristics, such as anatomical variability and intensity differences across different sequences and scanners. To address this problem, we present a Locally Linear Representation-based Classification (LLRC) method for brain extraction. A novel classification framework is derived by introducing the locally linear representation to the classical classification model. Under this classification framework, a common label fusion approach can be considered as a special case and thoroughly interpreted. Locality is important to calculate fusion weights for LLRC; this factor is also considered to determine local anchor embedding is more applicable in solving locally linear coefficients compared with other linear representation approaches. Moreover, LLRC supplies a way to learn the optimal classification scores of the training samples in the dictionary to obtain accurate classification. The International Consortium for Brain Mapping and the Alzheimer's Disease Neuroimaging Initiative databases were used to build a training dataset containing 70 scans. To evaluate the proposed method, we used three public available datasets (IBSR1, IBSR2, and LPBA40, with a total of 78 scans). Experimental results demonstrate that the proposed method outperforms the four common brain extraction methods (BET, BSE, GCUT, and ROBEX), and is comparable to the performance of BEaST, while being more accurate on some datasets compared with BEaST.
    NeuroImage 05/2014; DOI:10.1016/j.neuroimage.2014.01.059 · 6.36 Impact Factor
  • Source
    • "Huang [9] applied an expectation-maximization algorithm to a mixture of Gaussian models to determine the initial brain contour for use in the geodesic active contour evolution. Rex [10] developed a brain extraction meta-algorithm (BEMA), which executes four extraction algorithms to obtain individual extraction results in concert with an atlas-based registration procedure, and combines the results to achieve the final improved brain extraction result using a variety of anatomically specified Boolean functions. It is obvious that BEMA is time consuming. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background The extraction of brain tissue from cerebral MRI volume is an important pre-procedure for neuroimage analyses. The authors have developed an accurate and robust brain extraction method using a hybrid level set based active contour neighborhood model. Methods The method uses a nonlinear speed function in the hybrid level set model to eliminate boundary leakage. When using the new hybrid level set model an active contour neighborhood model is applied iteratively in the neighborhood of brain boundary. A slice by slice contour initial method is proposed to obtain the neighborhood of the brain boundary. The method was applied to the internet brain MRI data provided by the Internet Brain Segmentation Repository (IBSR). Results In testing, a mean Dice similarity coefficient of 0.95±0.02 and a mean Hausdorff distance of 12.4±4.5 were obtained when performing our method across the IBSR data set (18 × 1.5 mm scans). The results obtained using our method were very similar to those produced using manual segmentation and achieved the smallest mean Hausdorff distance on the IBSR data. Conclusions An automatic method of brain extraction from cerebral MRI volume was achieved and produced competitively accurate results.
    BioMedical Engineering OnLine 04/2013; 12(1):31. DOI:10.1186/1475-925X-12-31 · 1.43 Impact Factor
  • Source
    • "Preprocessing and definition of cortical and subcortical gray matter regions on structural images were conducted in the UCLA Laboratory of Neuro Imaging (LONI) Pipeline Processing Environment [33-35] and using FreeSurfer’s automated brain segmentation software (FreeSurfer 4.0.5,, as described in the work of Fischl and Dale [36-38]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Published structural neuroimaging studies of prenatal cocaine exposure (PCE) in humans have yielded somewhat inconsistent results, with several studies reporting no significant differences in brain structure between exposed subjects and controls. Here, we sought to clarify some of these discrepancies by applying methodologies that allow for the detection of subtle alterations in brain structure. Methods We applied surface-based anatomical modeling methods to magnetic resonance imaging (MRI) data to examine regional changes in the shape and volume of the caudate and putamen in adolescents with prenatal cocaine exposure (n = 40, including 28 exposed participants and 12 unexposed controls, age range 14 to 16 years). We also sought to determine whether changes in regional brain volumes in frontal and subcortical regions occurred in adolescents with PCE compared to control participants. Results The overall volumes of the caudate and putamen did not significantly differ between PCE participants and controls. However, we found significant (P <0.05, uncorrected) effects of levels of prenatal exposure to cocaine on regional patterns of striatal morphology. Higher levels of prenatal cocaine exposure were associated with expansion of certain striatal subregions and with contraction in others. Volumetric analyses revealed no significant changes in the volume of any subcortical region of interest, but there were subtle group differences in the volumes of some frontal cortical regions, in particular reduced volumes of caudal middle frontal cortices and left lateral orbitofrontal cortex in exposed participants compared to controls. Conclusions Prenatal cocaine exposure may lead to subtle and regionally specific patterns of regional dysmorphology in the striatum and volumetric changes in the frontal lobes. The localized and bidirectional nature of effects may explain in part the contradictions in the existing literature.
    Journal of Neurodevelopmental Disorders 08/2012; 4(1):22. DOI:10.1186/1866-1955-4-22 · 3.27 Impact Factor
Show more

Preview (2 Sources)

24 Reads
Available from