A Novel Algorithm for Automatic Brain Structure Segmentation from MRI

He, Qiang; Karsch, Kevin; Duan, Ye

doi:10.1007/978-3-540-89639-5_53

Cited by 2 publications

(3 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, research has been focused on automated MRI segmentation of anatomical brain structures (gray matter, GM; white matter, WM; and cerebrospinal fluid, CSF) using a variety of methods including fuzzy connectedness (16), constrained Gaussian mixture models (GMM) (17), deformable models with K‐means clustering (18), spatially constrained fuzzy kernel clustering (14), graph cuts with tissue prior models (19), and hidden Markov models (HMM) (20). In some cases, the lesion has been modeled as a reject class (21) or outliers of the normal tissue models.…”

mentioning

confidence: 99%

Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Ghosh¹,

Recker²,

Shah³

et al. 2011

Magnetic Resonance Imaging

View full text Add to dashboard Cite

Purpose: To develop and compare an automated detection system for ischemic lesions in a neonatal model of bilateral carotid artery occlusion with hypoxia (BCAO-H) from T2 weighted MRI (T2WI) to the currently used ''gold standard'' of manual segmentation.Materials and Methods: Forty-three P10 BCAO-H rat pups and 8 controls underwent T2WI at 1 day and 28 days. A computational imaging method, Hierarchical Region Splitting (HRS), was developed to automatically and rapidly detect and quantify 3D lesion and normal appearing brain matter (NABM) volumes.Results: HRS quantified lesion and NABM volumes within 15 s in comparison to 3 h for its manual counterpart, with a high correlation for injury (r 2 ¼ 0. 95; P ¼ 8.6 Â 10 À7 ) and NABM (r 2 ¼ 0. 92; P ¼ 1.4 Â 10 À22 ). Average lesion volumes for mild, moderate, and severe injuries were 3.85%, 28.85%, and 52.98% for HRS and 0.51%, 24.22%, and 48.74% for manual detection. Lesion volumes and locations were similar for both methods (sensitivity: 0.82, specificity: 0.86, and similarity: 1.47). Conclusion:HRS is an accurate, objective, and rapid method to quantify injury evolution in neonatal hypoxic ischemic injury models.

show abstract

mentioning

confidence: 99%

Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Ghosh¹,

Recker²,

Shah³

et al. 2011

Magnetic Resonance Imaging

View full text Add to dashboard Cite

show abstract

“…Since these methods use MRI signal contrast to detect abnormalities, they can be modifi ed or extended to evaluate lesion characteristics seen with ischemia or stroke [ 23,24 ] . Reviewed below are some of the different available approaches, focused on automated methods such as GMMs [ 25 ] , MRFs [ 26 ] , normalized graph cut [ 27 ] , and K-means clustering [ 28 ] for the detection of brain abnormalities. Although many of these methods have considerable similarities and are sometimes indistinguishable, we describe them separately.…”

Section: Current Computational Approaches For Lesion Detectionmentioning

confidence: 99%

“…In some instances, after CGMM based initial tissue segmentation, deformable models along with K-means clustering have been used to improve MRI tissue classifi cation within the brain so as to generate a tissue type atlas [ 28 ] . In these heavily model driven approaches, lesions of a known shape can be readily identifi ed if the lesion encompasses an entire brain region (e.g., putamen) but this approach does not work when the lesions cross anatomical boundaries, as is the case in the majority of stroke patients.…”

Section: K-means Clusteringmentioning

confidence: 99%

Computational Analysis: A Bridge to Translational Stroke Treatment

Ghosh

Sun

Turenius

et al. 2012

Translational Stroke Research

View full text Add to dashboard Cite

Objective rapid quantifi cation of injury using computational methods can improve the assessment of the degree of stroke injury, aid in the selection of patients for early or specifi c treatments, and monitor the evolution of injury and recovery. In this chapter, we use neonatal ischemia as a case-study of the application of several computational methods that in fact are generic and applicable across the age and disease spectrum. We provide a summary of current computational approaches used for injury detection, including Gaussian mixture models (GMM), Markov random fi elds (MRFs), normalized graph cut, and K-means clustering. We also describe more recent automated approaches to segment the region(s) of ischemic injury including hierarchical region splitting, support vector machine, a brain symmetry/asymmetry integrated model, and a watershed method that are robust at different developmental stages. We conclude with our assessment of probable future research directions in the fi eld of computational noninvasive stroke analysis such as automated detection of the ischemic core and penumbra, monitoring of implanted neuronal stem cells in the ischemic brain, injury localization specifi c to different brain anatomical regions, and quantifi cation of stroke evolution, recovery and spatiotemporal interactions between injury volume/severity and treatment. Computational analysis is expected to open a new horizon in current clinical and translational stroke research by exploratory data mining that is not detectable using the standard "methods" of visual assessment of imaging data.

show abstract

A Novel Algorithm for Automatic Brain Structure Segmentation from MRI

Cited by 2 publications

References 20 publications

Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Computational Analysis: A Bridge to Translational Stroke Treatment

Contact Info

Product

Resources

About