A novel region-based level set method initialized with mean shift clustering for automated medical image segmentation

Bai, Pei Rui; Liu, Qing Yi; Li, Lei; Teng, Sheng Hua; Li, Jing; Cao, Mao Yong

doi:10.1016/j.compbiomed.2013.08.024

Cited by 42 publications

(24 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Curve fitting of tissue-boundaries based on partial differential equations, including parametric methods like active contour snakes (Kazerooni et al, 2011; Liang et al, 2006; Zhou and Xie, 2013), non-parametric methods like level-set propagation (Droske et al, 2001), and their several modified versions (Bai et al, 2013; Kazemifar et al, 2014; Le Guyader and Vese, 2008; Liang et al, 2006; Mesejo et al, 2014; Somkantha et al, 2011; Wang et al, 2013) have been adopted in different medical image analysis applications. Success of these methods heavily depends on MRI intensity contrast edge detection (Kazerooni et al, 2011; Liang et al, 2006; Somkantha et al, 2011), seed-initialization (which is often manual) (Zhou and Xie, 2013) and choice of energy function (Kazerooni et al, 2011; Liang et al, 2006) that is to be minimized by heuristic gradient descent algorithms (Wang et al, 2013).…”

Section: Discussionmentioning

confidence: 99%

“…Success of these methods heavily depends on MRI intensity contrast edge detection (Kazerooni et al, 2011; Liang et al, 2006; Somkantha et al, 2011), seed-initialization (which is often manual) (Zhou and Xie, 2013) and choice of energy function (Kazerooni et al, 2011; Liang et al, 2006) that is to be minimized by heuristic gradient descent algorithms (Wang et al, 2013). These methods are often computationally expensive due to the need for an initial training module (Kazemifar et al, 2014; Mesejo et al, 2014), need for manual interventions (Liang et al, 2006; Zhou and Xie, 2013), need for obtaining local-minima during energy minimization instead of global minima (inherent problem of greedy search in gradient descent algorithms) (Mesejo et al, 2014; Wang et al, 2013) and sometime requires prior models (Bai et al, 2013; Le Guyader and Vese, 2008; Wang et al, 2013) or atlas-registration (Kazemifar et al, 2014) with their associated limitations as mentioned earlier. These pitfalls often restrict use of level-set and active contour snake methods in real-time clinical applications.…”

Section: Discussionmentioning

confidence: 99%

“…While prior probabilistic models for brain anatomy and specific diseases have been used (Corso et al, 2008), they are not robust to injury induced brain distortions, as model-subject co-registration often fails (Cuadra et al, 2004; Ghosh et al, 2012a). Curve fitting based methods like active contour snakes (Droske et al, 2001; Liang et al, 2006; Zhou and Xie, 2013), level-set propagation (Droske et al, 2001) and their combined and/or modified versions (Bai et al, 2013; Kazemifar et al, 2014; Le Guyader and Vese, 2008; Liang et al, 2006; Mesejo et al, 2014; Somkantha et al, 2011; Wang et al, 2013) have been applied to medical image segmentation. But these methods suffer from manual interventions (Liang et al, 2006; Zhou and Xie, 2013), computational complexity (Kazemifar et al, 2014; Mesejo et al, 2014), dependence on MRI contrast levels (Kazerooni et al, 2011; Liang et al, 2006; Somkantha et al, 2011) and inadequate cues for efficient registration to prior-models (Bai et al, 2013; Le Guyader and Vese, 2008; Wang et al, 2013) and atlases (Kazemifar et al, 2014), specifically in low-contrast noisy MRI data (Zhou and Xie, 2013).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automated detection of brain abnormalities in neonatal hypoxia ischemic injury from MR images

Ghosh

Sun

Bhanu

et al. 2014

Medical Image Analysis

View full text Add to dashboard Cite

We compared the efficacy of three automated brain injury detection methods, namely symmetry-integrated region growing (SIRG), hierarchical region splitting (HRS) and modified watershed segmentation (MWS) in human and animal magnetic resonance imaging (MRI) datasets for the detection of hypoxic ischemic injuries (HII). Diffusion weighted imaging (DWI, 1.5T) data from neonatal arterial ischemic stroke (AIS) patients, as well as T2-weighted imaging (T2WI, 11.7T, 4.7T) at seven different time-points (1, 4, 7, 10, 17, 24 and 31 days post HII) in rat-pup model of hypoxic ischemic injury were used to check the temporal efficacy of our computational approaches. Sensitivity, specificity, similarity were used as performance metrics based on manual (‘gold standard’) injury detection to quantify comparisons. When compared to the manual gold standard, automated injury location results from SIRG performed the best in 62% of the data, while 29% for HRS and 9% for MWS. Injury severity detection revealed that SIRG performed the best in 67% cases while HRS for 33% data. Prior information is required by HRS and MWS, but not by SIRG. However, SIRG is sensitive to parameter-tuning, while HRS and MWS are not. Among these methods, SIRG performs the best in detecting lesion volumes; HRS is the most robust, while MWS lags behind in both respects.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automated detection of brain abnormalities in neonatal hypoxia ischemic injury from MR images

Ghosh

Sun

Bhanu

et al. 2014

Medical Image Analysis

View full text Add to dashboard Cite

show abstract

“…The level set method affords numerous advantages: it is implicit, is parameter-free, provides a direct way to estimate the geometric properties of the evolving structure, allows for change of topology, and is intrinsic. [13,14] It can be used to define an optimization framework, as proposed by Zhao, Merriman and Osher in 1996. One can conclude that it is a very convenient framework for addressing numerous applications of computer vision and medical image analysis [15,16].…”

Section: Materials and Methodsologymentioning

confidence: 99%

Brain Image Segmentation Based on Fuzzy Clustering

Mohammed

2018

MJS

View full text Add to dashboard Cite

The segmentation performance is topic to suitable initialization and best configuration of supervisory parameters. In medical image segmentation, the segmentation is very important when the diagnosing becomes very hard in medical images which are not properly illuminated. This paper proposes segmentation of brain tumour image of MRI images based on spatial fuzzy clustering and level set algorithm. After performance evaluation of the proposed algorithm was carried on brain tumour images, the results showed confirm its effectiveness for medical image segmentation, where the brain tumour is detected properly.

show abstract

“…In our previous work, a novel region-based level set method (MS-RSF level set method) by integrating the mean shift clustering with the RSF model have been proposed [16]. The MS-RSF model can obtain appropriate initialization by mapping the clustering results of mean shift to a binary step function.…”

Section: Introductionmentioning

confidence: 99%

A Novel Integration Scheme Based on Mean Shift and Region-Scalable Fitting Level Set for Medical Image Segmentation

Bai¹,

Song²,

Bi³

et al. 2015

Proceedings of the 2015 3rd International Conference on Machinery, Materials and Information Technology Applications

Self Cite

View full text Add to dashboard Cite

Abstract. In this study, a novel integration scheme for coupling the results of mean shift with initial contours of the region-scalable fitting level set method (RSF model) is presented. There are two main contributions in the study. First, a new adaptive threshold formula to fit dynamic range of the mean shift clustering results is proposed. Second, a double-side mapping mode is presented to improve the robustness of initialization. Experimental results demonstrate the adaptability and robustness of the proposed method, and accurate segmentation could be obtained for medical images.

show abstract

A novel region-based level set method initialized with mean shift clustering for automated medical image segmentation

Cited by 42 publications

References 22 publications

Automated detection of brain abnormalities in neonatal hypoxia ischemic injury from MR images

Automated detection of brain abnormalities in neonatal hypoxia ischemic injury from MR images

Brain Image Segmentation Based on Fuzzy Clustering

A Novel Integration Scheme Based on Mean Shift and Region-Scalable Fitting Level Set for Medical Image Segmentation

Contact Info

Product

Resources

About