Automatic segmentation of corpus collasum using Gaussian mixture modeling and Fuzzy C means methods

doi:10.1016/j.cmpb.2013.06.006

Computer Methods and Programs in Biomedicine

Volume 112, Issue 1, October 2013, Pages 38-46

https://doi.org/10.1016/j.cmpb.2013.06.006 Get rights and content

Abstract

This paper presents a comparative study of the success and performance of the Gaussian mixture modeling and Fuzzy C means methods to determine the volume and cross-sectionals areas of the corpus callosum (CC) using simulated and real MR brain images. The Gaussian mixture model (GMM) utilizes weighted sum of Gaussian distributions by applying statistical decision procedures to define image classes. In the Fuzzy C means (FCM), the image classes are represented by certain membership function according to fuzziness information expressing the distance from the cluster centers. In this study, automatic segmentation for midsagittal section of the CC was achieved from simulated and real brain images. The volume of CC was obtained using sagittal sections areas. To compare the success of the methods, segmentation accuracy, Jaccard similarity and time consuming for segmentation were calculated. The results show that the GMM method resulted by a small margin in more accurate segmentation (midsagittal section segmentation accuracy 98.3% and 97.01% for GMM and FCM); however the FCM method resulted in faster segmentation than GMM. With this study, an accurate and automatic segmentation system that allows opportunity for quantitative comparison to doctors in the planning of treatment and the diagnosis of diseases affecting the size of the CC was developed. This study can be adapted to perform segmentation on other regions of the brain, thus, it can be operated as practical use in the clinic.

Introduction

Magnetic resonance imaging (MRI) has superior contrast properties according to other medical imaging modalities in soft tissue imaging [1]. Therefore, MRI has been extensively preferred in quantitative image processing studies for brain and other organs. Quantitative volumetric measurement and qualitative representation of brain tissues are quite helpful to evaluate various pathologies in the brain. For this aim, brain image segmentation plays first and an important role [2]. Each region in the brain has very different and complex functions in connection with each other.

The corpus callosum (CC) is located underneath the cerebrum at the center of the brain. CC is the largest white matter region connecting right and left cerebral hemispheres and it is consist of about 200–350 million fibers in humans. CC plays a vital role in the integration by transferring sensorial, cognitive, learning, mnemonic and motor information between the two brain hemispheres [3], [4], [5].

Many neurological diseases cause changes in the structure and size of the corpus callosum. Alterations in the structure of the corpus callosum (CC) have been observed in schizophrenia [5], autism [6], epilepsy [7], childhood stuttering [8], and in the effects of smoking on corpus callosum volume [9]. Therefore, quantitative calculation by segmentation as precisely as possible of the CC has a great importance in the process of diagnosis and treatment.

In recent years, two popular segmentation procedures have been used. These methods are Gaussian mixture modeling (GMM) and Fuzzy-C-means (FCM) known as soft segmentation algorithms. In the soft clustering, one piece of data can be including to two or more clusters in certain degree. The Fuzzy C means algorithm works by estimating the parameters which minimize distance of each voxel in the image to cluster centers according to a certain membership function. Whereas, the method based on the Gaussian mixture modeling (GMM) is a parametric probability density function represented as a weighted sum of Gaussian component densities. GMM parameters are estimated from training data using the iterative expectation-maximization (EM) [10]. These segmentation algorithms have been widely used recently in many studies, especially on biomedical image processing [2].

Ji et al. studied the generalized rough Fuzzy C means algorithm, weighted image patch-based FCM and modified possibilistic Fuzzy C means clustering algorithm for the segmentation of gray and white matter on MR brain images with various levels of noise [2], [11], [12]. Feng et al. were concentrated for SAR image segmentation using the non-local Fuzzy C-means algorithm with edge preservation [13]. Cai et al. proposed fast and robust Fuzzy C-means clustering algorithms on eight images in Matlab with mixed noise and MR brain images [14].

Forouzanfar et al. studied parameter optimization of the improved Fuzzy C means clustering algorithm for brain MR image segmentation using genetic algorithms (GA) and particle swarm optimization (PSO) in the case of noisy data on a synthetic square image and real T1-weighted MR image [1]. Enciso et al. studied a mixture of Gaussian functions with the parameters calculated using three nature inspired algorithms (particle swarm optimization, artificial bee colony optimization and differential evolution) on blood smear images [15]. Tang et al. studied a neighborhood weighted Gaussian mixture model in synthetic data and slice kidney CT image [10].

Merisaari et al. worked on watershed segmentation with Gaussian mixture model clustering for segmenting the cerebrospinal fluid from brain matter and other head tissues in premature infant brain MR images [16]. Qin et al. proposed a cloud model for four different images have multimodal histogram. They compared the results with related segmentation methods, including Otsu threshold, type-2 fuzzy threshold, Fuzzy C means clustering, and Gaussian mixture models. The index of misclassification error of their results competed well with FCM and GMM [17]. Ertekin et al. compared the Cavalieri method (point-counting) with the semi-automatic FCM algorithm to calculate the volumes of subcortical brain structures [18].

The aim of this study is to perform automatic segmentation of the CC by using Gaussian mixture modeling and Fuzzy C means algorithms for midsagittal section, and to compare these algorithms. In this study, the volume and other section areas of the CC were calculated using the Gaussian mixture model that estimated the maximum likelihood parameters by the EM algorithm and Fuzzy C means algorithm. For this purpose, the cross-section areas and volume of the CC, success of the segmentation were calculated by performing segmentation of the CC from simulated and real brain images. The quantitative and qualitative comparisons were performed with the results obtained from the two methods. With this study, an accurate and automatic segmentation system that allows opportunity for quantitative comparison to doctors in the planning of treatment and the diagnosis of diseases affecting the size of the CC was developed. This study can be adapted to perform segmentation on other regions of the brain, thus, it can be operated as practical use in the clinic.

Section snippets

Segmentation with Gaussian mixture modeling

Mixture models are very preferable in areas where the statistical modeling of data is needed, for instance in signal and image processing, pattern recognition, bioinformatics, and machine learning [19], [20]. Gaussian mixture modeling performs segmentation by extracting global statistics from Gaussian distributions of pixel intensity in image data set. The linear mixture of weighted sum of Gaussian distribution determines to how much is involved to which cluster of pixels. GMM is especially

Segmentation results with synthetic brain MR images

The simulated brain MR sample image was used from the BrainWeb [25] data base to compare the Gaussian mixture modeling and Fuzzy C means method. Brain Web provides T1, T2, and proton density (PD) weighted images and a variety of slice thicknesses, noise levels, and levels of intensity non-uniformity. In this study, a normal brain MR image (T1 modality, 1 mm slice thickness) was selected with 0% noise and intensity non-uniformity. The simulated image size was (181 × 217 × 181) voxels of (1 × 1 × 1) mm and

Discussion and conclusion

The segmentation process in brain images is aimed at classifying the tissues component of the brain, and quantifying the volume and other morphological, architectural features, and at separating different brain tissues to aid in various neurological and neurosurgical applications (neuropathological changes) [16], [17], [18]. The segmentation of brain images into the three main regions (GM, WM, CSF) is a fundamental step in the analysis of MR brain images. Many studies that used different

Conflict of interest statement

The author declares that there is no conflict of interest with any financial organization regarding the material discussed in the manuscript.

Acknowledgment

The author would like to thank Prof. Dr. Abdulhakim Coşkun head of Medical Faculty Radiology Department in Erciyes University for his help in recording of real brain images.

References (35)

M. Forouzanfar et al.
Parameter optimization of improved fuzzy c-means clustering algorithm for brain MR image segmentation
Engineering Applications of Artificial Intelligence
(2010)
Z. Ji et al.
A modified possibilistic fuzzy c-means clustering algorithm for bias field estimation and segmentation of brain MR image
Computerized Medical Imaging and Graphics
(2011)
F. Aboitiz et al.
Fiber composition of the human corpus callosum
Brain Research
(1992)
A.N. Francis et al.
Abnormalities of the corpus callosum in non-psychotic high-risk offspring of schizophrenia patients
Psychiatry Research: Neuroimaging
(2011)
M.B.D. Prigge et al.
Corpus callosum area in children and adults with autism
Research in Autism Spectrum Disorders
(2013)
R. O’Dwyer et al.
Differences in corpus callosum volume and diffusivity between temporal and frontal lobe epilepsy
Epilepsy & Behavior
(2010)
A.L. Choo et al.
Corpus callosum morphology in children who stutter
Journal of Communication Disorders
(2012)
M.H. Choi et al.
Difference between smokers and non-smokers in the corpus callosum volume
Neuroscience Letters
(2010)
H. Tang et al.
A vectorial image soft segmentation method based on neighborhood weighted Gaussian mixture model
Computerized Medical Imaging and Graphics
(2009)
Z. Ji et al.
Fuzzy c-means clustering with weighted image patch for image segmentation
Applied Soft Computing
(2012)

Z. Ji et al.

Generalized rough fuzzy c-means algorithm for brain MR image segmentation

Computer Methods and Programs in Biomedicine

(2012)

J. Feng et al.

Robust non-local fuzzy c-means algorithm with edge preservation for SAR image segmentation

Signal Processing

(2013)

W. Cai et al.

Fast and robust fuzzy c-means clustering algorithms incorporating local information for image segmentation

Pattern Recognition

(2007)

H. Merisaari et al.

Gaussian mixture model-based segmentation of MR images taken from premature infant brains

Journal of Neuroscience Methods

(2009)

K. Qin et al.

Image segmentation based on histogram analysis utilizing the cloud model

Computers and Mathematics with Applications

(2011)

T. Elguebaly et al.

Bayesian learning of finite generalized Gaussian mixture models on images

Signal Processing

(2011)

J. Liu et al.

Expectation-maximization algorithm with total variation regularization for vector-valued image segmentation

Journal of Visual Communication and Image Representation

(2012)

Cited by (27)

A Fuzzy-Based Duo-Secure Multi-Modal Framework for IoMT Anomaly Detection
2023, Journal of King Saud University - Computer and Information Sciences
With the advancement in the Internet of Medical Things (IoMT) infrastructure, network security issues have become a serious concern for hospitals and medical facilities. For this, a variety of customized network security tools and frameworks are used to distract several generalized attacks such as botnet-based distributed denial of services attacks (DDoS) and zero-day network attacks. Thus, it becomes difficult to operate routine IoMT services and tasks in between the under-attack scenario. This paper discusses a novel approach named Duo-Secure IoMT framework that uses multi-modal sensory signals’ data to differentiate the attack pattern and routine IoMT devices’ data. The proposed model uses a combination of two techniques such as dynamic Fuzzy C-Means clustering along with customized Bi-LSTM technique that processes sensory medical data securely along with identifying attack patterns within the IoMT network. As a case study, we are using a dataset to evaluate heart disease which consists of 36 attributes and 18940 instances. The performance evaluation shows that the proposed model evaluates a) prediction of heart issues and b) identification of network malware with an individual accuracy of 92.95% and multi-modal joint accuracy of 89.67% in the IoMT-based distributed network environment.
Measuring the hydration product proportion in composite cement paste by using quantitative BSE-EDS image analysis: A comparative study
2022, Measurement: Journal of the International Measurement Confederation
Composite cement has attracted increasing interest in the construction material field from the view of environmental and economic benefits. However, the measurement of hydration product proportion in the composite cementitious system remains challenging due to the heterogeneity of components and the existence of amorphous phases. In this study, three grey value histogram-based Backscattered Electron (BSE) image analysis approaches were compared mutually at first. Then, two color clustering-based Energy Dispersive X-ray Spectroscopy (EDS) image analysis approaches are comparatively developed for intuitively quantifying the hydration product proportion. In the BSE image analysis, results indicate that Overflow Method and Gaussian Mixture Model (GMM) show more reliable and accurate results than Frequency Histogram Method. In the EDS image analysis, GMM is better than K-means in terms of accuracy and reproducibility, and the cluster number of three sufficiently meets the requirement of grain extraction. Finally, the BSE-EDS image analysis, which integrates the merits of above approaches was proposed, achieving the best measurement performance. The hydration product proportions determined by BSE, EDS, and BSE-EDS image analysis approaches in the current research are 73.5%, 74.7%, and 68.1%, respectively. The proposed BSE-EDS image analysis offers a way to identify the hydration products practically, reliably, and visually in the cementitious system.
Computational methods for corpus callosum segmentation on MRI: A systematic literature review
2018, Computer Methods and Programs in Biomedicine
Citation Excerpt :
The metric and labels of voxels are then optimized in a graph-based semi-supervised learning model to achieve the segmentation result. Içer [83] proposed a comparative model of two clustering methods. Firstly, the Gaussian mixture model is applied by utilizing the weighted sum of Gaussian distributions through statistical decision procedures to define image classes.
Background and objective: The corpus callosum (CC) is the largest white matter structure in the brain and has a significant role in central nervous system diseases. Its volume correlates with the severity and/or extent of neurodegenerative disease. Even though the CC’s role has been extensively studied over the last decades, and different algorithms and methods have been published regarding CC segmentation and parcellation, no reviews or surveys covering such developments have been reported so far. To bridge this gap, this paper presents a systematic literature review of computational methods focusing on CC segmentation and parcellation acquired on magnetic resonance imaging.
Methods: IEEExplore, PubMed, EBSCO Host, and Scopus database were searched with the following search terms: ((Segmentation OR Parcellation) AND (Corpus Callosum) AND (DTI OR MRI OR Diffusion Tensor Imag* OR Diffusion Tractography OR Magnetic Resonance Imag*)), resulting in 802 publications. Two reviewers independently evaluated all articles and 36 studies were selected through the systematic literature review process.
Results: This work reviewed four main segmentation methods groups: model-based, region-based, thresholding, and machine learning; 32 different validity metrics were reported. Even though model-based techniques are the most recurrently used for the segmentation task (13 articles), machine learning approaches achieved better outcomes of 95% when analyzing mean values for segmentation and classification metrics results. Moreover, CC segmentation is better established in T₁-weighted images, having more methods implemented and also being tested in larger datasets, compared with diffusion tensor images.
Conclusions: The analyzed computational methods used to perform CC segmentation on magnetic resonance imaging have not yet overcome all presented challenges owing to metrics variability and lack of traceable materials.
A segmentation method based on HMRF for the aided diagnosis of acute myeloid leukemia
2017, Computer Methods and Programs in Biomedicine
The diagnosis of acute myeloid leukemia (AML) is purely dependent on counting the percentages of blasts (>20%) in the peripheral blood or bone marrow. Manual microscopic examination of peripheral blood or bone marrow aspirate smears is time consuming and less accurate. The first and very important step in blast recognition is the segmentation of the cells from the background for further cell feature extraction and cell classification. In this paper, we aimed to utilize computer technologies in image analysis and artificial intelligence to develop an automatic program for blast recognition and counting in the aspirate smears.
We proposed a method to analyze the aspirate smear images, which first performs segmentation of the cells by k-means cluster, then builds cell image representing model by HMRF (Hidden-Markov Random Field), estimates model parameters through probability of EM (expectation maximization), carries out convergence iteration until optimal value, and finally achieves second stage refined segmentation. Furthermore, the segmentation results are compared with several other methods using six classes of cells respectively.
The proposed method was applied to six groups of cells from 61 bone marrow aspirate images, and compared with other algorithms for its performance on the analysis of the whole images, the segmentation of nucleus, and the efficiency of calculation. It showed improved segmentation results in both the cropped images and the whole images, which provide the base for down-stream cell feature extraction and identification.
Segmentation of the aspirate smear images using the proposed method helps the analyst in differentiating six groups of cells and in the determination of blasts counting, which will be of great significance for the diagnosis of acute myeloid leukemia.
A Swarm Optimization approach for clinical knowledge mining
2015, Computer Methods and Programs in Biomedicine
Rule-based classification is a typical data mining task that is being used in several medical diagnosis and decision support systems. The rules stored in the rule base have an impact on classification efficiency. Rule sets that are extracted with data mining tools and techniques are optimized using heuristic or meta-heuristic approaches in order to improve the quality of the rule base. In this work, a meta-heuristic approach called Wind-driven Swarm Optimization (WSO) is used. The uniqueness of this work lies in the biological inspiration that underlies the algorithm.
WSO uses Jval, a new metric, to evaluate the efficiency of a rule-based classifier. Rules are extracted from decision trees. WSO is used to obtain different permutations and combinations of rules whereby the optimal ruleset that satisfies the requirement of the developer is used for predicting the test data. The performance of various extensions of decision trees, namely, RIPPER, PART, FURIA and Decision Tables are analyzed. The efficiency of WSO is also compared with the traditional Particle Swarm Optimization.
Experiments were carried out with six benchmark medical datasets. The traditional C4.5 algorithm yields 62.89% accuracy with 43 rules for liver disorders dataset where as WSO yields 64.60% with 19 rules. For Heart disease dataset, C4.5 is 68.64% accurate with 98 rules where as WSO is 77.8% accurate with 34 rules. The normalized standard deviation for accuracy of PSO and WSO are 0.5921 and 0.5846 respectively.
WSO provides accurate and concise rulesets. PSO yields results similar to that of WSO but the novelty of WSO lies in its biological motivation and it is customization for rule base optimization. The trade-off between the prediction accuracy and the size of the rule base is optimized during the design and development of rule-based clinical decision support system. The efficiency of a decision support system relies on the content of the rule base and classification accuracy.
Alpha-plane based automatic general type-2 fuzzy clustering based on simulated annealing meta-heuristic algorithm for analyzing gene expression data
2015, Computers in Biology and Medicine
This paper considers microarray gene expression data clustering using a novel two stage meta-heuristic algorithm based on the concept of α-planes in general type-2 fuzzy sets. The main aim of this research is to present a powerful data clustering approach capable of dealing with highly uncertain environments. In this regard, first, a new objective function using α-planes for general type-2 fuzzy c-means clustering algorithm is represented. Then, based on the philosophy of the meta-heuristic optimization framework ‘Simulated Annealing’, a two stage optimization algorithm is proposed. The first stage of the proposed approach is devoted to the annealing process accompanied by its proposed perturbation mechanisms. After termination of the first stage, its output is inserted to the second stage where it is checked with other possible local optima through a heuristic algorithm. The output of this stage is then re-entered to the first stage until no better solution is obtained. The proposed approach has been evaluated using several synthesized datasets and three microarray gene expression datasets. Extensive experiments demonstrate the capabilities of the proposed approach compared with some of the state-of-the-art techniques in the literature.

View all citing articles on Scopus

View full text

Automatic segmentation of corpus collasum using Gaussian mixture modeling and Fuzzy C means methods

Abstract

Introduction

Section snippets

Segmentation with Gaussian mixture modeling

Segmentation results with synthetic brain MR images

Discussion and conclusion

Conflict of interest statement

Acknowledgment

Engineering Applications of Artificial Intelligence

Computerized Medical Imaging and Graphics

Brain Research

Psychiatry Research: Neuroimaging

Research in Autism Spectrum Disorders

Epilepsy & Behavior

Journal of Communication Disorders

Neuroscience Letters

Computerized Medical Imaging and Graphics

Applied Soft Computing

Computer Methods and Programs in Biomedicine

Signal Processing

Pattern Recognition

Journal of Neuroscience Methods

Computers and Mathematics with Applications

Signal Processing

Journal of Visual Communication and Image Representation