Q

The study is about the medical aspects divided into three sections including radiology, brain and its sensitivity

Home, - Medical study

Section 1

 Studies have shown that radiologists can overlook the detection of a significant proportion of abnormalities in addition to obtaining high false positive rates. Pattern recognition in image processing requires the extraction of features from image regions along with the processing of these features with a pattern recognition algorithm. Since it is a crucial matter to segment the calcification and hemorrhage based on selected parameters, a process which simulates the selected parameters with fuzzy c-mean and adaptive neuro-fuzzy inference (ANFIS) was developed in this article. ANFIS is a specific form of the ANN family and shows very good learning and prediction capabilities, making it an efficient tool for dealing with uncertainties encountered in any system. For the purpose of early treatment with radiotherapy and surgery, the newly proposed AFCM is preferred to provide more information for medical images used by radiologist. Fuzzy clustering segmentation techniques provide useful information and good results.  AFCM method has better detection of abnormal tissues of calcification and hemorrhage. Overall, the newly proposed AFCM segmentation technique is recommended in medical image segmentation.This unsupervised segmentation algorithm can help radiologist to reduce the medical imaging noise effects originating from low resolution sensors and/or the structures that move during the data acquisition. They may be particularly helpful in the clinical field as an aid to the diagnosis of calcification and hemorrhage disease.

 

Besides acquiring excessiveand false positive rates, a considerable amount of deformities can be neglected by the radiotherapists as per the research studies. From the image areas, the mining of features in addition to their processing with a pattern detection algorithm is required within image processing. In this article, a process was developed which represents the chosen parameters with the ANFIS (adaptive neuro-fuzzy inference) and fuzzy c-meanas it is a critical matter to divide the hemorrhage and contractionon the basis of chosenfactors. ANFIS can be referred to as the extension of the ANN family and it displays excellent learning skills and estimatingcompetences, so it is believed to be a highly productive tool through which ambiguities in any system can be efficiently managed. For medical descriptions used by radiologist,the newly recommended AFCM is desired to offer more information in order to acknowledge early cure with surgery and radiation treatment. The desirable outcome and useful informationis most likely to be delivered by the fuzzy clustering segmentation methods. Theirregular tissues of hemorrhageand calcificationcan be better identified with the help of AFCM techniques. Generally, the medical image segmentation is the witness to the recentlyrecommended AFCM segmentation technique.The radiotherapist can reduce the medical imaging noise effects initiating from low resolution sensors or/and the movement of the structures during the data collection by applying this unsupervised segmentation algorithm. The medicaldiscipline is realizing the benefits of this system, such as the identification of hemorrhage and calcification syndrome.

 

Section 2

Brain is a very sensitive, complex and central part of the body. Overall functionality of the body is controlled by brain. Any damage in the brain affects the body badly. Brain is hidden from direct view by the protective skull. This skull gives brain protection from injuries and it hinders the study of its function. Brain can be affected by a hemorrhage which causes change in its normal structure and its normal behavior. Brain hemorrhage occurs due to the rupture of blood vessels and abnormal cells in the brain. This blood affects the normal functionality of the brain. Brain hemorrhage is the leading cause of solid tumor cancer death in human beings. In United States 22,070 are new estimated cases and 12,920 are death cases due to brain hemorrhage and other nervous system cancer in 2009 [28]. 

In the human body, pathological calcification means the deposition of calcium salts in necrotic tissue as well as the increase of calcified materials in local lesions. The calcified lesions on CT normally depicted as high density Hounsfield unit containing calcium, phosphorus, silicon, potassium, magnesium and zinc. For detecting intracranial calcifications, CT scanning is superior to routine magnetic resonance imaging (MRI) as it works principally on x-ray attenuation coefficient. The common MRI sequence such as the T-weighted (W) and T2W could not confidently define calcification’s appearance. The conventional spin echo T1W and T2W shows calcification in wide range of intensities[1, 2].  In gradient echo acquisition, calcification appears, as low signal intensity and this will give rise to confusion as blood may appears as low signal. Calcification appears in low intensity due to its low relative abundance of hydrogen proton in its compound. Calcification emits low signal and hence the intensity on image is low. However, Tawil et al. reported paradoxical hyper intense appearance of calcification in T1 weighted image. Tawil et al. [3] explained that the appearance might be due to the paramagnetic metals contents in the calcified tissue which caused the shortening of T1[3]. The contents of calcification vary and this explains the varying appearance of calcification in MRI.

 

The brain controls the whole functions of the body, since it is a very complicated, delicate and dominant part of the body. Moreover the body is badly affected by any injury to the brain. The defensive skull is likely to hide the brain from the direct view. Through the skull, the brain can be defended from woundsand it is likely to thwart the study of its utility. The hemorrhage can have adverse impacts on the brain by transforming its normal performance and its normal structure. When the abnormal cells and blood vessels are separated in the brain, it usually results in the function known as the brain hemorrhage and the normal functionality of the brain is also affected due to this blood. In human beings, the main reason of solid tumor cancer death is referred to as the brain hemorrhage. Reportedly, as a result of brain hemorrhage and other nervous system cancer, 22,070 new estimated cases and 12,920 death cases were registered in 2009 across United States [28]. 

Theadmission of calcium salts in necrotic tissue besides the rise of calcified materials in local wounds is described as the pathological calcification in human body. Across CT systems, generally the calcified woundsare illustrated as high density Hounsfield unit containing phosphorus, calcium, potassium, silicon, zinc and magnesium. As compared to routine MRI (magnetic resonance imaging), the CT scanning is better in spottingthe intracranial calcifications as it performs functioningmainly on x-ray attenuation coefficient. The calcification’s appearance could not be positively defined by the common MRI sequence for example the T2W and T-weighted (W). The chemical processis demonstrated in extensive range of intensities[1, 2] through the traditional spin echo T1W and T2W. The appearance of calcification is observedas low signal intensity in gradient echo acquisition and resultantly chaos would be created as if the blood is being displayed as a low signal. The low relative abundance of hydrogen proton in the compound of calcification is the result of its low intensity. Low signal is discharged by the calcification and therefore there is a low intensity on image. On the other hand, mysterious hyper extreme appearance of calcification was reported by Tawil et al.in T1 weighted image. According to Tawil et al. [3] the paramagnetic metal contents in the calcified material which yielded the shortening of T1[3] can be the result of such appearance. Fluctuation is found among the contents of calcification and resultantly the variable appearance of calcification in MRI is justified.

Intracranial hemorrhages (ICH) require rapid diagnosis and management, because it is a devastating disorder. The challenge in imaging is due to the fact that intracranial hemorrhage is a dynamic process with various imaging characteristics at the hyper acute, sub-acute and chronic stage [4] Diagnosis of brain hemorrhage and calcification is a very critical task because wrong diagnosis can lead to severe results. Surgery for brain hemorrhage is also very imperative errand because brain has a very intricate interconnected formation [5]. Each cell in the brain is bounded together in a very complex way. Internal structure of brain is very delicate and complicated. Normal laboratory test are inadequate to analyze the chemistry of brain  [6]. Positron Emission Tomography (PET), Computed tomography (CT) scans and Magnetic Resonance Imaging (MRI) are noninvasive medical imaging modalities used for visualizing calcification and hemorrhage. These medical imaging modalities help the doctors and researchers for analyzing the brain functionalities [7].

Since the ICH (Intracranial hemorrhages) is a shocking syndrome, therefore it needsquick diagnosis and management. Since the intracranial hemorrhage is a dynamic process containingseveral imaging attributes at the sub-acute, hyper acute and chronic stage, so the medical experts encounter challenges at times[4]It is very important to accuratelyspot the calcification and brain hemorrhage because we can have adverse impacts due to incorrect diagnosis. Likewise, brain is formed of a very complex and integrated structure, so its surgery also holds key significance[5]. There seems to be a very difficult connection of brain cells with each other. The brain is very complex and delicate from inner formation. The chemistry of brain cannot be examined through the general lab tests[6]. In order to visualize hemorrhage and calcification,the acute and broad-spectrum medical imaging modalities are among CT (Computed tomography) scans, PET (Positron Emission Tomography) and MRI (Magnetic Resonance Imaging). The researchers and surgeons can comprehensively scan the brain utilities through these medical imaging modalities [7].

In the medical field, physicians depend on different clinical frameworks, different anatomical evidences and different theoretical approaches, to diagnose calcification and hemorrhage. It is often impossible to establish rule-based systems[14]. MRI or computer-assisted approaches may be particularly helpful in the clinical field as a support in the diagnosis of hemorrhage and calcification disease in brain. For the purpose of differentiation and segmentation of calcification and hemorrhage, the FCM clustering algorithm was introduced to the diagnosis of every effected patient with hemorrhage and calcification. Recently, Wu and Yang [15] proposed a new algorithm called alternative fuzzy, c-means (AFCM) which provides more information available in medical images. In this paper, we use AFCM segmentation techniques for differentiation of calcification and hemorrhage in brain medical images. AFCM segmentation techniques provide useful information and good results. 

In orderto spothemorrhage and calcification,doctorsrely on different structural evidences, different clinical frameworks and diverseacademic approaches in the clinical domain. Establishing rule-based systems[14] often seems to be adifficult task. In the medicaldiscipline,calcification and hemorrhage syndrome in brain can be diagnosed through the MRI or computer based methodologies in a desirable way. The FCM clustering algorithm was introduced to the diagnosis of every effected patient with calcificationand hemorrhagefor the function of segregation, diagnosis and subdivisionof calcification and hemorrhage. In recent times, a new algorithm known asthe AFCM (alternative fuzzy, c-means) has been developed by Wu and Yang [15] through which comprehensive data can be acquired through the medical images. Forseparation of hemorrhage and calcification in brain medical images, the AFCM segmentation techniques would be used in this study. The suitable information and desirable results are delivered through the AFCM segmentation methods.

In clinical diagnosis, the segmentation of medical images is an important step. The clustering segmentation can be successfully used in the discrimination of different tissues from the medical images [16]. However, most medical images always present overlapping grayscale intensities for different tissues. MRI medical imaging uncertainties are widely presented in data because of the noise and blur in the acquisition and the partial volume effects originating from the low resolution of the sensors. In particular, borders between tissues are not clearly defined and memberships in the boundary regions are intrinsically fuzzy. The conventional (hard) clustering methods restrict each point of the data set to exactly one cluster. Fuzzy sets give the idea of uncertainty of belonging described by a membership function. Therefore, fuzzy clustering methods turn out to be particularly suitable for the segmentation of medical images. In this paper, we focus our attention on the fuzzy c-mean (FCM) and alternative fuzzy, c-mean (AFCM) methods and the resolution of MRI segmentation uncertainty in Ophthalmology cases. 

Amajor step within clinical diagnosis is the segmentation of medical images. From the medical images, the different tissues can be clearly judged by the successful application of clustering segmentation[16]. But, for different tissues,coinciding gray-scale intensities are always exhibited by most medical images. Due tothe partial volume effects initiating from the low resolution of the sensors and due to the blur and noise in the acquisition,there is a broad display of MRI medical imaging suspicionsacross the data. Specifically, there are inherently vague memberships found in the boundary regions and the borders among tissues are not evidently defined. Each point of the data set is limitedto just one clusterwithin the traditional (hard) clustering techniques. The idea of uncertainty of belonging described by a membership function is delivered by the fuzzy sets. Consequently, the segmentation of medical images can be suitably carried out through the fuzzy clustering methods. In this dissertation, we would mainly emphasize on the FCM (fuzzy c-mean) and AFCM (alternative fuzzy, c-mean) methods in addition to the resolution of MRI segmentation ambiguity in Ophthalmological scenarios.

Section 3

 For brain MRI tissue segmentation Cocosco et al. [17]  proposed an adaptive method. Pruning strategy is used for customizing the training set. By doing this, segmentation can accommodate the anatomical variability and pathology of brain MRI. Minimum spanning tree is used for reducing the incorrect label samples generated by prior tissue probability maps. These samples are then used by the KNN classifier for classification of the tissues from brain MR image. It cannot classify accurately tumorous tissues of the brain, which is its main drawback. For classification of brain images as normal and abnormal El-Syed et al. [18] proposed a hybrid technique. In this technique [18] features of the brain MR image are extracted using Discrete Wavelet Transform (DWT), which are then reduced using principal component analysis. The last step is the classification based on these features. Two types of classifiers are used for classification, one is feed forward back propagation neural network and other is K-nearest neighbors. Maximum 98.6% accuracy is achieved by [18].  Zhang et al. [19] proposed a Hidden Markov Random Field Model and the Expectation-Maximization algorithm for segmentation of brain MR images [19]. This is a fully automatic technique for brain MR image segmentation. This method is based on estimation of threshold that is heuristic in nature. Thus, most of the time, this method does not produce accurate results. The method in [19] is also computationally very expensive. For brain image segmentation Saha and Bandyopadhyay[20] proposed a technique which is based on fuzzy symmetry based genetic clustering method. In this technique  [20] fuzzy variable string length genetic point symmetry is used for clustering and then it is used for segmentation. Instead of Euclidean distance, point symmetry based distance is used for membership. The aim here is to determine the optimum fuzzy partitioning of the MR image. Maximum value of FSym-index best partitions the data. Genetic algo is used for evolving the number of clusters and evolving appropriate fuzzy clustering of the data. For measuring quality of cluster fuzzy point symmetry based cluster validity index is proposed in this paper. Experiments are performed on different T1, T2 and PD brain images. This technique performs better than FCM and Expectation Maximization algorithm. But this technique does not consider spatial information and sometime does not segment brain image correctly. This technique also does not work properly for the data sets which have the same point as a center for different clusters.

 

An adaptive method for brain MRI tissue segmentation has been proposed by Cocosco et al. [17]. The training set can be tailored by the pruning approach. The pathology of brain MRIand anatomical variabilitycan be adjustedin this wayby the segmentation method. The incorrect label samples delivered by previous tissue probability maps can be curtailed through minimum spanning tree. Subsequently, for grouping of the tissues from brain MR image,these samples are used by the KNN classifier. The major downside of this classifier is its inability to accurately classify tumorous tissues of the brain. A hybrid technique has been recommended by El-Syed et al. [18]for classification of brain images as normal and abnormal. The DWT (Discrete Wavelet Transform) method is used in this technique to extract the[18] features of the brain MR image, which are then condensedthrough principal component analysis. Classification is the final step based on these features. The classification basically includes two types of classifiers, one is referred to as the K-nearest neighbors and other is known as feed forward back propagation neural network. Through [18], you can maximum achieve 98.6% precision rate.  For subdivision of brain MR images [19], the Expectation-Maximization algorithm and a Hidden Markov Random Field Modelhas been suggested by Zhang et al. [19]. For brain MR image subdivision, this is a completelymechanical and automated technique. The threshold assessment is likely to drive this method and this assessment is experimental in nature. Therefore, accurate resultsare not delivered most of the time by this method. Numerically, the method in [19] is also very high priced. A technique based on fuzzy symmetry based genetic clustering methodhas been proposed by Saha and Bandyopadhyay [20]for brain image segmentation. For the purpose of clustering, the fuzzy variable string length genetic point symmetry is used in this technique [20] and afterwards it is applied for segmentationpurposes. For membership role, the point symmetry based distance is used rather than Euclidean distance. To determine the optimum fuzzy partitioning of the MR image is the basic goal at this point. The data is ideally separated by the maximum value of FSym-index. The number of clusters can be developed and suitable fuzzy clustering of the data can be progressed through the genetic algo. In this paper, the fuzzy point symmetry based cluster validity index is suggestedfordetermining quality of cluster. The different T1, T2 and PD brain images have witnessed certain experiments. This technique has better output than Expectation Maximization algorithm and the FCM. However, at times, this technique does not properlysegment brain image and spatial information is also not considered by this method. Also, the data sets containing the same point as a center for different clusters are not suitable for the application of this technique.

Ganesan and Sukanesh[21] proposed a fuzzy clustering method for segmentation of brain MR images. In this proposed technique feature of brain MR image is extracted by applying her wavelet transform of an image. When her wavelet is applied to an image, then approximate and detail coefficients are obtained. Approximate coefficients are taken as features of the brain MR image. These features are used for brain segmentation. Fuzzy C mean clustering algorithm is used for the classification purpose. FCM determines cluster centers for each cluster. FCM assigns membership values for each data point in the cluster. FCM iteratively modifies the cluster centers to get the right center within a dataset. Objective function minimization is the aim of FCM. Objective fuction represents the distance of any given data point from the cluster center. Cluster center is weighted by the membership value of the data point. Membership function matrix is able to classify the different objects in an image. This matrix is also used for the image reconstruction. Sobel edge detection method is applied for edge detection of the output image. Silhouette method is used for finding the strength of the clusters. Ganesan and Sukanesh  visually provide the result of their proposed technique but quantitatively they does not proved that their results are better. Neither have they compared their results with some other technique to validate their results [20]. 

In For division of brain MR images,a fuzzy clustering method was proposed by Ganesan and Sukanesh[21]. The feature of brain MR image is removedin this suggested method by applying her wavelet transform of an image. We can obtain the approximate and detail coefficients upon applying her wavelet to an image. Approximate coefficients are considered as attributes of the brain MR image. Then brain segmentation observes the application of these features. For the classification purpose,the experts use the Fuzzy C mean clustering algorithm. Afterwards, for each cluster, cluster centers are determined by the FCM. For each data point in the cluster, the membership values are assigned by the FCM. In order to acquire the right center within a dataset, the cluster centers are iteratively reformed by the FCM. The goal of the FCM is objective function minimization. From the cluster center, the distance of any given data point is represented by the objective function. The membership value of the data point is likely to weight the cluster center. Within an image, the different objects can be easily classified through the membership function matrix. We can also use this matrix for the image reconstruction. For edge detection of the output image,we can apply the sobel edge detection technique. The strength of the clusters can be revealed through silhouette method. The result of the proposed method was graphically illustrated by Ganesan and Sukanesh,however, they did not quantitatively prove that their results are enhanced. In addition, they also didn’t perform comparison of their outcomes with other methods to certify their results [20].

Li et al. proposed membership constraint FCM algorithm by incorporating spatial information for image segmentation [22]. This proposed algorithm is quite similar to conventional FCM algorithm but it gives improve results for MR images on different noise level as compared to original FCM algorithm because it considers spatial information of MR images as well. Ahmed and Mohamad[23] proposed segmentation of brain MR images for tumor extraction by combining K-means Clustering for grouping tissues and peronamalik anisotropic diffusion model for image enhancement. This method also yields good results for segmentation. Fuzzy c-means (FCM) clustering[9]  is an unsupervised system that has been effectively applied in fields such as astronomy, geology, medical imaging, target recognition, and image segmentation for clustering, classifier designs and feature analysis. Cluster Validity is the procedure of evaluating quantitatively the results of a clustering algorithm.  Wang and Wang [24] proposed modified FCM algorithm by modifying membership weighting of every cluster and integrating spatial information for brain MR image segmentation.  Wang and Wang [24]applied this method on different MR images and this technique gives appropriate results for MR images of different noise type. Dubey et al. [25] proposed a semi-automatic segmentation technique for brain tumor from MRI. Level set evolution with contrast propagation is used for segmentation in [25]. Pre and post contrast difference images and mixture modeling fit of the histogram are used for calculating probabilities for background and tumor region. For segmentation of the tumor boundaries, level set evolution is initialized using whole image. Ratan et al.[26]  proposed a method for brain tumor detection based on segmentation of the image. Segmentation of the brain MRI is done using watershed algorithm in the proposed technique. The method in  [26] does not require any initialization for the segmentation of the brain MRI. Cluster analysis is a tool for clustering a data set into groups of similar individuals. Image segmentation is a method to partition image pixels into similar regions. Thus, clustering algorithms would naturally be applied to enhance images in using segmentation areas. The conventional (hard) clustering methods restrict each point of the data set to exactly one cluster. Since Zadeh[8] proposed fuzzy sets which produced the idea of overlapping membership in two or more sets described by a membership function, fuzzy clustering has been widely studied and applied in various areas [9, 10]. In the fuzzy clustering literature, the FCM clustering algorithm is the most used method. Recently, this FCM algorithm has been more frequently used in segmenting the medical images. However, most FCM segmentation techniques are usually used in segmenting brain MRIs [11-13]. In this paper, we pay particular attention to the segmentation of hemorrhage and calcification in the brain images.

For image segmentation, a membership constraint FCM algorithm was developed by Li et al. by integrating spatial information [22]. There is uniqueness in this proposed algorithm and the traditional FCM algorithm, however, with this new technique; we acquire enhanced results for MR images along with their spatial data on different noise level as compared to original FCM algorithm. For tumor extraction,segmentation of brain MR imageshave been proposed by Ahmed and Mohamad [23] by integrating perona malik anisotropic diffusion model for image enhancement and K-means Clustering for grouping tissues. Desirable results are observed by applying this method for segmentation. Fuzzy c-means (FCM) clustering[9]  has been successfully applied in fields,for instance, geology,astronomy, target recognition, medical imaging,and image segmentation for clustering, feature analysis and classifier designs and FCM is also referred to as an unsupervised system. The procedure of quantitatively evaluatingthe results of a clustering algorithm is known as Cluster Validity.  By adjusting membership weightage of every cluster and integrating spatial data for brain MR image segmentation, a modified FCM algorithm was developed by Wang and Wang [24].  This method was applied by them on different MR images and suitable results are achievedfor MR images of different noise type by this technique. A semi-automatic segmentation technique was proposed by Dubey et al. [25]for brain tumor from MRI. For segmentation in [25], the specialists are likely to use level set evolution with contrast propagation. Fordetermininglikelihoods for background and tumor region,mixture modeling fit of the histogramin addition to the pre and post contrast difference images are used. By using whole image, the level set evolution is preparedfor segmentation of the tumor boundaries. On the basis of segmentation of the image, a method for brain tumor detection was proposed by Ratan et al.[26]. In the proposed technique, watershed algorithm is used to carry out the segmentation of the brain MRI. For the segmentation of the brain MRI, no initialization is required in the method [26]. A tool for grouping a data set into groups of analogous individuals is called cluster analysis. A method to split image pixels into similar regions is known as image segmentation. Therefore, images in segmentation areas can be greatly enhanced by the natural application of clustering algorithms. Each point of the data set is restricted to exactly one cluster within the standard (hard) clustering methods. In several areas, fuzzy clustering has been extensively studied and applied [9, 10]as Zadeh [8] proposed fuzzy sets, by which the idea of coinciding membership in two or more sets described by a membership function was produced. The FCM clustering algorithm is the generally appliedprocess across the fuzzy clustering literature. In recent times, the medical images are likely to be segmented through the application of this FCM algorithm. On the other hand, brain MRI is the area that mostly observes the FCM segmentation techniques [11-13]. The segmentation of calcification and hemorrhagein the brain images would be greatly emphasized in this paper.

 


Leave a comment


       
Captcha

Related :-