Brain MRI Segmentation Research Articles

Quality-driven deep active learning method for 3D brain MRI segmentation

Automatic segmentation of the brain Magnetic Resonance Imaging (MRI) plays a crucial role in many brain MRI processing algorithms, which is effective for the prevention, detection, monitoring, and treatment planning of brain disease. Currently, deep learning algorithms have shown outstanding performance in brain segmentation. Most algorithms train models with fully annotated brain MRI datasets. However, full annotation of 3D brain MRI is laborious and time-consuming. Using sparsely annotated datasets to train models may be one appropriate solution to reduce annotation cost. However, the approach of 3D dense segmentation with sparse annotation has not been fully studied. In this paper, we develop a segmentation framework combined with the quality-driven active learning (QDAL) module for suggestive annotation. In the proposed Active Learning module, attention mechanism and deep supervision mode are used to improve the segmentation accuracy and feedback segmentation quality information. Meanwhile, we observe a high correlation coefficient between the proposed two surrogate metrics and the real segmentation accuracy of per slice in one scan. We validate our framework on two public brain MRI datasets for brain region extraction and brain tissue segmentation. The comparative experiments demonstrate that the QDAL method outperforms the other four popular sampling strategies. The segmentation network with the guidance of the QDAL method only needs 15–20% annotated slices in brain extraction task, and 30–40% annotated slices in tissue segmentation task to achieve competitive results compared with training with full supervision.

Adaptive trapezoid region intercept histogram based Otsu method for brain MR image segmentation

In brain magnetic resonance (MR) image segmentation, the current Otsu method is often difficult to take both accuracy and anti-noise capability into consideration. So, in this paper, an adaptive trapezoid region intercept histogram based Otsu method is proposed. On the basis of bilateral filtering, the method uses Sigmoid function to identify the noise and adaptively calculate the weight of neighborhood pixel, and then constructs a 2D histogram of gray value-adaptive weight neighborhood gray mean to enhance the algorithm’s anti-noise capability and detail retention. The hierarchical threshold model is adopted: the macro-threshold T1 is determined by the trapezoid region intercept histogram based Otsu method, and the micro-threshold T2 is determined by the between-class variance criterion again in the trapezoid region corresponding to T1. The image is segmented by T2 to improve the accuracy of image segmentation. Based on the neighborhood information, an adaptive parameter l is designed to identify and correct noise, thus enhancing the universality of the algorithm. The experimental results show that the proposed method is effective and can be well applied to MR image segmentation.

Sparse Coding for Brain Tumor Segmentation Based on the Non-Linear Features

The main aim of brain Magnetic Resonance Image (MRI) segmentation is to extractthe significant objects like tumors for better diagnosis and proper treatment. As the brain tumors are distinct in the sense of shapes, location, and intensity it is hard to define a general algorithm for the tumor segmentation. Accurate extraction of tumors from the brain MRIs is a challenging task due to the complex anatomical structure of brain tissues in addition to the existence of intensity inhomogeneity, partial volume effects, and noise. In this paper, a method of Sparse coding based on non-linear features is proposed for the tumor segmentation from MR images of the brain. Initially, first and second-order statistical eigenvectors of 3 × 3 size are extracted from the MRIs then the process of Sparse coding is applied to them. The kernel dictionary learning algorithm is employed to obtain the non-linear features from these processed vectors to build two individual adaptive dictionaries for healthy and pathological tissues. This work proposes dictionary learning based kernel clustering algorithm to code the pixels, and then the target pixelsare classified by utilizing the method of linear discrimination. The proposed technique is applied to several tumor MRIs, taken from the BRATS database. This technique overcomes the problem of linear inseparability produced by the high level intensity similarity between the normal and abnormal tissues of the given brain image. All the performance parameters are high for the proposed technique. Comparison of the results with some other existing methods such as Fuzzy – C- Means (FCM), Hybrid k-Mean Graph Cut (HKMGC) and Neutrosophic Set – Expert Maximum Fuzzy Sure Entropy (NS-EMFSE) demonstrates that the proposed sparse coding method is effective in segmenting the brain tumor regions.

Multi-branch sharing network for real-time 3D brain tumor segmentation

Brain tumors are one of the most lethal diseases in the world. The segmentation of brain tumor is of great significance for physician in formulating appropriate diagnostic and treatment plans, not only accurate but also efficient 3D segmentation algorithms are urgently demanded in clinical practice. Nowadays, several 3D convolution neural networks have achieved impressive segmentation performance. However, these architectures come with extremely high computational overheads due to the extra depth dimensionality in 3D convolution, which may make these models prohibitive from practical large-scale clinic application. In this work, we aim at designing a more efficient and lightweight network without accuracy reduction for real-time segmentation of magnetic resonance images. To this end, we propose a multi-branch sharing network which consists of novel multi-branch sharing units. Different from other works, our proposed multi-branch sharing units focus the information sharing and communication between grouped layers by leveraging a Multiplexer operation, which can reduce the computational cost significantly while maintaining decent performance. Extensive experimental results on the BraTS2018 challenge dataset show that the proposed architecture achieve real-time inference while maintaining high accuracy for 3D brain magnetic resonance image segmentation.

BrainSeg-Net: Brain Tumor MR Image Segmentation via Enhanced Encoder-Decoder Network.

Efficient segmentation of Magnetic Resonance (MR) brain tumor images is of the utmost value for the diagnosis of tumor region. In recent years, advancement in the field of neural networks has been used to refine the segmentation performance of brain tumor sub-regions. The brain tumor segmentation has proven to be a complicated task even for neural networks, due to the small-scale tumor regions. These small-scale tumor regions are unable to be identified, the reason being their tiny size and the huge difference between area occupancy by different tumor classes. In previous state-of-the-art neural network models, the biggest problem was that the location information along with spatial details gets lost in deeper layers. To address these problems, we have proposed an encoder–decoder based model named BrainSeg-Net. The Feature Enhancer (FE) block is incorporated into the BrainSeg-Net architecture which extracts the middle-level features from low-level features from the shallow layers and shares them with the dense layers. This feature aggregation helps to achieve better performance of tumor identification. To address the problem associated with imbalance class, we have used a custom-designed loss function. For evaluation of BrainSeg-Net architecture, three benchmark datasets are utilized: BraTS2017, BraTS 2018, and BraTS 2019. Segmentation of Enhancing Core (EC), Whole Tumor (WT), and Tumor Core (TC) is carried out. The proposed architecture have exhibited good improvement when compared with existing baseline and state-of-the-art techniques. The MR brain tumor segmentation by BrainSeg-Net uses enhanced location and spatial features, which performs better than the existing plethora of brain MR image segmentation approaches.

MR Brain Image Segmentation to Detect White Matter, Gray Matter, and Cerebro Spinal Fluid Using TLBO Algorithm

This paper proposes an advanced and precise technique for the segmentation of Magnetic Resonance Image (MRI) of the brain. Brain MRI segmentation is to be familiar with the anatomical structure, to recognize the deformities, and to distinguish different tissues which help in treatment planning and diagnosis. Nature’s inspired population-based evolutionary algorithms are extremely popular for a wide range of applications due to their best solutions. Teaching Learning Based Optimization (TLBO) is an advanced population-based evolutionary algorithm designed based on Teaching and Learning process of a classroom. TLBO uses common controlling parameters and it won’t require algorithm-specific parameters. TLBO is more appropriate to optimize the real variables which are fuzzy valued, computationally efficient, and does not require parameter tuning. In this work, the pixels of the brain image are automatically grouped into three distinct homogeneous tissues such as White Matter (WM), Gray Matter (GM), and Cerebro Spinal Fluid (CSF) using the TLBO algorithm. The methodology includes skull stripping and filtering in the pre-processing stage. The outcomes for 10 MR brain images acquired by utilizing the proposed strategy proved that the three brain tissues are segmented accurately. The segmentation outputs are compared with the ground truth images and high values are obtained for the measure’s sensitivity, specificity, and segmentation accuracy. Four different approaches, namely Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Bacterial Foraging Algorithm (BFA), and Electromagnetic Optimization (EMO) are likewise implemented to compare with the results of the proposed methodology. From the results, it can be proved that the proposed method performed effectively than the other.

A stochastic approach for automated brain MRI segmentation

This paper presents an approach to segment lesions from brain magnetic resonance images in a fully automatic manner. The proposed idea leverages the strength of classical random walker algorithm and graph cut optimization technique in a single framework. We demonstrate that formulating a “prior” from a stochastic model can ameliorate the need of manually selected seed selection process in the random walker framework, thus making the algorithm fully automatic in a generic manner. By analytically solving a linear system of equations in the random walk process, initial labelling ∈ [0, 1] of each pixel in the image are computed to obtain the likelihood probability. These probabilities are then used to compute the likelihood for the data fidelity term in the energy function to compute the final segmentation, which is minimized by min cut max flow algorithm. Experimental results show the superiority of the proposed method over the state-of-the-art techniques on publicly available datasets.

Systematic volumetric analysis predicts response to CSF drainage and outcome to shunt surgery in idiopathic normal pressure hydrocephalus

ObjectivesIdiopathic normal pressure hydrocephalus (INPH) is a neurodegenerative disorder characterized by excess cerebrospinal fluid (CSF) in the ventricles, which can be diagnosed by invasive CSF drainage test and treated by shunt placement. Here, we aim to investigate the diagnostic and prognostic power of systematic volumetric analysis based on brain structural MRI for INPH.MethodsWe performed a retrospective study with a cohort of 104 probable INPH patients who underwent CSF drainage tests and another cohort of 41 INPH patients who had shunt placement. High-resolution T1-weighted images of the patients were segmented using an automated pipeline into 283 structures that are grouped into different granularity levels for volumetric analysis. Volumes at multi-granularity levels were used in a recursive feature elimination model to classify CSF drainage responders and non-responders. We then used pre-surgical brain volumes to predict Tinetti and MMSE scores after shunting, based on the least absolute shrinkage and selection operator.ResultsThe classification accuracy of differentiating the CSF drainage responders and non-responders increased as the granularity increased. The highest diagnostic accuracy was achieved at the finest segmentation with a sensitivity/specificity/precision/accuracy of 0.89/0.91/0.84/0.90 and an area under the curve of 0.94. The predicted post-surgical neurological scores showed high correlations with the ground truth, with r = 0.80 for Tinetti and r = 0.88 for MMSE. The anatomical features that played important roles in the diagnostic and prognostic tasks were also illustrated.ConclusionsWe demonstrated that volumetric analysis with fine segmentation could reliably differentiate CSF drainage responders from other INPH-like patients, and it could accurately predict the neurological outcomes after shunting.Key Points• We performed a fully automated segmentation of brain MRI at multiple granularity levels for systematic volumetric analysis of idiopathic normal pressure hydrocephalus (INPH) patients.• We were able to differentiate patients that responded to CSF drainage test with an accuracy of 0.90 and area under the curve of 0.94 in a cohort of 104 probable INPH patients, as well as to predict the post-shunt gait and cognitive scores with a coefficient of 0.80 for Tinetti and 0.88 for MMSE.• Feature analysis showed the inferior lateral ventricle, bilateral hippocampus, and orbital cortex are positive indicators of CSF drainage responders, whereas the posterior deep white matter and parietal subcortical white matter were negative predictors.

MRI brain segmentation using correlation based on adaptively regularised kernel-based fuzzy C-means clustering

Brain tumour detection is the most challenging aspect in the field of medical image processing and analysis. Whereas, the traditional techniques used in MRI brain tumour segmentation are extremely time consuming tasks. In this paper, the proposed approach contains two major steps: image acquisition and segmentation. Initially, the brain tumour detection was assessed by employing T1-weighted contrast enhanced magnetic resonance imaging (T1-WCEMRI) database. After image acquisition, segmentation was carried-out by using correlation based adaptively regularised kernel-based fuzzy C-means (ARKFCM) clustering along with Otsu thresholding. In conventional clustering methodologies, it was very hard to detect the ill-defined masses that highly decrease the segmentation accuracy. To address this concern, the kernel function in ARKFCM was replaced by a correlation function for localising the object in a complex template. In experimental analysis, the proposed approach distinguishes the normal brain region and brain tumour region by means of dice coefficient, Jaccard coefficient, true positive rate (TPR), false positive rate (FPR) and accuracy. The experimental outcome shows that the proposed methodology delivered average accuracy of 99.213% in brain tumour detection. The proposed methodology improved accuracy in brain tumour segmentation up to 3-3.2% compared to the existing methods: FCM and ARKFCM.

Evolution of Deep Learning Algorithms for MRI-Based Brain Tumor Image Segmentation.

Brain tumor textures are among the most challenging features for neuroradiologists to extract from magnetic resonance images (MRIs). Exceptionally high-grade tumors such as gliomas require quick and precise diagnosis and medical intervention due to their infiltrative and fast-spreading nature. Therefore, they require computer assistance instead of manual methods. Deep learning (DL) methods are currently on the rise and have become an active field of research in several domains varying from stock market analysis to deep space object detection. They have very promising potential in brain tumor feature extraction from MRIs. Convolutional neural network (CNN) architectures, one of the most influential families of DL algorithms, have undergone a profound transformation since their first successes. This has led to increasing feature extraction quality and algorithm generalizability over various brain tumor types and grades. This review paper presents an explanatory and comparative survey on MRI-based brain tumor image segmentation. First, it provides the survey background and the typical process chain for brain MRI segmentation using CNNs. Second, it details the typical CNN architecture structure and its advantages over other machine learning algorithms. CNN architectures proposed for this purpose are enumerated and classified corresponding to their complexity, and then compared using specific metrics that consider the datasets they use.

VNS Metaheuristic Based on Thresholding Functions for Brain MRI Segmentation

Image segmentation is a very crucial step in medical image analysis which is the first and the most important task in many clinical interventions. The authors propose in this paper to apply the variable neighborhood search (VNS) metaheuristic on the problem of brain magnetic resonance images (MRI) segmentation. In fact, by reviewing the literature, they notice that when the number of classes increases the computational time of the exhaustive methods grows exponentially with the number of required classes. That's why they exploit the VNS algorithm to optimize two maximizing thresholding functions which are the between-class variance (the Otsu's function) and the entropy thresholding (the Kapur's function). Thus, two versions of the VNS metaheuristic are respectively obtained: the VNS-Otsu and the VNS-Kapur. These two novel proposed thresholding methods are tested on a set of benchmark brain MRI to show their robustness and proficiency.

Segmentation of brain magnetic resonance images using a novel fuzzy clustering based method

Segmentation of tissues in brain magnetic resonance (MR) images has a crucial role in computer-aided diagnosis (CAD) of various brain diseases. However, due to the complex anatomical structure and the presence of intensity non-uniformity (INU) artefact, the segmentation of brain MR images is considered as a complicated task. In this study, the authors propose a novel locally influenced fuzzy C-means (LIFCM) clustering for segmentation of tissues in MR brain images. The proposed method incorporates local information in the clustering process to achieve accurate labelling of pixels. A novel local influence factor is proposed, which estimates the influence of a neighbouring pixel on the centre pixel. Furthermore, they have introduced the kernel-induced distance in LIFCM, which deals with complex brain MR data and produces effective segmentation. To evaluate the performance of the proposed method, they have used one simulated and one real MRI data set. Extensive experimental findings suggest that the authors' method not only produces effective segmentation but also retains crucial image details. The statistical significance test has been further conducted to support their experimental observations.

Quantitative Comparison of four brain MRI segmentation Techniques (Dept.E)

Magnetic resonance jmaging (MRI) is an advanced medical imaging technique providing rich information about the human soft tissue anatomy. The goal of brain magnetic resonance image segmentation is to accurately identify the principal tissue structures in these image volumes. There are many methods that exist to segment the brain. One of these, conventional methods that use pure image processing techniques are not preferred because they need human interaction for accurate and reliable segmentation. Unsupervised methods, on the other hand, do not require any human interference and can segment the brain with high precision, in the light to this fact, we in this paper compare the performance of our image segmentation techniques in the subject of brain MR image. Results show that fuzzy Kohonen's Competitive Learning Algorithms performs better in terms of segmentation accuracy, while FCM performs better in terms of speed of computation

Gray Matter Segmentation of Brain MRI Using Hybrid Enhanced Independent Component Analysis in Noisy and Noise Free Environment

Medical segmentation is the primary task performed to diagnosis the abnormalities in the human body. The brain is the complex organ and anatomical segmentation of brain tissues is a challenging task. In this paper, we used Enhanced Independent component analysis to perform the segmentation of gray matter. We used modified K means, Expected Maximization and Hidden Markov random field to provide better spatial correlation that overcomes in-homogeneity, noise and low contrast. Our objective is achieved in two steps (i) initially unwanted tissues are clipped from the MRI image using skull stripped Algorithm (ii) Enhanced Independent Component analysis is used to perform the segmentation of gray matter. We apply the proposed method on both T1w and T2w MRI to perform segmentation of gray matter at different noisy environments. We evaluate the the performance of our proposed system with Jaccard Index, Dice Coefficient and Accuracy. We further compared the proposed system performance with the existing frameworks. Our proposed method gives better segmentation of gray matter useful for diagnosis neurodegenerative disorders.

Inhomogeneous Brain Magnetic Resonance Images Segmentation Using a Novel Double Level Set Method

Magnetic resonance (MR) image segmentation plays an important role in the clinical diagnosis and pathological analysis of brain diseases, and has become a focus in the field of medical image processing. However, MR image segmentation is also a complex task because it is easily corrupted by inhomogeneous intensity and noise during the process of imaging. In this paper, we use double level set function to replace single level set of the data energy fitting model and propose a model based on Legendre polynomial and Heaviside function, which is used to segment brain magnetic resonance images. The double level set method (DLSM) can extract simultaneously the white matter (WM) and gray matter (GM) of brain tissue and ensure the robustness of level set initialization. Moreover, the bias field caused by intensity inhomogeneity is represented by a set of smooth basis functions, which can satisfy its property of slow variety. Finally, compared with the local intensity clustering model and multiplicative intrinsic component optimization model, both visual and objective results can prove the superior of the proposed DLSM model, and the computational speed is faster.

Vascular Risk Factors of Hippocampal Subfield Volumes in Persons without Dementia: The Medea 7T Study.

Background:Vascular risk factors have been associated with risk of Alzheimer’s disease (AD) and volume loss of the hippocampus, but the associations with subfields of the hippocampus are understudied. Knowing if vascular risk factors contribute to hippocampal subfield atrophy may improve our understanding of vascular contributions to neurodegenerative diseases.Objective:To investigate the associations between age, sex, and vascular risk factors with hippocampal subfields volumes on 7T MRI in older persons without dementia.Methods:From the Medea 7T study, 283 participants (67±9 years, 68% men) without dementia had 7T brain MRI and hippocampal subfield segmentation. Subfields were automatically segmented on the 3D T2-weighted 7T images with ASHS software. Using linear mixed models, we estimated adjusted associations of age, sex, and vascular risk factors with z-scores of volumes of the entorhinal cortex (ERC), subiculum (SUB), Cornu Ammonis (CA)1, CA2, CA3, CA4, and dentate gyrus (DG), and tail as multivariate correlated outcomes.Results:Increasing age was associated with smaller volumes in all subfields, except CA4/DG. Current smoking was associated with smaller ERC and SUB volumes; moderate alcohol use with smaller CA1 and CA4/DG, obesity with smaller volumes of ERC, SUB, CA2, CA3, and tail; and diabetes mellitus with smaller SUB volume. Sex, former smoking, and hypertension were not associated with subfield volumes. When formally tested, no risk factor affected the subfield volumes differentially.Conclusion:Several vascular risk factors were associated with smaller volumes of specific hippocampal subfields. However, no statistical evidence was found that subfields were differentially affected by these risk factors.

Automatic segmentation of brain MRI using a novel patch-wise U-net deep architecture.

Accurate segmentation of brain magnetic resonance imaging (MRI) is an essential step in quantifying the changes in brain structure. Deep learning in recent years has been extensively used for brain image segmentation with highly promising performance. In particular, the U-net architecture has been widely used for segmentation in various biomedical related fields. In this paper, we propose a patch-wise U-net architecture for the automatic segmentation of brain structures in structural MRI. In the proposed brain segmentation method, the non-overlapping patch-wise U-net is used to overcome the drawbacks of conventional U-net with more retention of local information. In our proposed method, the slices from an MRI scan are divided into non-overlapping patches that are fed into the U-net model along with their corresponding patches of ground truth so as to train the network. The experimental results show that the proposed patch-wise U-net model achieves a Dice similarity coefficient (DSC) score of 0.93 in average and outperforms the conventional U-net and the SegNet-based methods by 3% and 10%, respectively, for on Open Access Series of Imaging Studies (OASIS) and Internet Brain Segmentation Repository (IBSR) dataset.

A Cooperative Approach Based on Local Detection of Similarities and Discontinuities for Brain MR Images Segmentation.

This paper introduces a new cooperative multi-agent approach for segmenting brain Magnetic Resonance Images (MRIs). MRIs are manually processed by human radiology experts for the identification of many diseases and the monitoring of their evolution. However, such a task is time-consuming and depends on expert decision, which can be affected by many factors. Therefore, various types of research were and are still conducted to automate MRI processing, mainly MRI segmentation. The approach presented in this paper, without any parametrization or prior knowledge, uses a set of situated agents, locally interacting to segment images according to two main phases: the detection of discontinuities and the detection of similarities. An implementation of this approach was tested on phantom brain MR images to assess the results and prove its efficiency. Experimental results ensure a minimum of 89% Dice coefficient with increasing values of the noise and the intensity non-uniformity.

Redescending intuitionistic fuzzy clustering to brain magnetic resonance image segmentation

Redescending intuitionistic fuzzy clustering to brain magnetic resonance image segmentation

Brain Magnetic Resonance Imaging Segmentation Using Possibilistic Clustering Algorithm Combined with Context Constraints

Because the FCM method is simple and effective, a series of research results based on this method are widely used in medical image segmentation. Compared with the traditional FCM, the probability clustering (PCM) algorithm cancels the constraint on the normalization of each sample membership degree in the iterative process, and the clustering effect of the method is improved within a certain range. However, the above two methods only use the gray value of the image pixels in the iterative process, ignoring the context constraint relationship between the high-dimensional image pixels. The two are easily affected by image noise during the segmentation process, resulting in poor robustness, which will affect the segmentation accuracy in practical applications. In order to alleviate this problem, this paper introduces the context constraint information of image based on PCM, and proposes a PCM algorithm that combines context constraints (CCPCM) and successfully applies it to human brain MR image segmentation to further improve the noise immunity of the new algorithm. Expand the applicability of new algorithms in the medical field. Through simulation results on medical images, it is found that compared with the previous classical clustering methods, such as FCM, PCM, etc., the CCPCM has better anti-interference to different noises, and the segmentation boundary is clearer. At the same time, CCPCM algorithm introduces the spatial neighbor information adaptive weighting mechanism in the clustering process, which can adaptively adjust the constraint weight of spatial information and optimize the clustering process, thus improving the segmentation efficiency.