Segmentation Algorithms

We develop and evaluate several algorithms that segment a speech signal into subword units without using phone or orthographic transcripts. These segmentation algorithms rely on a scoring function, termed the local score, that is applied at the feature level and indicates where the characteristics of the audio signal change. The predominant approach in the literature to segmentation is to apply a threshold to the local score, and local maxima (peaks) that are above the threshold result in the hypothesis of a segment boundary. Scoring mechanisms of a select number of such algorithms are investigated, and it is found that these local scores frequently exhibit clusters of peaks near phoneme transitions that cause spurious segment boundaries. As a consequence, very short segments are sometimes postulated by the algorithms. To counteract this, ad-hoc remedies are proposed in the literature. We propose a dynamic programming (DP) framework for speech segmentation that employs a probabilist...

The selection of features is a crucial part of machine learning and data mining. The feature sets that are used for classification are always prone to having redundant and correlated features that can affect the performance. The goal of this study is to remove redundant and irrelevant features from the system and retain only relevant ones. This study presents Beetle-Swarm optimization process which involves selecting the features from a segmented image with a Random Forest classifier. The process is performed through a series of steps such as pre-processing, feature extraction, and feature classification. Two objective functions are used to perform the process: image entropy and accuracy function. The proposed method is evaluated on publicly available Kaggle brain tumor dataset. The results of the study revealed that the BSO+RF approach performed well compared to other techniques such as the PSO, ABC, and MVO. The proposed BSO+RF outperforms other similar algorithms in terms of accuracy. It has a performance of 0.8% compared to PSO, while it is slightly better than ABC, and slightly better than MVO. The performance of the proposed BSO+RF algorithm is also higher than that of the comparative techniques, with a learning percentage of 80. It has a low FDR value of less than PSO, ABC, and MVO, which suggests that it has better performance The proposed BSO-RF technique is more accurate than the existing algorithms when it comes to training and testing. In addition, it requires less features to achieve better accuracy. This results in faster computing time and more accuracy. This study presents a new approach to predict cancer using the combination of Beetle Swarm Optimization (BSO) and Random Forest. Beetle-swarm optimization is used to find threshold. This is used to segment the tumor from MR images resulting in better accuracy.

Segmentation algorithms for the quantitative description of the surface microstructure and the evaluation of the grain-boundary conductivity of ceramic materials surface using SEM microphotographs analysis was developed and applied in the present work for first time in barium cerate based solid solutions. To this purpose novel polycrystalline ceramic materials based on Cu and Gd co-doped BaCeO 3 proton conductors exhibiting high proton conductivity have been prepared and characterized. The influence of copper oxide as a dopant on the microstructure and the electrical properties of BaCe 0.9-x Gd 0.1 Cu x O 3-δ (0≤x≤0.1) have been examined in detail. It is shown that the dependence of the electrical conductivity is correlated with the average oxide grain diameter variation in air at various temperatures. Moreover, it was found that the algorithmic segmentation evaluation results are in good agreement with the experimental ones, verifying the considerable contribution of the grain-boundary conductivity to the electrical transport of Cu and Gd co-doped BaCeO 3 proton conductors.

People are dying these days from numerous deadliest diseases. One such illness is brain tumour, in which the unusual cells within the tumour quickly begin to damage the brain's healthy cells. Owing to this rapid growth, a person may pass away before the disease receives a correct diagnosis. Early disease detection is essential for any disease to help save the patient by providing them with better care. In a similar vein, a patient's life depends on early brain tumour detection. Brain tumour detection is an extremely challenging procedure that we would like to simplify in order to save time. The proposed model facilitates the quicker and more accurate identification of abnormal brain cells, leading to the early detection of brain tumours. In this work, an improved binomial thresholding-based segmentation (IBTBS) is introduced for segmentation purpose. From this segmented image, information theoretic based, wavelet transform (WT) based, and wavelet scattering transform (WST) based features are extracted. An optimization-based feature selection approach (OBFSA) is incorporated between feature selection and tumour classification in order to reduce the dimension of this retrieved feature. Finally, classification is performed using the Sparse Bayesian extreme learning machine (SBELM) classifier. The execution process of this proposed methodology takes an MRI image from the free accessible source. By computing and detecting four different parameters, the experimental analysis of the proposed approach displays the accuracy, specificity, and sensitivity values. This model can assist us in quickly diagnosing brain tumours, potentially saving the lives of patients.

Lung cancer remains a substantial global fatality; early detection is imperative for successful intervention and treatment. Deep learning (DL) models have shown promise in predicting lung cancer from medical images, but optimizing their parameters remains a challenging task. To improve prediction capability, this study introduces an approach by merging Particle Swarm Optimization and Bayesian Optimization (PSbBO) to optimize deep learning parameters. PSO provides an effective way for exploring the hyperparameter space, while Bayesian optimization provides a probabilistic framework for the effective evaluation and refining of a DL network. The simulation study showcases the effectiveness of the proposed model, achieving notable metrics for histopathological images, including an accuracy of 99.5%, precision of 98.3%, recall of 99.2%, F1-score of 99.4%, and an error rate of 1.19%. Furthermore, when applied to lung CT images, the proposed PSbBO demonstrates an accuracy of 98.8%, precision of 97.4%, recall of 98.3%, F1-score of 98.6%, and an error rate of 1.21%.

Nowadays heart diseases and their diagnosis have emerged as a prominent subject in health care systems, given that the heart performs a crucial role in the human body. Several computational techniques have been explored for the recognition and classification of cardiac diseases using Electrocardiogram (ECG) signals. Deep Learning (DL) is a present focus in healthcare solicitations, particularly in the classification of heartbeats in ECG signals. Many studies have utilized dissimilar DL models, including RNN (Recurrent Neural Networks), GRU (Gated Recurrent Unit), and CNN (Convolutional Neural Networks), to classify heartbeats using the MIT-BIH arrhythmia dataset. This article presents a methodical exploration of Bi-LSTM (Bi directional Long Short-Term Memory) based DL models for heartbeat classification using various quality metrics. Proposed variants include the Bi-LSTM model, demonstrating remarkable accuracy in classifying the heartbeats into five classes: Normal (N) beat, Supraventricular (S) beat, Ventricular contraction (V), Fusion beats (F), and Unclassifiable Beat (Q). The proposed technique outperforms present classifiers with Accuracy, Sensitivity, Specificity, and F1 score values of 98%, 96.9%, 97.4%, and 97.5% respectively. The simulations are conducted using MATLAB 2020a.

Nowadays heart diseases and their diagnosis have emerged as a prominent subject in health care systems, given that the heart performs a crucial role in the human body. Several computational techniques have been explored for the recognition and classification of cardiac diseases using Electrocardiogram (ECG) signals. Deep Learning (DL) is a present focus in healthcare solicitations, particularly in the classification of heartbeats in ECG signals. Many studies have utilized dissimilar DL models, including RNN (Recurrent Neural Networks), GRU (Gated Recurrent Unit), and CNN (Convolutional Neural Networks), to classify heartbeats using the MIT-BIH arrhythmia dataset. This article presents a methodical exploration of Bi-LSTM (Bi directional Long Short-Term Memory) based DL models for heartbeat classification using various quality metrics. Proposed variants include the Bi-LSTM model, demonstrating remarkable accuracy in classifying the heartbeats into five classes: Normal (N) beat, Supraventricular (S) beat, Ventricular contraction (V), Fusion beats (F), and Unclassifiable Beat (Q). The proposed technique outperforms present classifiers with Accuracy, Sensitivity, Specificity, and F1 score values of 98%, 96.9%, 97.4%, and 97.5% respectively. The simulations are conducted using MATLAB 2020a.

The emergence of deepfake technology has spurred the need for robust and adaptive methods to detect manipulated media content. This study explores the integration of the Integrate-backward-integrate (IbI) Logic Optimization Algorithm with Convolutional Neural Networks (CNNs) for enhanced deepfake detection. The proposed approach involves a multi-phase iterative process: the CNN initially trained on a diverse dataset encompassing both real and deepfake images. The CNN serves as the foundation for the IbI-driven optimization. The integration phase employs the trained CNN to forward-integrate images, classifying them as real or deepfake. Subsequently, the IbI Logic Optimization Algorithm engages in the backward phase, utilizing feedback from the CNN's performance to iteratively refine the network's parameters, architecture, and feature extraction capabilities. This iterative optimization process aims to adaptively enhance the CNN's ability to discern subtle nuances between authentic and manipulated visuals. The re-integration phase evaluates the refined CNN's performance through multiple iterations, seeking to iteratively improve deepfake detection accuracy. Validation occurs using separate datasets to prevent overfitting and ensure the model's generalizability. The proposed method aims to enhance the CNN's adaptability to evolving deepfake techniques, addressing the dynamic nature of manipulative media creation. This fusion of IbI Logic Optimization with CNNs presents a promising avenue for bolstering deepfake detection capabilities. However, the effectiveness of this approach relies on dataset quality, network architecture, and the dynamic nature of deepfake generation techniques. Continuous refinement and validation are essential to adapt the model to new challenges posed by advancing deepfake technologies.

Segmentation algorithms for the quantitative description of the surface microstructure and the evaluation of the grain-boundary conductivity of ceramic materials surface using SEM microphotographs analysis was developed and applied in the present work for first time in barium cerate based solid solutions. To this purpose novel polycrystalline ceramic materials based on Cu and Gd co-doped BaCeO 3 proton conductors exhibiting high proton conductivity have been prepared and characterized. The influence of copper oxide as a dopant on the microstructure and the electrical properties of BaCe 0.9-x Gd 0.1 Cu x O 3-δ (0≤x≤0.1) have been examined in detail. It is shown that the dependence of the electrical conductivity is correlated with the average oxide grain diameter variation in air at various temperatures. Moreover, it was found that the algorithmic segmentation evaluation results are in good agreement with the experimental ones, verifying the considerable contribution of the grain-boundary conductivity to the electrical transport of Cu and Gd co-doped BaCeO 3 proton conductors.

In this varying environment, a correct and appropriate disease diagnosis including early preclusion has never been more significant. Our study on disease identification of groundnut originated by Groundnut Bud Necrosis Virus will cover the way to the effective use of image processing approach in agriculture. The difficulty of capable plant disease protection is very much linked to the problems of sustainable agriculture and climate change. Due to the fast advancement of Artificial Intelligence, the work in this paper is primarily focused on applying Pattern Recognition based techniques. The purpose is to determine the grade of disease to control by developing a model for the selection of bud blight disease caused by GBNV in tomatoes. The images are classified according to the grade of the disease. Different methods have been applied to make a proper diagnosis by bringing clarity in the diagnostic results. Linear Vector Quantization works well than, Radial Basis Function, Back Propag...

Visually cancer is the abnormal pattern with predefined structure could be found in liver Computed Tomography (CT) images. Using deep convolution neural network computation and image processing, this detected abnormal pattern cluster can be classified in different liver issue types. Full size liver CT scan images consisting different body parts, and these are ultrasonic based gray scaled image construction. The primary challenge in the cancer symptoms detection process is to extract the liver area out of image then finding out the actual area of abnormality to conclude whether abnormality is cancer or any other issues on liver. This is two stage processes, first is to segment the abnormality area and second is to perform pattern matching to identify the abnormality. This research paper primarily focuses on different pre-processing techniques and stages involved in liver abnormality segmentation.

Analysis of intracranial neoplasm using multimodal MR images requires accurate and automatic segmentation. However, manually classifying tumors with similar structures or appearances in magnetic resonance imaging (MRI) with similar anatomy or appearances is more challenging, requiring experience to detect brain tumors. Precise segmentation of brain tumors gives clinicians with a foundation for surgical planning and treatment. Due to its capacity to segment brain tumor images automatically, Deep Neural Networks (DNN) have been widely used in image segmentation applications. To classify, segment and marking the occurrence of the brain tumor area accurately, we present custom Deep Convolution Neural Network (CNN) based Residual block U-Net (RB-ResUnet) architecture. Our technique is tested on publicly available Kaggle datasets utilizing quantitative metrics. Comparative results demonstrate that the custom CNN-based RB-ResUnet model can more reliably identify tumor locations and give accurate segmentation masks to tumor locations that are defined by bounding boxes. The findings of the experiment reveal that our proposed model RB-ResUnet can effectively aid in the identification, toxicological evaluation of the brain tumor and has clinical research as well as practical application.

The field of Printed Electronics (PE) is experiencing significant growth in the industrial sector and generating considerable interest across various industries due to its ability to produce intricate components. The functionality of printed electronic products heavily relies on the utilization of conductive ink during the printing process, which plays a vital role in developing flexible electronic circuits and improving the communicative functionalities of objects. Selecting the right ink for printing is crucial to meet consumer requirements. However, the conventional approach to this process has been manual, labor-intensive, and time-consuming, relying on the expertise of designers. This paper presents an automated ink selection model for printed circuits. This novel method has been incorporated with Multilayer Perceptron Neural Network (MLPNN) and Particle Swarm Optimization (PSO), named PSO-MLPNN. A dataset containing material features is generated by gathering information from both literature and experimental observations. To ensure uniformity, the data undergoes preprocessing using the min-max method, which scales all features to a standardized range between 0 and 1. A four-layer MLPNN is constructed to choose the most suitable ink. The network is trained with the PSO algorithm. The bias and weight values of MLPNN are tuned using the PSO algorithm to attain high accuracy. The computed findings confirm that the ink selection is highly effective and more accurate when compared to both the standard MLPNN.

Fast Computer-Aided Diagnostic Systems (CAD) have become instrumental in diagnosing diseases. Brain tumors, in particular, pose a significant health challenge. Traditional tumor detection methods relied on radiologists and biopsy, which are time-consuming and detrimental to patients. Early detection is crucial for effective treatment. This system leverages image processing, SWARM intelligence, and Support Vector Machines (SVMs) to detect and classify brain tumors swiftly and accurately. Image processing encompasses preprocessing, segmentation, and feature extraction, with the Particle Swarm Optimization (PSO) method optimizing feature selection. SVMs identify tumor types. While various techniques exist for tumor detection, none achieve 100% accuracy. This system is engineered to provide precise detection.

In the agricultural industry, plant diseases and pests pose the greatest risks. Lime is rich 10 source of vitamin C which works as an immunity booster in human body. Because of the late and manually diseases detection in lime causes a vast loss in crop production worldwide. The most common diseases are found in limes are lime canker, lemon scab, brown rot, sooty mould and Armillaria. In this paper we used imaging and non-imaging (spectral based sensing) methods with the combination of machine learning technique to detect the lime canker and sooty mould diseases. Image acquirement, pre-processing, segmentation and classification are all steps in the imaging methodology, which is then followed by feature extraction. In non-imaging methodology a multispectral sensor (Spectrometer) is used with 400 nm to 1000 nm wavelength to detect the diseases. training set and test set ratio is fixed for both techniques are 75% and 25% respectively. When it comes to identifying and classifying lime d...

Visually cancer is the abnormal pattern with predefined structure could be found in liver Computed Tomography (CT) images. Using deep convolution neural network computation and image processing, this detected abnormal pattern cluster can be classified in different liver issue types. Full size liver CT scan images consisting different body parts, and these are ultrasonic based gray scaled image construction. The primary challenge in the cancer symptoms detection process is to extract the liver area out of image then finding out the actual area of abnormality to conclude whether abnormality is cancer or any other issues on liver. This is two stage processes, first is to segment the abnormality area and second is to perform pattern matching to identify the abnormality. This research paper primarily focuses on different pre-processing techniques and stages involved in liver abnormality segmentation.

In this paper we incorporate recent results from AM-FM models for texture analysis into the variational model of image segmentation and examine the potential benefits of using the combination of these two approaches for texture segmentation. Using the Dominant Components Analysis (DCA) technique we obtain a low-dimensional, yet rich texture feature vector that proves to be useful for texture segmentation. We use an unsupervised scheme for texture segmentation, where only the number of regions is known apriori. Experimental results on both synthetic and challenging real-world images demonstrate the potential of the proposed combination.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Visually cancer is the abnormal pattern with predefined structure could be found in liver Computed Tomography (CT) images. Using deep convolution neural network computation and image processing, this detected abnormal pattern cluster can be classified in different liver issue types. Full size liver CT scan images consisting different body parts, and these are ultrasonic based gray scaled image construction. The primary challenge in the cancer symptoms detection process is to extract the liver area out of image then finding out the actual area of abnormality to conclude whether abnormality is cancer or any other issues on liver. This is two stage processes, first is to segment the abnormality area and second is to perform pattern matching to identify the abnormality. This research paper primarily focuses on different pre-processing techniques and stages involved in liver abnormality segmentation.

Segmentation algorithms for the quantitative description of the surface microstructure and the evaluation of the grain-boundary conductivity of ceramic materials surface using SEM microphotographs analysis was developed and applied in the present work for first time in barium cerate based solid solutions. To this purpose novel polycrystalline ceramic materials based on Cu and Gd co-doped BaCeO 3 proton conductors exhibiting high proton conductivity have been prepared and characterized. The influence of copper oxide as a dopant on the microstructure and the electrical properties of BaCe 0.9-x Gd 0.1 Cu x O 3-δ (0≤x≤0.1) have been examined in detail. It is shown that the dependence of the electrical conductivity is correlated with the average oxide grain diameter variation in air at various temperatures. Moreover, it was found that the algorithmic segmentation evaluation results are in good agreement with the experimental ones, verifying the considerable contribution of the grain-boundary conductivity to the electrical transport of Cu and Gd co-doped BaCeO 3 proton conductors.

Immediate treatment of a stroke can minimize long-term effects and even help reduce death risk. In the ischemic stroke cases, there are two zones of injury which are ischemic core and ischemic penumbra zone. The ischemic penumbra indicates the part that is located around the infarct core that is at risk of developing a brain infarction. Recently, various segmentation methods of infarct lesion from the MRI input images were developed and these methods gave a high accuracy in the extraction and detection of the infarct core. However, only some limited works have been reported to isolate the penumbra tissues and infarct core separately. The challenges exist in ischemic core identification are traditional approach prone to error, time-consuming and tedious for medical expert which could delay the treatment. In this paper, we study and analyse the segmentation algorithms for brain MRI ischemic of different categories. The focus of the review is mainly on the segmentation algorithms of infarct core with penumbra and infarct core only. We highlight the advantages and limitations alongside the discussion of the capabilities of these segmentation algorithms and its key challenges. The paper also devised a generic structure for automated stroke lesion segmentation. The performance of these algorithms was investigated by comparing different parameters of the surveyed algorithms. In addition, a new structure of the segmentation process for segmentation of penumbra is proposed by considering the challenges remains. The best accuracy for segmentation of infarct core and penumbra tissues is 82.1% whereas 99.1% for segmentation infarct core only. Meanwhile, the shortest average computational time recorded was 3.42 seconds for segmenting 10 slices of MR images. This paper presents an inclusive analysis of the discussed papers based on different categories of the segmentation algorithm. The proposed structure is important to enable a more robust and accurate assessment in clinical practice. This could be an opportunity for the medical and engineering sector to work together in designing a complete end-to-end automatic framework in detecting stroke lesion and penumbra.

Segmentation algorithms for the quantitative description of the surface microstructure and the evaluation of the grain-boundary conductivity of ceramic materials surface using SEM microphotographs analysis was developed and applied in the present work for first time in barium cerate based solid solutions. To this purpose novel polycrystalline ceramic materials based on Cu and Gd co-doped BaCeO 3 proton conductors exhibiting high proton conductivity have been prepared and characterized. The influence of copper oxide as a dopant on the microstructure and the electrical properties of BaCe 0.9-x Gd 0.1 Cu x O 3-δ (0≤x≤0.1) have been examined in detail. It is shown that the dependence of the electrical conductivity is correlated with the average oxide grain diameter variation in air at various temperatures. Moreover, it was found that the algorithmic segmentation evaluation results are in good agreement with the experimental ones, verifying the considerable contribution of the grain-boundary conductivity to the electrical transport of Cu and Gd co-doped BaCeO 3 proton conductors.

On account of the uncontrolled and quick growth of cells, Brain Tumor (BT) occurs. It may bring about death if not treated at an early phase. Brain Tumor Detection (BTD) has turned out to be a propitious research field in the current decennia. Precise segmentation along with classification sustains to be a difficult task in spite of several important efforts and propitious results in this field. The main complexity of BTD emerges from the change in tumor location, shape, along with size. Providing detailed literature on BTD via Magnetic Resonance Imaging (MRI) utilizing Machine Learning (ML) methods to aid the researchers is the goal of this review. Diverse datasets are mentioned which are utilized most often in the surveyed articles as a prime source of Brain Disease (BD) data. Furthermore, a concise epitome of diverse segmentation methods that are utilized in diagnosing BDs has been offered. Lastly, an outline of key outcomes from the surveyed articles is exhibited, and several ma...

In the analysis of brain Magnetic Resonance Images (MRI), classification of normality and abnormality is an important issue. Many works have been done to classify the brain MR images. This paper presents a new technique to classify the brain MRI images by using segmentation and KNN classifier. Initially, the brain MRI images obtained from brain databases are preprocessed using the Gaussian filter and the pre-processed images are normalized. Subsequently, the normalized images are subjected to segmentation by employing texture and intensity oriented region growing technique (TIORGW). Then texture features are extracted from the segmented brain MRI images. Later that, the well-known optimization algorithm called Genetic Algorithm (GA) is utilized to select the optimal texture features. Following that, the optimal features are passed in to KNN in order to classify whether the brain MRI image is normal or not. The proposed technique is implemented in the working platform of MATLAB and the performance is analysed by utilizing more number of brain MRI images.

On account of the uncontrolled and quick growth of cells, Brain Tumor (BT) occurs. It may bring about death if not treated at an early phase. Brain Tumor Detection (BTD) has turned out to be a propitious research field in the current decennia. Precise segmentation along with classification sustains to be a difficult task in spite of several important efforts and propitious results in this field. The main complexity of BTD emerges from the change in tumor location, shape, along with size. Providing detailed literature on BTD via Magnetic Resonance Imaging (MRI) utilizing Machine Learning (ML) methods to aid the researchers is the goal of this review. Diverse datasets are mentioned which are utilized most often in the surveyed articles as a prime source of Brain Disease (BD) data. Furthermore, a concise epitome of diverse segmentation methods that are utilized in diagnosing BDs has been offered. Lastly, an outline of key outcomes from the surveyed articles is exhibited, and several main problems related to ML-centred BD diagnostic methodologies are elucidated. The most precise method to detect diverse BDs can be engaged for future advancement via this study.