Search

Deep learning methods have shown promise for automated medical image analysis tasks. However, class imbalance is a common challenge that can negatively impact model performance, especially for tasks with minority classes that are clinically significant. This study aims to address this challenge through a novel hyperparameter optimization technique for training convolutional neural networks on imbalanced data. We developed a custom Convolutional Neural Network (CNN) architecture and introduced a Tangent Optimization Algorithm (TOA) based on the trigonometric properties of the tangent function. The TOA optimizes hyperparameters during training without requiring data preprocessing or augmentation steps. We applied our approach to classifying B-mode ultrasound carotid artery plaque images as symptomatic or asymptomatic using a dataset with significant class imbalance. On k-fold cross-validation, our method achieved an average accuracy of 98.82%, a sensitivity of 99.41%, and a specificity of 95.74%. The proposed optimization technique provides a computationally efficient and interpretable solution for training deep learning models on unbalanced medical image datasets.

The objective of this paper is to develop a novel emotion recognition system from electroencephalogram (EEG) signals using effective connectivity and deep learning methods. Emotion recognition is an important task for various applications such as human-computer interaction and, mental health diagnosis. The paper aims to improve the accuracy and robustness of emotion recognition by combining different effective connectivity (EC) methods and pre-trained convolutional neural networks (CNNs), as well as long short-term memory (LSTM). EC methods measure information flow in the brain during emotional states using EEG signals. We used three EC methods: transfer entropy (TE), partial directed coherence (PDC), and direct directed transfer function (dDTF). We estimated a fused image from these methods for each five-second window of 32-channel EEG signals. Then, we applied six pre-trained CNNs to classify the images into four emotion classes based on the two-dimensional valence-arousal model. We used the leave-one-subject-out cross-validation strategy to evaluate the classification results. We also used an ensemble model to select the best results from the best pre-trained CNNs using the majority voting approach. Moreover, we combined the CNNs with LSTM to improve recognition performance. We achieved the average accuracy and F-score of 98.76%, 98.86%, 98.66 and 98.88% for classifying emotions using DEAP and MAHNOB-HCI datasets, respectively. Our results show that fused images can increase the accuracy and that an ensemble and combination of pre-trained CNNs and LSTM can achieve high accuracy for automated emotion recognition. Our model outperformed other state-of-the-art systems using the same datasets for four-class emotion classification.

The current work proposes an efficient assessment of hypertension (HTN) using a Directional-Guided Motion Sensitive (DGMS) descriptor and Machine Learning (ML) algorithm. The main objective of the proposed work is to automate the detection of HTN using ultrasound (US) images. The four-chamber US images from 70 healthy subjects and 70 HTN patients are collected. A novel pipelined architecture has been developed in two stages with four phases: preprocessing, feature extraction using DGMS descriptor, feature ranking and selection, and classification using shallow K-Nearest Neighbor classifier. The proposed model has achieved a classification accuracy of 98% using a set of prominent features, predominating the performance attained by other approaches. This study suggests US contains predictive signals even when standard measures are normal and lays the groundwork for artificial intelligence-assisted cardiac assessment to aid quicker, more objective diagnosis and earlier treatment. If further validated on additional diverse patient data, the technology could be integrated into clinics to enhance HTN detection through automated, early discernment of subtle manifestations missed by human eyes and traditional metrics.

Electroencephalogram (EEG) signals contain information about the brain’s state as they reflect the brain’s functioning. However, the manual interpretation of EEG signals is tedious and time-consuming. Therefore, automatic EEG translation models need to be proposed using machine learning methods. In this study, we proposed an innovative method to achieve high classification performance with explainable results. We introduce channel-based transformation, a channel pattern (ChannelPat), the t algorithm, and Lobish (a symbolic language). By using channel-based transformation, EEG signals were encoded using the index of the channels. The proposed ChannelPat feature extractor encoded the transition between two channels and served as a histogram-based feature extractor. An iterative neighborhood component analysis (INCA) feature selector was employed to select the most informative features, and the selected features were fed into a new ensemble k-nearest neighbor (tkNN) classifier. To evaluate the classification capability of the proposed channel-based EEG language detection model, a new EEG language dataset comprising Arabic and Turkish was collected. Additionally, Lobish was introduced to obtain explainable outcomes from the proposed EEG language detection model. The proposed channel-based feature engineering model was applied to the collected EEG language dataset, achieving a classification accuracy of 98.59%. Lobish extracted meaningful information from the cortex of the brain for language detection.

Nowadays, the forward-forward (FF) algorithm is very popular in the machine learning society, and it uses a square-based activation function. In this research, we inspired the FF algorithm and presented a new kernel for a local binary pattern named square-kernelled local binary pattern (SKLBP). By deploying the proposed one-dimensional SKLBP, a new feature engineering model has been presented. To measure the classification ability of the proposed SKLBP-based model, we have collected a new textural environmental sound classification (ESC) dataset. The collected dataset is a balanced dataset, and it contains 15 classes. There are 100 sounds in each class. Our proposed model has mimicked the deep learning structure. Therefore, it uses multileveled feature extraction methodology by using discrete wavelet transform. The features generated have been considered as input for the iterative feature selector. The chosen feature vector has been utilized as input of the k nearest neighbor classifier. The proposed SKLBP-based signal classification model reached 94% classification accuracy. In this aspect, we contributed to the ESC methodology by collecting the new textural ESC dataset and proposing the SKLBP-based ESC model.

Pulmonary hypertension (PH) is a chronic and progressive disease. We introduced a novel automated self-organized feature engineering architecture for PH detection, which was trained and refined using a new thoracic CT image dataset. This study’s dataset includes 807 transverse contrast-enhanced CT images from 313 patients, categorized into four groups: Group 1 with 20 mmHg $\le $ mean pulmonary artery pressure (mPAP) < 25 mmHg; Group 2 with 25 mmHg $\le $ mPAP $\le30$ mmHg; Group 3 where mPAP > 30 mmHg; and a control group with no PH. Our model consists of four primary stages: (i) generation of features based on combinations from nested patches, (ii) feature selection, (iii) classification and (iv) majority voting. CT images were segmented into nested patches, each being processed through pretrained EfficientNetB0 and DenseNet201 to derive four deep feature vectors, utilizing both the global average pooling and fully connected layers of these networks. These four extracted features underwent combinatorial operations, resulting in 15 feature vectors. Subsequently, these vectors were introduced to neighborhood component analysis, ReliefF, and Chi2 feature selectors. This process yielded 45 refined feature vectors with diminished data dimensions. These selected vectors were then processed through a support vector machine and k-nearest neighbors classifiers, producing 90 predictive vectors. By applying mode-based iterative majority voting to these vectors, an additional 88 voted prediction vectors were generated, leading to a total of 178 classifier-generated and voted prediction vectors. The optimal classification result was selected from these 178 vectors. With the use of 10-fold cross-validation, our model achieved a remarkable 97.27% overall accuracy for the 4-class classification on the study dataset. Owing to its reduced time complexity, this model is practical for CT-based PH screenings.

Lung cancer is the leading cause of cancer death, and early diagnosis is associated with a positive prognosis. Chest X-ray (CXR) provides an inexpensive imaging mode for lung cancer diagnosis. Computer vision algorithms have previously been proposed to assist human radiologists in this task; however, leading studies use down-sampled images and computationally expensive methods with unproven generalization. In contrast, this study localizes lung nodules from CXR images using efficient encoder-decoder neural networks that have been crafted to process full resolution input images, thereby avoiding signal loss resulting from down-sampling. Encoder-decoder networks are trained and tested using the Japanese Society of Radiological Technology dataset. The networks are used to localize lung nodules from an independent CXR dataset. These experiments allow for the determination of the optimal network depth, image resolution, and pre-processing pipeline for generalized lung nodule localization. We find that more subtle nodules are detected in earlier training epochs. Therefore, we propose a novel self-ensemble model from three consecutive epochs centered on the validation optimum. This ensemble achieved a sensitivity of 85% in 10-fold internal testing with false positives of 8 per image. A sensitivity of 81% is achieved at a false positive rate of 6 following morphological false positive reduction. This result is comparable to more computationally complex systems, but with a sub-second inference time that is faster than other methods presented in the literature. The proposed algorithm achieved excellent generalization results against a challenging external dataset with a sensitivity of 77% at a false positive rate of 7.6.

Intracranial hematoma due to traumatic brain injury is a serious health concern with rates of morbidity and mortality that are increasing worldwide. Manual identification is slow, subject to observer variabilities, and the existing automated techniques for intracranial hematoma detection in non-contrast computed tomography images cannot effectively detect multiple lesions of irregular sizes and shapes. Therefore, a computer-aided system using different window settings, YOLOv5s, cascaded attention module, and spatial pyramid pooling-fast is proposed to detect hematoma types, namely acute intraparenchymal, intraventricular, subdural, epidural, subarachnoid, and chronic subdural. Firstly, the computed tomography images are pre-processed using a window-based stacking approach wherein a three-channel image is generated by stacking grayscale images obtained with the aid of multiple window settings, i.e, brain, bone, and subdural. Secondly, a cascaded attention module is constructed in the neck of the YOLOv5s model to improve its detection performance by placing the convolution block attention module in serial with the efficient channel attention module. The cascaded attention module enriches the feature representation of various hematoma types in complex backgrounds especially when they are small and inconspicuous. The spatial pyramid pooling is replaced by a spatial pyramid pooling-fast to reduce the computational parameters and accelerate the feature fusion ability. The proposed deep learning model is trained, validated, and tested with 15,921 images from the brain haemorrhage extended dataset and it achieved overall precision, recall, F1-score, and mean average precision at 0.5, and mean average precision at 0.5:0.95 of 0.935, 0.908, 0.921, 0.943 and 0.65 respectively. The experimental results show that in comparison to the original YOLOv5s model and state-of-the-art methods, the model was able to localize and classify the acute or chronic instances of five hematoma subtypes in an individual image with improved precision and recall values. Hence the proposed system can be used in hospitals for the early and accurate detection of hematoma.

Sleep is an essential criterion for health. However, sleep disorders degrade the sleep quality. Hence, to diagnose sleep disorders, sleep monitoring is crucial. The cyclic alternating patterns (CAP) phases describe the sleep quality. However, CAP detection is a time-consuming, hectic, and uncertain process. Therefore, an automatic detection of CAP phases is necessary. This study proposes a hierarchical approach to identify sleep disorders and classify CAP phases. Single-channel EEG recording provided by the CAP sleep database has been utilized in this study. The proposed approach classifies CAP sequence into healthy or unhealthy. Further, it identifies sleep disorder of unhealthy sequence among periodic leg movement (PLM), rapid eye movement behaviour disorder (RBD), nocturnal frontal lobe epilepsy (NFLE), narcolepsy (NARCO), and insomnia (INS). Further using our prior work, the CAP phase of the sequence can be identified. The best model was obtained by long short-term memory (LSTM) along with convolutional neural network (CNN) for healthy-unhealthy, and disease classification with an accuracy of 91.45% and 90.55%, respectively. The same models gave an accuracy of 92.79% for healthy-unhealthy and 93.31% for disease classification when evaluated using dataset of only phase B, highlighting the importance of phase B for identifying sleep disorders.

One of the foremost causes of death in males worldwide is prostate cancer. The identification, detection and diagnosis of the same is very crucial in saving lives. In this paper, we present an efficient gland segmentation model using digital histopathology and deep learning. These methods have the potential to revolutionize medicine by identifying hidden patterns within the image. The recent improvements in data acquisition, processing and analysis of Deep Learning Models has made Artificial Intelligence driven healthcare a very lucrative area, in terms of data inference and delivering meaningful insights. This study presents an automated method for segmenting histopathological images of human prostate glands. The main focus is developing new methods for segmenting histopathological images of prostate gland using a multi-channel algorithm with an attention mechanism to detect important areas. We compare our results with a host of contemporary techniques and show that our method performs better at the segmentation task for histopathological imagery. Our method is able to delineate gland and background parts with an average Dice-coefficient of 0.9168. In this attention-based model we propose for semantic segmentation of prostate glands the potential to provide accurate segmentation versus tumor features, which has significant implications for medical screening applications.

Most emotion classification schemes to date have concentrated on individual inputs rather than crowd-level signals. In addressing this gap, we introduce Sound-based Community Emotion Recognition (SCED) as a fresh challenge in the machine learning domain. In this pursuit, we crafted the FF-BTP-based feature engineering model inspired by deep learning principles, specifically designed for discerning crowd sentiments. Our unique dataset was derived from 187 YouTube videos, summing up to 2733 segments each of 3 seconds (sampled at 44.1 KHz). These segments, capturing overlapping speech, ambient sounds, and more, were meticulously categorized into negative, neutral, and positive emotional content. Our architectural design fuses the BTP, a textural feature extractor, and an innovative handcrafted feature selector inspired by Hinton’s FF algorithm. This combination identifies the most salient feature vector using calculated mean square error. Further enhancements include the incorporation of a multilevel discrete wavelet transform for spatial and frequency domain feature extraction, and a sophisticated iterative neighborhood component analysis for feature selection, eventually employing a support vector machine for classification. On testing, our FF-BTP model showcased an impressive 97.22% classification accuracy across three categories using the SCED dataset. This handcrafted approach, although inspired by deep learning’s feature analysis depth, requires significantly lower computational resources and still delivers outstanding results. It holds promise for future SCED-centric applications.

Fibromyalgia is a chronic pain syndrome associated with sleep disturbances, which may manifest as altered electroencephalography and electrocardiography (ECG) signal alterations during sleep. We aimed to develop a lightweight machine learning model for diagnosing fibromyalgia using single-lead ECG signals recorded during sleep. We analyzed 139 single-lead ECGs recorded during Stage 2 and Sleep Stage 3 of 16 patients with fibromyalgia and 16 age and sex matched controls. ECG records were divided into 15-second segments: 3308 and 1783 in healthy vs fibromyalgia classes, respectively. Our model comprised (1) feature extraction that combined an 8-wavelet filter and 4-level multiple filters-based multilevel discrete wavelet transform signal decomposition with a novel local binary pattern (LBP)-like function, 3LBP, that generated multiple patterns (analogous to quantum superposition) for feature map value extraction (the optimal input-specific pattern was dynamically selected using a novel forward-forward algorithm); (2) feature selection using neighborhood component analysis and Chi-square functions; (3) classification with k-nearest neighbors and support vector machine classifiers using leave-one-record-out cross-validation; and (4) mode function-based iterative majority voting to generate voted results, from which the best model result was derived. Our model attained binary classification accuracies of 93.87% and 92.02% for Sleep Stage 2 and Sleep Stage 3, respectively. The observed outcomes and empirical evidence unequivocally demonstrate the efficacy of our proposed methodology in differentiating the electrocardiographic signatures of fibromyalgia patients from control subjects. The model exhibited self-organizational properties and computational efficiency, rendering it amenable to facile clinical integration.

Mental fatigue is a psychophysical condition with a significant adverse effect on daily life, compromising both physical and mental wellness. We are experiencing challenges in this fast-changing environment, and mental fatigue problems are becoming more prominent. This demands an urgent need to explore an effective and accurate automated system for timely mental fatigue detection. Therefore, we present a systematic review of brain-computer interface (BCI) studies for mental fatigue detection using artificial intelligent (AI) techniques published in Scopus, IEEE Explore, PubMed and Web of Science (WOS) between 2011 and 2022. The Boolean search expression that comprised (((ELECTROENCEPHALOGRAM) AND (BCI)) AND (FATIGUE CLASSIFICATION)) AND (BRAIN-COMPUTER INTERFACE) has been used to select the articles. Through the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) methodology, we selected 39 out of 562 articles. Our review identified the research gap in employing BCI for mental fatigue intervention through automated neurofeedback. The AI techniques employed to develop EEG-based mental fatigue detection are also discussed. We have presented comprehensive challenges and future recommendations from the gaps identified in discussions. The future direction includes data fusion, hybrid classification models, availability of public datasets, uncertainty, explainability, and hardware implementation strategies.

An automated Neurological Disorder detection system can be considered as a cost-effective and resource efficient tool for medical and healthcare applications. In automated Neurological Disorder detection, electroencephalograms are commonly used, but their low signal intensity and nonlinear features are difficult to analyze visually. A promising approach for processing of electroencephalogram signals is the concept of entropy, a nonlinear signal processing method to measure the chaos in the signal. The aim of this study was to find out the effective entropy measures and the machine learning approaches that produced promising output. Using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines as our method, we have identified 84 studies published between 2012 and 2022 that has investigated epilepsy, Parkinson’s disease, autism, Attention Deficit Hyperactive disorder, schizophrenia, Alzheimer’s disease, depression, and alcohol use disorder with machine learning approaches considering entropy measures. We show that Support Vector Machines was the most commonly used machine learning model, with consistent performance in most of the studies whereas sample entropy was the most commonly used entropy measure, followed by the approximate entropy. For epilepsy detection, the most used entropy feature was the log energy entropy, whereas the multi-scale entropy was commonly used for Alzheimer’s Disease, approximate and sample entropy used for Parkinson’s Disease, multi scale and Shannon entropy applied for autism, approximate and Shannon entropy used for attention deficit hyperactive disorder, sample entropy used for depression, approximate and spectral entropy adopted for schizophrenia, and the approximate and sample entropy employed for alcohol use disorder. According to the majority of the studies, there is growing concern about the increase in neuro patients and the heavy resource burden that is associated with their prevalence and diagnosis. Based on these studies, we conclude that Computer-Aided Design systems would be economically advantageous in detecting Neurological Disorders. To incorporate Computer-Aided Design system into the mainstream health care system, future research could focus on multi-modal approaches to the disorder and its interpretation and explanation. We believe this is the first review that has combined the electroencephalograms, entropy, and automated detection possibility of the 8 distinct neurological disorders. The study is limited to the papers that used accuracy as their performance evaluation metric. The findings and synthesis of previous studies provides a clear pathway that identifies the entropy approach as a practical solution for automated detection of neurological disorder using electroencephalograms with potential applications in other kinds of signal analysis.

Sleep is a crucial component of health and well-being. It maintains the metabolism of the body and covers one-third of total life. The assessment of sleep quality is typically done by evaluating the macrostructure-based sleep stages, however, it does not take into account transient phenomena like K-complexes and transient fluctuations, which are crucial for the diagnosis of various sleep disorders. Cyclic alternating pattern (CAP) is a recurrent physiological electroencephalogram (EEG) activity that takes place in the brain during sleep and it is considered as a microstructure of sleep that can provide more accurate and relevant evaluation of sleep. The traditional way of CAP phase division is done manually by sleep specialists, which is sensitive, time-consuming, and prone to inaccuracies. Hence, there is a need for automated detection techniques that can solve the problems. This study proposes an automated, computerized approach for developing a machine learning model with explainable artificial intelligence (XAI) capabilities, using wavelet-based Hjorth parameters for classifying CAP A & B phases and phases A sub-phases (A1, A2, A3). The study utilizes SHAP (SHapley Additive exPlanations)-based feature ranking to provide insights into the model. This study uses the publicly accessible Physionet CAP sleep database. The model is developed using single-channel standardized EEG recordings from healthy subjects and patients with five types of sleep disorders, namely, insomnia, nocturnal frontal lobe epilepsy (NFLE), periodic leg movement disorder (PLM), rapid eye movement behavior disorder (RBD) and narcolepsy. The best performance is obtained using k-nearest neighbors (KNN) and ensemble bagged trees (EbagT) classifiers. The proposed model achieved a average classification accuracy of 91.6% for healthy subjects and 94.33%, 86.3%, 88.68%, 84.43%, and 88.5% for narcolepsy, RBD, PLM, NFLE, and insomnia subjects respectively, for classifying phases A and B. Our model achieved a average classification accuracy of 92.85% for healthy subjects and 93.9%, 84.9%, 88.0%, 80.92%, and 89.41% for narcolepsy, RBD, PLM, NFLE, and insomnia subjects, respectively while categorizing A subphases (A1, A2, A3). The proposed method may help sleep experts to examine a person’s sleep quality automatically using the microstructure of sleep.

The voltage fluctuations are caused by variations in renewable energy source outputs, increased usage of nonlinear loads, high reactive power consumption of loads, etc. This results in damage to electric components and enormous economic losses. It is necessary to measure the parameters of voltage fluctuations and flicker levels to achieve a secure supply of power. However, higher-order harmonics and noises in voltage waveform make such assessment a burdensome task. The traditional techniques fail to provide exact monitoring due to the uncharacteristic variations in the input voltage signal. Nowadays, data-driven methods have become a prominent choice for non-stationary, nonlinear signal analysis as it efficiently tackles atypical changes by its capability to identify spatio-temporal information from given data. The present paper proposes a data-driven hybrid methodology for monitoring voltage fluctuations using the low-rank approximation of dynamic mode decomposition (DMD) and flat-top finite impulse response (FIR) filter. The low-rank approximation of DMD enables to identify the spatio-temporal coherent structures from the data by mapping the data into a lower-dimensional space. Adaptive flat-top FIR filter tracks the fundamental dynamic phasor by identifying the instantaneous frequency, magnitude and phase variations. The performance of the proposed hybrid methodology is assessed using different test cases, and is compared with various existing approaches. The promising results suggest that the proposed methodology can be used for accurate parameter identification and monitoring voltage fluctuations in smart grid scenarios. It can also be used as an accurate tool in distribution grids to detect higher-order harmonics.

Abstract Predicting treatment outcomes in traumatic brain injury (TBI) patients is challenging worldwide. The present study aimed to achieve the most accurate machine learning (ML) algorithms to predict the outcomes of TBI treatment by evaluating demographic features, laboratory data, imaging indices, and clinical features. We used data from 3347 patients admitted to a tertiary trauma centre in Iran from 2016 to 2021. After the exclusion of incomplete data, 1653 patients remained. We used ML algorithms such as random forest (RF) and decision tree (DT) with ten-fold cross-validation to develop the best prediction model. Our findings reveal that among different variables included in this study, the motor component of the Glasgow coma scale, the condition of pupils, and the condition of cisterns were the most reliable features for predicting in-hospital mortality, while the patients’ age takes the place of cisterns condition when considering the long-term survival of TBI patients. Also, we found that the RF algorithm is the best model to predict the short-term mortality of TBI patients. However, the generalized linear model (GLM) algorithm showed the best performance (with an accuracy rate of 82.03 ± 2.34) in predicting the long-term survival of patients. Our results showed that using appropriate markers and with further development, ML has the potential to predict TBI patients’ survival in the short- and long-term.

Abstract Pain intensity classification using facial images is a challenging problem in computer vision research. This work proposed a patch and transfer learning-based model to classify various pain intensities using facial images. The input facial images were segmented into dynamic-sized horizontal patches or “shutter blinds”. A lightweight deep network DarkNet19 pre-trained on ImageNet1K was used to generate deep features from the shutter blinds and the undivided resized segmented input facial image. The most discriminative features were selected from these deep features using iterative neighborhood component analysis, which were then fed to a standard shallow fine k-nearest neighbor classifier for classification using tenfold cross-validation. The proposed shutter blinds-based model was trained and tested on datasets derived from two public databases—University of Northern British Columbia-McMaster Shoulder Pain Expression Archive Database and Denver Intensity of Spontaneous Facial Action Database—which both comprised four pain intensity classes that had been labeled by human experts using validated facial action coding system methodology. Our shutter blinds-based classification model attained more than 95% overall accuracy rates on both datasets. The excellent performance suggests that the automated pain intensity classification model can be deployed to assist doctors in the non-verbal detection of pain using facial images in various situations (e.g., non-communicative patients or during surgery). This system can facilitate timely detection and management of pain.

Abstract Coronary artery disease (CAD) is a prevalent disease with high morbidity and mortality rates. Invasive coronary angiography is the reference standard for diagnosing CAD but is costly and associated with risks. Noninvasive imaging like cardiac magnetic resonance (CMR) facilitates CAD assessment and can serve as a gatekeeper to downstream invasive testing. Machine learning methods are increasingly applied for automated interpretation of imaging and other clinical results for medical diagnosis. In this study, we proposed a novel CAD detection method based on CMR images by utilizing the feature extraction ability of deep neural networks and combining the features with the aid of a random forest for the very first time. It is necessary to convert image data to numeric features so that they can be used in the nodes of the decision trees. To this end, the predictions of multiple stand-alone convolutional neural networks (CNNs) were considered as input features for the decision trees. The capability of CNNs in representing image data renders our method a generic classification approach applicable to any image dataset. We named our method RF-CNN-F, which stands for Random Forest with CNN Features. We conducted experiments on a large CMR dataset that we have collected and made publicly accessible. Our method achieved excellent accuracy (99.18%) using Adam optimizer compared to a stand-alone CNN trained using fivefold cross validation (93.92%) tested on the same dataset.

Background: Seismic signals record earthquakes and also noise from different sources. The influence of noise makes it difficult to interpret seismograph signals correctly. This study aims to develop a computationally lightweight, accurate, and explainable machine learning model for the automated detection of seismogram signals that could serve as an effective warning system for earthquake prediction. Material and method: We developed a handcrafted model for earthquake detection using a balanced dataset of 5001 earthquakes and 5001 non-earthquake signal samples. The model included multilevel feature extraction, selector-based feature selection, classification, and post-processing. Input signals were decomposed using tunable Q wave transform and fed to a statistical and textural feature extractor based on the most complicated lock pattern (MCLP). Four feature selectors were used to choose the most valuable features for the support vector machine classifier. Additionally, voted vectors were generated using iterative hard majority voting. Finally, the best model was chosen using a greedy algorithm. Results: The presented self-organized MCLP-based feature engineering model yielded 96.82% classification accuracy with 10-fold cross-validation using the seismic signal dataset. Conclusions: Our model attained high seismological signal detection performance comparable with more computationally expensive deep learning models. Our handcrafted explainable feature engineering model is computationally less expensive and can be easily implemented. Furthermore, we have introduced a competitive feature engineering model to the deep learning models for the seismic signal classification model.

Background: Chest computed tomography (CT) has a high sensitivity for detecting COVID-19 lung involvement and is widely used for diagnosis and disease monitoring. We proposed a new image classification model, swin-textural, that combined swin-based patch division with textual feature extraction for automated diagnosis of COVID-19 on chest CT images. The main objective of this work is to evaluate the performance of the swin architecture in feature engineering. Material and method: We used a public dataset comprising 2167, 1247, and 757 (total 4171) transverse chest CT images belonging to 80, 80, and 50 (total 210) subjects with COVID-19, other non-COVID lung conditions, and normal lung findings. In our model, resized 420 × 420 input images were divided using uniform square patches of incremental dimensions, which yielded ten feature extraction layers. At each layer, local binary pattern and local phase quantization operations extracted textural features from individual patches as well as the undivided input image. Iterative neighborhood component analysis was used to select the most informative set of features to form ten selected feature vectors and also used to select the 11th vector from among the top selected feature vectors with accuracy >97.5%. The downstream kNN classifier calculated 11 prediction vectors. From these, iterative hard majority voting generated another nine voted prediction vectors. Finally, the best result among the twenty was determined using a greedy algorithm. Results: Swin-textural attained 98.71% three-class classification accuracy, outperforming published deep learning models trained on the same dataset. The model has linear time complexity. Conclusions: Our handcrafted computationally lightweight swin-textural model can detect COVID-19 accurately on chest CT images with low misclassification rates. The model can be implemented in hospitals for efficient automated screening of COVID-19 on chest CT images. Moreover, findings demonstrate that our presented swin-textural is a self-organized, highly accurate, and lightweight image classification model and is better than the compared deep learning models for this dataset.

Social media contains a plethora of information in the form of text, images, videos, and other data. Users across the globe are increasingly sharing their data on various social media platforms. Sentiment analysis of data, such as text, images, and videos are widely used to understand the feelings of users. In recent years, the convolutional neural network (CNN) has been extensively applied for various applications. The cloud computing environment is a popular service due to its reliability, availability, and easy software integration. However, CNN models are deep neural networks that have a high computational cost. There is a need for CNN models which utilize lesser computational resources especially when these models are deployed in a cloud environment due to the remote physicality of servers, resource optimization, and infrastructure cost reduction. In this research, Gabor filters are integrated with CNN models to improve image sentiment analysis in a cloud environment, with advantages such as the reduction in computation energy and time, the elimination of the need for pre-trained models, and a perceived accuracy improvement. Two variants of Gabor-CNN (G-CNN) models with a different number of pooling and normalization layers are developed. The proposed G-CNN is trained and tested using five standard databases as SentiBank, Twitter, MVSO, MultiView_I, and MultiView_II. Maximum classification accuracies of 91.71%, 92.52%, 97.39%, 90.88%, and 91.31% are obtained on SentiBank, Twitter, MVSO, MultiView_I, and MultiView_II databases respectively using the developed models. The proposed G-CNN model has provided an accuracy of 92.76% on average.

COVID-19 caused by the transmission of SARS-CoV-2 virus taking a huge toll on global health and caused life-threatening medical complications and elevated mortality rates, especially among older adults and people with existing morbidity. Current evidence suggests that the virus spreads primarily through respiratory droplets emitted by infected persons when breathing, coughing, sneezing, or speaking. These droplets can reach another person through their mouth, nose, or eyes, resulting in infection. The “gold standard” for clinical diagnosis of SARS-CoV-2 is the laboratory-based nucleic acid amplification test, which includes the reverse transcription-polymerase chain reaction (RT-PCR) test on nasopharyngeal swab samples. The main concerns with this type of test are the relatively high cost, long processing time, and considerable false-positive or false-negative results. Alternative approaches have been suggested to detect the SARS-CoV-2 virus so that those infected and the people they have been in contact with can be quickly isolated to break the transmission chains and hopefully, control the pandemic. These alternative approaches include electrochemical biosensing and deep learning. In this review, we discuss the current state-of-the-art technology used in both fields for public health surveillance of SARS-CoV-2 and present a comparison of both methods in terms of cost, sampling, timing, accuracy, instrument complexity, global accessibility, feasibility, and adaptability to mutations. Finally, we discuss the issues and potential future research approaches for detecting the SARS-CoV-2 virus utilizing electrochemical biosensing and deep learning.

Anxiety, learning disabilities, and depression are the symptoms of attention deficit hyperactivity disorder (ADHD), an isogenous pattern of hyperactivity, impulsivity, and inattention. For the early diagnosis of ADHD, electroencephalogram (EEG) signals are widely used. However, the direct analysis of an EEG is highly challenging as it is time-consuming, nonlinear, and nonstationary in nature. Thus, in this paper, a novel approach (LSGP-USFNet) is developed based on the patterns obtained from Ulam’s spiral and Sophia Germain’s prime numbers. The EEG signals are initially filtered to remove the noise and segmented with a non-overlapping sliding window of a length of 512 samples. Then, a time–frequency analysis approach, namely continuous wavelet transform, is applied to each channel of the segmented EEG signal to interpret it in the time and frequency domain. The obtained time–frequency representation is saved as a time–frequency image, and a non-overlapping n × n sliding window is applied to this image for patch extraction. An n × n Ulam’s spiral is localized on each patch, and the gray levels are acquired from this patch as features where Sophie Germain’s primes are located in Ulam’s spiral. All gray tones from all patches are concatenated to construct the features for ADHD and normal classes. A gray tone selection algorithm, namely ReliefF, is employed on the representative features to acquire the final most important gray tones. The support vector machine classifier is used with a 10-fold cross-validation criteria. Our proposed approach, LSGP-USFNet, was developed using a publicly available dataset and obtained an accuracy of 97.46% in detecting ADHD automatically. Our generated model is ready to be validated using a bigger database and it can also be used to detect other children’s neurological disorders.

Diabetic Macular Edema (DME) is a severe ocular complication commonly found in patients with diabetes. The condition can precipitate a significant drop in VA and, in extreme cases, may result in irreversible vision loss. Optical Coherence Tomography (OCT), a technique that yields high-resolution retinal images, is often employed by clinicians to assess the extent of DME in patients. However, the manual interpretation of OCT B-scan images for DME identification and severity grading can be error-prone, with false negatives potentially resulting in serious repercussions. In this paper, we investigate an Artificial Intelligence (AI) driven system that offers an end-to-end automated model, designed to accurately determine DME severity using OCT B-Scan images. This model operates by extracting specific biomarkers such as Disorganization of Retinal Inner Layers (DRIL), Hyper Reflective Foci (HRF), and cystoids from the OCT image, which are then utilized to ascertain DME severity. The rules guiding the fuzzy logic engine are derived from contemporary research in the field of DME and its association with various biomarkers evident in the OCT image. The proposed model demonstrates high efficacy, identifying images with DRIL with 93.3% accuracy and successfully segmenting HRF and cystoids from OCT images with dice similarity coefficients of 91.30% and 95.07% respectively. This study presents a comprehensive system capable of accurately grading DME severity using OCT B-scan images, serving as a potentially invaluable tool in the clinical assessment and treatment of DME.

Screening programs for early lung cancer diagnosis are uncommon, primarily due to the challenge of reaching at-risk patients located in rural areas far from medical facilities. To overcome this obstacle, a comprehensive approach is needed that combines mobility, low cost, speed, accuracy, and privacy. One potential solution lies in combining the chest X-ray imaging mode with federated deep learning, ensuring that no single data source can bias the model adversely. This study presents a pre-processing pipeline designed to debias chest X-ray images, thereby enhancing internal classification and external generalization. The pipeline employs a pruning mechanism to train a deep learning model for nodule detection, utilizing the most informative images from a publicly available lung nodule X-ray dataset. Histogram equalization is used to remove systematic differences in image brightness and contrast. Model training is then performed using combinations of lung field segmentation, close cropping, and rib/bone suppression. The resulting deep learning models, generated through this pre-processing pipeline, demonstrate successful generalization on an independent lung nodule dataset. By eliminating confounding variables in chest X-ray images and suppressing signal noise from the bone structures, the proposed deep learning lung nodule detection algorithm achieves an external generalization accuracy of 89%. This approach paves the way for the development of a low-cost and accessible deep learning-based clinical system for lung cancer screening.

Major Depressive Disorder (MDD) is a neurohormonal disorder that causes persistent negative thoughts, mood and feelings, often accompanied with suicidal ideation (SI). Current clinical diagnostic approaches are solely based on psychiatric interview questionnaires. Thus, a computational intelligence tool for the automated detection of MDD with and without suicidal ideation is presented in this study. Since MDD is proven to affect cardiovascular and respiratory systems, the aim of the study is to automatically identify the disorder severity in MDD patients using corresponding multi-modal physiological signals, including electrocardiogram (ECG), finger photoplethysmography (PPG) and respiratory signals (RSP). Data from 88 subjects were used in this study, out of which 25 were MDD patients without SI (MDDSI−), 18 MDD patients with SI (MDDSI+), and 45 normal subjects. Multi-modal physiological signals were acquired from each subject, including ECG, RSP, and PPG signals, and then pre-processed. Discrete wavelet transform (DWT) was applied to the signals, which were decomposed up to six levels, and then eleven nonlinear features were extracted. The features were ranked according to the analysis of variance test and Marginal Fisher Analysis was employed to reduce the feature set, after which the reduced features were ranked again to select the most discriminatory features. Support vector machine with polynomial radial basis function (SVM-RBF) as well as k-nearest neighbor (KNN) classifiers were used to classify the significant features. The performance of the classifiers was evaluated in a 10-fold cross validation scheme. The best performance achieved for the classification of MDDSI+ patients was up to 85.2%, by using selected features from the obtained multi-modal signals with SVM-RBF, while it was up to 96.6% for the detection of MDD patients against healthy subjects. This work is a step toward the utilization of automated tools in diagnostics and monitoring of MDD patients in a personalized and wearable healthcare system.

Autism spectrum disorder (ASD) is a brain condition characterized by diverse signs and symptoms that appear in early childhood. ASD is also associated with communication deficits and repetitive behavior in affected individuals. Various ASD detection methods have been developed, including neuroimaging modalities and psychological tests. Among these methods, magnetic resonance imaging (MRI) imaging modalities are of paramount importance to physicians. Clinicians rely on MRI modalities to diagnose ASD accurately. The MRI modalities are non-invasive methods that include functional (fMRI) and structural (sMRI) neuroimaging methods. However, diagnosing ASD with fMRI and sMRI for specialists is often laborious and time-consuming; therefore, several computer-aided design systems (CADS) based on artificial intelligence (AI) have been developed to assist specialist physicians. Conventional machine learning (ML) and deep learning (DL) are the most popular schemes of AI used for diagnosing ASD. This study aims to review the automated detection of ASD using AI. We review several CADS that have been developed using ML techniques for the automated diagnosis of ASD using MRI modalities. There has been very limited work on the use of DL techniques to develop automated diagnostic models for ASD. A summary of the studies developed using DL is provided in the Supplementary Appendix. Then, the challenges encountered during the automated diagnosis of ASD using MRI and AI techniques are described in detail. Additionally, a graphical comparison of studies using ML and DL to diagnose ASD automatically is discussed. We suggest future approaches to detecting ASDs using AI techniques and MRI neuroimaging.

Monkeypox or Mpox is an infectious virus predominantly found in Africa. It has spread to many countries since its latest outbreak. Symptoms such as headaches, chills, and fever are observed in humans. Lumps and rashes also appear on the skin (similar to smallpox, measles, and chickenpox). Many artificial intelligence (AI) models have been developed for accurate and early diagnosis. In this work, we systematically reviewed recent studies that used AI for mpox-related research. After a literature search, 34 studies fulfilling prespecified criteria were selected with the following subject categories: diagnostic testing of mpox, epidemiological modeling of mpox infection spread, drug and vaccine discovery, and media risk management. In the beginning, mpox detection using AI and various modalities was described. Other applications of ML and DL in mitigating mpox were categorized later. The various machine and deep learning algorithms used in the studies and their performance were discussed. We believe that a state-of-the-art review will be a valuable resource for researchers and data scientists in developing measures to counter the mpox virus and its spread.

Continuous advancements of technologies such as machine-to-machine interactions and big data analysis have led to the internet of things (IoT) making information sharing and smart decision-making possible using everyday devices. On the other hand, swarm intelligence (SI) algorithms seek to establish constructive interaction among agents regardless of their intelligence level. In SI algorithms, multiple individuals run simultaneously and possibly in a cooperative manner to address complex nonlinear problems. In this paper, the application of SI algorithms in IoT is investigated with a special focus on the internet of medical things (IoMT). The role of wearable devices in IoMT is briefly reviewed. Existing works on applications of SI in addressing IoMT problems are discussed. Possible problems include disease prediction, data encryption, missing values prediction, resource allocation, network routing, and hardware failure management. Finally, research perspectives and future trends are outlined.

Dental caries is the most frequent dental health issue in the general population. Dental caries can result in extreme pain or infections, lowering people’s quality of life. Applying machine learning models to automatically identify dental caries can lead to earlier treatment. However, physicians frequently find the model results unsatisfactory due to a lack of explainability. Our study attempts to address this issue with an explainable deep learning model for detecting dental caries. We tested three prominent pre-trained models, EfficientNet-B0, DenseNet-121, and ResNet-50, to determine which is best for the caries detection task. These models take panoramic images as the input, producing a caries–non-caries classification result and a heat map, which visualizes areas of interest on the tooth. The model performance was evaluated using whole panoramic images of 562 subjects. All three models produced remarkably similar results. However, the ResNet-50 model exhibited a slightly better performance when compared to EfficientNet-B0 and DenseNet-121. This model obtained an accuracy of 92.00%, a sensitivity of 87.33%, and an F1-score of 91.61%. Visual inspection showed us that the heat maps were also located in the areas with caries. The proposed explainable deep learning model diagnosed dental caries with high accuracy and reliability. The heat maps help to explain the classification results by indicating a region of suspected caries on the teeth. Dentists could use these heat maps to validate the classification results and reduce misclassification.

Next basket recommendation is a critical task in market basket data analysis. It is particularly important in grocery shopping, where grocery lists are an essential part of shopping habits of many customers. In this work, we first present a new grocery Recommender System available on the MyGroceryTour platform. Our online system uses different traditional machine learning (ML) and deep learning (DL) algorithms, and provides recommendations to users in a real-time manner. It aims to help Canadian customers create their personalized intelligent weekly grocery lists based on their individual purchase histories, weekly specials offered in local stores, and product cost and availability information. We perform clustering analysis to partition given customer profiles into four non-overlapping clusters according to their grocery shopping habits. Then, we conduct computational experiments to compare several traditional ML algorithms and our new DL algorithm based on the use of a gated recurrent unit (GRU)-based recurrent neural network (RNN) architecture. Our DL algorithm can be viewed as an extension of DREAM (Dynamic REcurrent bAsket Model) adapted to multi-class (i.e. multi-store) classification, since a given user can purchase recommended products in different grocery stores in which these products are available. Among traditional ML algorithms, the highest average F-score of 0.516 for the considered data set of 831 customers was obtained using Random Forest, whereas our proposed DL algorithm yielded the average F-score of 0.559 for this data set. The main advantage of the presented Recommender System is that our intelligent recommendation is personalized, since a separate traditional ML or DL model is built for each customer considered. Such a personalized approach allows us to outperform the prediction results provided by general state-of-the-art DL models.

Sleep apnea is a severe sleep disorder that degrades the quality of sleep and causes patients to feel restless and fatigued even after a full night's sleep. Sleep-disordered breathing (SDB) research has grown its importance in recent decades, owing to patients' substantial health effects and poor quality of life. Obstructive sleep apnea (OSA), the most prevalent kind of sleep disorder, that is caused due to a partial blockage in the nasal airway, resulting in low oxygen levels in the patient's blood. The drop in oxygen level may cause high blood pressure, which can lead to major cardiac issues. Further, OSA is also a risk for pregnancy complications in first-time pregnant women (nulliparous pregnancy), as emerging research and studies suggest a strong relationship between sleep-disordered breathing patients and critical pregnancy outcomes like preeclampsia, preterm birth, gestational diabetes, and intrauterine growth retardation.Polysomnography (PSG) recordings with multiple-channel various modalities signals are often used to detect sleep problems. This study aims to detect sleep apnea in nulliparous pregnant women using a simplified PSG-based system that employed only airflow signals. We used the ‘nulliparous pregnancy outcomes study monitoring mothers-to-be (nuMOM2b)’ database, containing 3012 women subjects for sleep-disordered breathing (SDB) related sub-studies. To the best of our knowledge, this is the first computer-aided diagnostic tool developed to automatically identify sleep apnea in pregnant women. In this work on the detection of OSA, we proposed the use of a single-channel airflow (AF) signal in combination with an orthogonal wavelet filter bank. The L1 − norm features are extracted from wavelet coefficients. The selected features are then fed into various machine learning classifiers to automatically detect apnea. The developed model obtained a highest accuracy of 83.9% with an F1 score of 0.91 using the RUSBoosted tree classifier with ten-fold cross-validation (CV) strategy. The system proposed can be used in a home-based compact clinical setup to detect apnea in pregnant women.

In the presence of gestational diabetes mellitus (GDM), the fetus is exposed to a hyperinsulinemia environment. This environment can cause a wide range of metabolic and fetal cardiac structural alterations. Fetal myocardial hypertrophy predominantly affecting the interventricular septum possesses a morphology of disarray similar to hypertrophic cardiomyopathy, and may be present in some GDM neonates after birth. Myocardial thickness may increase in GDM fetuses independent of glycemic control status and fetal weight. Fetal echocardiography performed on fetuses with GDM helps in assessing cardiac structure and function, and to diagnose myocardial hypertrophy. There are few studies in the literature which have established evidence for morphologic variation associated with cardiac hypertrophy among fetuses of GDM mothers. In this study, fetal ultrasound images of normal, pregestational diabetes mellitus (preGDM) and GDM mothers were used to develop a computer aided diagnostic (CAD) tool. We proposed a new method called local preserving class separation (LPCS) framework to preserve the geometrical configuration of normal and preGDM/GDM subjects. The generated shearlet based texture features under LPCS framework showed promising results compared with deep learning algorithms. The proposed method achieved a maximum accuracy of 98.15% using a support vector machine (SVM) classifier. Hence, this paradigm can be helpful to physicians in detecting fetal myocardial hypertrophy in preGDM/GDM mothers.

Background: Sleep stage classification is a crucial process for the diagnosis of sleep or sleep-related diseases. Currently, this process is based on manual electroencephalogram (EEG) analysis, which is resource-intensive and error-prone. Various machine learning models have been recommended to standardize and automate the analysis process to address these problems. Materials and methods: The well-known cyclic alternating pattern (CAP) sleep dataset is used to train and test an L-tetrolet pattern-based sleep stage classification model in this research. By using this dataset, the following three cases are created, and they are: Insomnia, Normal, and Fused cases. For each of these cases, the machine learning model is tasked with identifying six sleep stages. The model is structured in terms of feature generation, feature selection, and classification. Feature generation is established with a new L-tetrolet (Tetris letter) function and multiple pooling decomposition for level creation. We fuse ReliefF and iterative neighborhood component analysis (INCA) feature selection using a threshold value. The hybrid and iterative feature selectors are named threshold selection-based ReliefF and INCA (TSRFINCA). The selected features are classified using a cubic support vector machine. Results: The presented L-tetrolet pattern and TSRFINCA-based sleep stage classification model yield 95.43%, 91.05%, and 92.31% accuracies for Insomnia, Normal dataset, and Fused cases, respectively. Conclusion: The recommended L-tetrolet pattern and TSRFINCA-based model push the envelope of current knowledge engineering by accurately classifying sleep stages even in the presence of sleep disorders.

Emotion recognition is one of the most important issues in human–computer interaction (HCI), neuroscience, and psychology fields. It is generally accepted that emotion recognition with neural data such as electroencephalography (EEG) signals, functional magnetic resonance imaging (fMRI), and near-infrared spectroscopy (NIRS) is better than other emotion detection methods such as speech, mimics, body language, facial expressions, etc., in terms of reliability and accuracy. In particular, EEG signals are bioelectrical signals that are frequently used because of the many advantages they offer in the field of emotion recognition. This study proposes an improved approach for EEG-based emotion recognition on a publicly available newly published dataset, VREED. Differential entropy (DE) features were extracted from four wavebands (theta 4–8 Hz, alpha 8–13 Hz, beta 13–30 Hz, and gamma 30–49 Hz) to classify two emotional states (positive/negative). Five classifiers, namely Support Vector Machine (SVM), k-Nearest Neighbor (kNN), Naïve Bayesian (NB), Decision Tree (DT), and Logistic Regression (LR) were employed with DE features for the automated classification of two emotional states. In this work, we obtained the best average accuracy of 76.22% ± 2.06 with the SVM classifier in the classification of two states. Moreover, we observed from the results that the highest average accuracy score was produced with the gamma band, as previously reported in studies in EEG-based emotion recognition.

Magnetic resonance imaging (MRI) is a standard tool for the diagnosis of stroke, but its manual interpretation by experts is arduous and time-consuming. Thus, there is a need for computer-aided-diagnosis (CAD) models for the automatic segmentation and classification of stroke on brain MRI. The heterogeneity of stroke pathogenesis, morphology, image acquisition modalities, sequences, and intralesional tissue signal intensity, as well as lesion-to-normal tissue contrast, pose significant challenges to the development of such systems. Machine learning (ML) is increasingly being used in predictive neuroimaging diagnosis and prognostication. This paper reviews image processing and machine learning techniques that have been applied to detect ischemic stroke on brain MRI, including details on image acquisition, pre-processing, techniques to segment, extraction of features, and classification into stroke types. The main objective of this work is to find the state-of-art machine learning techniques used to predict the ischemic stroke and their application in clinical set-up. The article selection is performed according to PRISMA guideline. The state-of-the-art on automated MRI stroke diagnosis, with a focus on machine learning, is discussed, along with its advantages and limitations. We found that the various machine learning models discussed in this article are able to detect the infarcts with an acceptable accuracy of 70–90%. However, no one has highlighted the time complexity to predict the stroke in the model developed, which is an important factor. The work concludes with proposals for future recommendations for building efficient and robust deep learning (DL) models for quantitative brain MRI analysis. In recent work, with the application of DL approaches, using large datasets to train the models has improved the detection accuracy and reduced computational complexity. We suggest that the design of a decision support system based on artificial intelligence (AI) and clinical data presenting symptoms is essential to support clinicians to accelerate diagnosis and timeous therapy in the emergency management of stroke.

Attention deficit hyperactivity disorder (ADHD) is a common neurodevelopmental condition worldwide. In this research, we used an ADHD electroencephalography (EEG) dataset containing more than 4000 EEG signals. Moreover, these EEGs are noisy signals. A new hand-modeled EEG classification model has been proposed to separate healthy versus ADHD individuals using the EEG signals. In this model, a new ternary motif pattern (TMP) has been incorporated. We have mimicked deep learning networks to create this hand-modeled classification method. The Tunable Q Wavelet Transform (TQWT) has been utilized to generate wavelet subbands. We applied the proposed TMP and statistics to construct informative features from both raw EEG signals and wavelet bands by generating TQWT. Herein, features have been generated by 18 subbands and the original EEG signal. Thus, this model is named TMP19. The most informative features have been chosen by deploying neighborhood component analysis (NCA), and the selected features have been classified using the k-nearest neighbor (kNN) classifier. The used ADHD EEG dataset has 14 channels. Thus, these three phases—(i) feature extraction with TQWT, TMP, and statistics; (ii) feature selection by deploying NCA; and (iii) classification with kNN—have been applied to each channel. Iterative hard majority voting (IHMV) has been applied to obtain a higher and more general classification response. Our model attained 95.57% and 77.93% classification accuracies by deploying 10-fold and leave one subject out (LOSO) cross-validations, respectively.

Diabetic retinopathy (DR) is a common complication of diabetes that can lead to progressive vision loss. Regular surveillance with fundal photography, early diagnosis, and prompt intervention are paramount to reducing the incidence of DR-induced vision loss. However, manual interpretation of fundal photographs is subject to human error. In this study, a new method based on horizontal and vertical patch division was proposed for the automated classification of DR images on fundal photographs. The novel sides of this study are given as follows. We proposed a new non-fixed-size patch division model to obtain high classification results and collected a new fundus image dataset. Moreover, two datasets are used to test the model: a newly collected three-class (normal, non-proliferative DR, and proliferative DR) dataset comprising 2355 DR images and the established open-access five-class Asia Pacific Tele-Ophthalmology Society (APTOS) 2019 dataset comprising 3662 images. Two analysis scenarios, Case 1 and Case 2, with three (normal, non-proliferative DR, and proliferative DR) and five classes (normal, mild DR, moderate DR, severe DR, and proliferative DR), respectively, were derived from the APTOS 2019 dataset. These datasets and these cases have been used to demonstrate the general classification performance of our proposal. By applying transfer learning, the last fully connected and global average pooling layers of the DenseNet201 architecture were used to extract deep features from input DR images and each of the eight subdivided horizontal and vertical patches. The most discriminative features are then selected using neighborhood component analysis. These were fed as input to a standard shallow cubic support vector machine for classification. Our new DR dataset obtained 94.06% and 91.55% accuracy values for three-class classification with 80:20 hold-out validation and 10-fold cross-validation, respectively. As can be seen from steps of the proposed model, a new patch-based deep-feature engineering model has been proposed. The proposed deep-feature engineering model is a cognitive model, since it uses efficient methods in each phase. Similar excellent results were seen for three-class classification with the Case 1 dataset. In addition, the model attained 87.43% and 84.90% five-class classification accuracy rates using 80:20 hold-out validation and 10-fold cross-validation, respectively, on the Case 2 dataset, which outperformed prior DR classification studies based on the five-class APTOS 2019 dataset. Our model attained about >2% classification results compared to others. These findings demonstrate the accuracy and robustness of the proposed model for classification of DR images.

Medical imaging and analysis are useful to visualize anatomic structure. However, analysis of the pathologic substrate is difficult and inefficient when using simple imaging tools. The manual detection and classification of brain abnormality is particularly tedious. Moreover, the currently used methodology suffers from interobserver variability during image interpretation. Magnetic resonance imaging (MRI) is an efficient imaging technique for revealing complex anatomical architecture, and it is highly efficacious for precise brain imaging. Herein, we describe a novel computer aided diagnosis method for automated processing of brain MRI images. The performances of two decomposition techniques, namely, bidimensional empirical mode decomposition and variational mode decomposition (VMD), are compared. Thereafter, bispectral feature extraction and supervised neighborhood projection embedding are implemented to represent each feature in a new subspace, for the automated classification of various categories of disease. A support vector machine classifier is used to train and test the performance accuracy. The level of classification accuracy of 90.68%, 99.43% sensitivity and 87.95% specificity is obtained using the VMD technique. Hence, the developed system can be used as an adjunct tool by radiologists to confirm their screening.

Video surveillance is gaining popularity in numerous applications, including facility management, traffic monitoring, crowd analysis, and urban security. Despite the increasing demand for closed-circuit television (CCTV) and related infrastructure in public spaces, there remains a notable lack of readily-deployable automated surveillance systems. In this study, we present a low-cost and efficient approach that integrates the use of computational object recognition to perform fully-automated identification, tracking, and counting of human traffic on camera video streams. Two software implementations are explored and the performance of these schemes is compared. Validation against controlled and non-controlled real-world environments is also demonstrated. The implementation provides automated video analytics for medium crowd density monitoring and tracking, eliminating labor-intensive tasks traditionally requiring human operation, with results indicating great reliability in real-life scenarios.

Coronary artery disease (CAD) is one of the main causes of cardiac death around the world. Due to its significant impact on the society, early and accurate detection of CAD is essential. This study proposes a novel nested ensemble nu-Support Vector Classification (NE-nu-SVC) model which combines several traditional machine learning methods and ensemble learning techniques for effective diagnosis of CAD. We validated our model using two well-known CAD datasets (Z-Alizadeh Sani and Cleveland). To improve the performance of the model, we selected clinically significant features from the datasets using a genetic search algorithm. To further improve our results, we applied a multi-level filtering technique to balance the data using the ClassBlancer and Resample methods. Our base algorithm, nu-SVC, is performed using four well-known kernel functions (linear, polynomial, radial basis (RBF) and sigmoid). The proposed NE-nu-SVC model provided the highest accuracy of 94.66% and 98.60% to predict CAD entities in the Z-Alizadeh Sani and Cleveland CAD datasets, respectively. Our system can aid the clinicians to diagnose CAD accurately and may probably replace other invasive diagnostic techniques.

Data centers are mission-critical infrastructures. There are strict service level requirements and standards imposed on operators and maintainers to ensure reliable run-the-clock operation. In the context of thermal management and data hall environmental control, the formation of hot and cold spots around server cabinets are especially undesirable, and can result in overheating, lifespan reductions, and performance throttling in the former and condensation damage in the latter. In this paper, we present a comprehensive multi-pronged methodology in data center environmental control, comprising computational fluid dynamics (CFD) simulation-aided predictive design first-stage approach, and a complementing Internet of Things (IoT) reactive management system that autonomously monitors and regulates fluctuations in thermal parameters. The novel hybrid methodology is demonstrated on various test scenarios derived from real-world context, and prototypes of the IoT system have been experimentally validated. The approach is shown to be efficient in eliminating unfavourable environmental variations and provides an enhanced understanding of common design problems and respective mitigation measures.

Breast cancer is the most commonly occurring cancer in women worldwide. While mammography remains the gold standard in breast cancer screening, ultrasound is an important imaging modality for both screening and cancer diagnosis. This paper presents a novel method for the detection of breast lesions in ultrasound images using texton filter banks, local configuration pattern features, and classification, without employing any segmentation technique. The developed method was able to accurately detect and classify breast lesions and achieved an accuracy, sensitivity, specificity, and positive predictive value of 96.1%, 96.5%, 95.3%, and 97.9%, respectively. The paradigm that we describe may, therefore, be useful as an effective tool to detect breast nodules during screening and in whole breast imaging, enabling clinicians to focus on images where a lesion is already known to be present. The developed method may also serve as a component for automatic breast nodule detection, and, when found, for the subsequent classification between lesion type benign versus malignant.

Multiple sclerosis (MS) is a chronic demyelinating condition characterized by plaques in the white matter of the central nervous system that can be detected using magnetic resonance imaging (MRI). Many deep learning models for automated MS detection based on MRI have been presented in the literature. We developed a computationally lightweight machine learning model for MS diagnosis using a novel handcrafted feature engineering approach. The study dataset comprised axial and sagittal brain MRI images that were prospectively acquired from 72 MS and 59 healthy subjects who attended the Ozal University Medical Faculty in 2021. The dataset was divided into three study subsets: axial images only (n = 1652), sagittal images only (n = 1775), and combined axial and sagittal images (n = 3427) of both MS and healthy classes. All images were resized to 224 × 224. Subsequently, the features were generated with a fixed-size patch-based (exemplar) feature extraction model based on local phase quantization (LPQ) with three-parameter settings. The resulting exemplar multiple parameters LPQ (ExMPLPQ) features were concatenated to form a large final feature vector. The top discriminative features were selected using iterative neighborhood component analysis (INCA). Finally, a k-nearest neighbor (kNN) algorithm, Fine kNN, was deployed to perform binary classification of the brain images into MS vs. healthy classes. The ExMPLPQ-based model attained 98.37%, 97.75%, and 98.22% binary classification accuracy rates for axial, sagittal, and hybrid datasets, respectively, using Fine kNN with 10-fold cross-validation. Furthermore, our model outperformed 19 established pre-trained deep learning models that were trained and tested with the same data. Unlike deep models, the ExMPLPQ-based model is computationally lightweight yet highly accurate. It has the potential to be implemented as an automated diagnostic tool to screen brain MRIs for white matter lesions in suspected MS patients.

The fetal echocardiogram is useful for monitoring and diagnosing cardiovascular diseases in the fetus in utero. Importantly, it can be used for assessing prenatal congenital heart disease, for which timely intervention can improve the unborn child’s outcomes. In this regard, artificial intelligence (AI) can be used for the automatic analysis of fetal heart ultrasound images. This study reviews nondeep and deep learning approaches for assessing the fetal heart using standard four-chamber ultrasound images. The state-of-the-art techniques in the field are described and discussed. The compendium demonstrates the capability of automatic assessment of the fetal heart using AI technology. This work can serve as a resource for research in the field.

Hypertrophic cardiomyopathy (HCM) is a genetic disorder that exhibits a wide spectrum of clinical presentations, including sudden death. Early diagnosis and intervention may avert the latter. Left ventricular hypertrophy on heart imaging is an important diagnostic criterion for HCM, and the most common imaging modality is heart ultrasound (US). The US is operator-dependent, and its interpretation is subject to human error and variability. We proposed an automated computer-aided diagnostic tool to discriminate HCM from healthy subjects on US images. We used a local directional pattern and the ResNet-50 pretrained network to classify heart US images acquired from 62 known HCM patients and 101 healthy subjects. Deep features were ranked using Student’s t-test, and the most significant feature (SigFea) was identified. An integrated index derived from the simulation was defined as 100·log10(SigFea/2) in each subject, and a diagnostic threshold value was empirically calculated as the mean of the minimum and maximum integrated indices among HCM and healthy subjects, respectively. An integrated index above a threshold of 0.5 separated HCM from healthy subjects with 100% accuracy in our test dataset.

Arterial hypertension (HT) is a common cardiovascular disease and, if not treated at an early stage, can lead to serious complications. It is difficult to precisely describe because it is a dynamic physiological state. Monitoring of HT is subjective and prone to mistakes. Therefore, various computer-assisted diagnostic methods have been designed. The proposed work is based on ballistocardiography (BCG) signals for the diagnosis of healthy control (HC) and HT subjects. The identification of HT based on BCG signals of 30-s duration using empirical mode decomposition (EMD) and wavelet transform (WT) with nonlinear techniques is proposed in this work. The BCG signal is decomposed into five sub-bands (SBs) using WT and five-level of intrinsic mode functions (IMFs) using EMD. Then, the various non-linear features are calculated for all five-level wavelet decomposition and IMFs. The non-linear features are extracted from the SBs of WT and IMFs of EMD. These features are fed to ensemble gentleboost (EGB) classifier with 10-fold cross-validation strategy for automated detection of HC and HT groups. In this study, the performance of WT and EMD techniques are compared. The proposed HT identification model accomplished the highest average classification accuracy of 89% using WT method. In future, we plan to extend this work using more BCG data.

Background and Purpose: Machine learning models have been used to diagnose schizophrenia. The main purpose of this research is to introduce an effective schizophrenia hand-modeled classification method. Method: A public electroencephalogram (EEG) signal data set was used in this work, and an automated schizophrenia detection model is presented using a cyclic group of prime order with a modulo 17 operator. Therefore, the presented feature extractor was named as the cyclic group of prime order pattern, CGP17Pat. Using the proposed CGP17Pat, a new multilevel feature extraction model is presented. To choose a highly distinctive feature, iterative neighborhood component analysis (INCA) was used, and these features were classified using k-nearest neighbors (kNN) with the 10-fold cross-validation and leave-one-subject-out (LOSO) validation techniques. Finally, iterative hard majority voting was employed in the last phase to obtain channel-wise results, and the general results were calculated. Results: The presented CGP17Pat-based EEG classification model attained 99.91% accuracy employing 10-fold cross-validation and 84.33% accuracy using the LOSO strategy. Conclusions: The findings and results depicted the high classification ability of the presented cryptologic pattern for the data set used.

Recently, deep models have been very popular because they achieve excellent performance with many classification problems. Deep networks have high computational complexities and require specific hardware. To overcome this problem (without decreasing classification ability), a hand-modeled feature selection method is proposed in this paper. A new shape-based local feature extractor is presented which uses the geometric shape of the frustum. By using a frustum pattern, textural features are generated. Moreover, statistical features have been extracted in this model. Textures and statistics features are fused, and a hybrid feature extraction phase is obtained; these features are low-level. To generate high level features, tunable Q factor wavelet transform (TQWT) is used. The presented hybrid feature generator creates 154 feature vectors; hence, it is named Frustum154. In the multilevel feature creation phase, this model can select the appropriate feature vectors automatically and create the final feature vector by merging the appropriate feature vectors. Iterative neighborhood component analysis (INCA) chooses the best feature vector, and shallow classifiers are then used. Frustum154 has been tested on three basic hand-movement sEMG datasets. Hand-movement sEMG datasets are commonly used in biomedical engineering, but there are some problems in this area. The presented models generally required one dataset to achieve high classification ability. In this work, three sEMG datasets have been used to test the performance of Frustum154. The presented model is self-organized and selects the most informative subbands and features automatically. It achieved 98.89%, 94.94%, and 95.30% classification accuracies using shallow classifiers, indicating that Frustum154 can improve classification accuracy.