Search

The study aims to investigate the potential of training efficient deep learning models by using 2.5D (2.5-Dimension) masks of sICH. Furthermore, it intends to evaluate and compare the predictive performance of a joint model incorporating four types of features with standalone 2.5D deep learning, radiomics, radiology, and clinical models for early expansion in sICH. A total of 254 sICH patients were enrolled retrospectively and divided into two groups according to whether the hematoma was enlarged or not. The 2.5D mask of sICH is constructed with the maximum axial, coronal and sagittal planes of the hematoma, which is used to train the deep learning model and extract deep learning features. Predictive models were built on clinic, radiology, radiomics and deep learning features separately and four type features jointly. The diagnostic performance of each model was measured using the receiver operating characteristic curve (AUC), Accuracy, Recall, F1 and decision curve analysis (DCA). The AUCs of the clinic model, radiology model, radiomics model, deep learning model, joint model, and nomogram model on the train set (training and Cross-validation) were 0.639, 0.682, 0.859, 0.807, 0.939, and 0.942, respectively, while the AUCs on the test set (external validation) were 0.680, 0.758, 0.802, 0.857, 0.929, and 0.926. Decision curve analysis showed that the joint model was superior to the other models and demonstrated good consistency between the predicted probability of early hematoma expansion and the actual occurrence probability. Our study demonstrates that the joint model is a more efficient and robust prediction model, as verified by multicenter data. This finding highlights the potential clinical utility of a multifactorial prediction model that integrates various data sources for prognostication in patients with intracerebral hemorrhage. The Critical Relevance Statement: Combining 2.5D deep learning features with clinic features, radiology markers, and radiomics signatures to establish a joint model enabling physicians to conduct better-individualized assessments the risk of early expansion of sICH. [ABSTRACT FROM AUTHOR]

Abstract The study aims to investigate the potential of training efficient deep learning models by using 2.5D (2.5-Dimension) masks of sICH. Furthermore, it intends to evaluate and compare the predictive performance of a joint model incorporating four types of features with standalone 2.5D deep learning, radiomics, radiology, and clinical models for early expansion in sICH. A total of 254 sICH patients were enrolled retrospectively and divided into two groups according to whether the hematoma was enlarged or not. The 2.5D mask of sICH is constructed with the maximum axial, coronal and sagittal planes of the hematoma, which is used to train the deep learning model and extract deep learning features. Predictive models were built on clinic, radiology, radiomics and deep learning features separately and four type features jointly. The diagnostic performance of each model was measured using the receiver operating characteristic curve (AUC), Accuracy, Recall, F1 and decision curve analysis (DCA). The AUCs of the clinic model, radiology model, radiomics model, deep learning model, joint model, and nomogram model on the train set (training and Cross-validation) were 0.639, 0.682, 0.859, 0.807, 0.939, and 0.942, respectively, while the AUCs on the test set (external validation) were 0.680, 0.758, 0.802, 0.857, 0.929, and 0.926. Decision curve analysis showed that the joint model was superior to the other models and demonstrated good consistency between the predicted probability of early hematoma expansion and the actual occurrence probability. Our study demonstrates that the joint model is a more efficient and robust prediction model, as verified by multicenter data. This finding highlights the potential clinical utility of a multifactorial prediction model that integrates various data sources for prognostication in patients with intracerebral hemorrhage. The Critical Relevance Statement: Combining 2.5D deep learning features with clinic features, radiology markers, and radiomics signatures to establish a joint model enabling physicians to conduct better-individualized assessments the risk of early expansion of sICH.

In this research, an intelligent approach for the identification of Indian medicinal plant leaves is presented. Herein, heterogeneous features extracted from leaves of Indian medicinal herbs serve as input attributes for classifier models. A bilateral approach is employed to bring out heterogeneous features. In the first approach, a framework for extracting handcrafted feature descriptors characterized by edge histograms, oriented gradients and binary patterns is proposed. In the second approach, deep features are extracted by Convolutional Neural Networks (CNN) using a transfer learning approach. Z score normalization is applied for the normalization of extracted features. Normalized features are fused using intermediate level fusion techniques to bring out heterogeneous features. 'Uniform Manifold Approximation and Projection (UMAP)' is employed to tackle the challenge offered by the high dimensional heterogeneous feature vector dataset. Finally, the reduced heterogeneous feature vectors are fed into different classifiers for the identification and classification of medicinal herbs. We compare the models' performances in terms of accuracy, precision, sensitivity, Probability of False Alarm (PFA), Matthews Correlation Coefficient Score (MCC), Jaccard Score and Receiver Operating Characteristics Curve (ROC). The results reveal that the Random Forest classifier with hybrid feature vector attains accuracy, precision and sensitivity of more than 99% and attains a very low PFA of 0.02%. Hence the proposed models with heterogeneous feature framework not only improve the identification and classification performance but at the same time reduce the probability of false alarm to substantial amounts. Our model presents remarkable results and marks up to 03% performance improvement in contrast to other available models and it demonstrates significant improvement in the identification and classification accuracy of Indian medicinal plant leaves. [ABSTRACT FROM AUTHOR]

Object detection in images has been identified as a critical area of research in computer vision image processing. Research has developed several novel methods for determining an object’s location and category from an image. However, there is still room for improvement in terms of detection efficiency. This study aims to develop a technique for detecting objects in images. To enhance overall detection performance, we considered object detection a two-fold problem, including localization and classification. The proposed method generates class-independent, high-quality, and precise proposals using an agglomerative clustering technique. We then combine these proposals with the relevant input image to train our network on convolutional features. Next, a network refinement module decreases the quantity of generated proposals to produce fewer high-quality candidate proposals. Finally, revised candidate proposals are sent into the network’s detection process to determine the object type. The algorithm’s performance is evaluated using publicly available the PASCAL Visual Object Classes Challenge 2007 (VOC2007), VOC2012, and Microsoft Common Objects in Context (MS-COCO) datasets. Using only 100 proposals per image at intersection over union (IoU) = 0.5 and 0.7), the proposed method attains Detection Recall (DR) rates of (93.17% and 79.35%) and (69.4% and 58.35%), and Mean Average Best Overlap (MABO) values of (79.25% and 62.65%), for the VOC2007 and MS-COCO datasets, respectively. Besides, it achieves a Mean Average Precision (mAP) of (84.7% and 81.5%) on both VOC datasets. The experiment findings reveal that our method exceeds previous approaches in terms of overall detection performance, proving its effectiveness. [ABSTRACT FROM AUTHOR]

Gait recognition is an active research area that uses a walking theme to identify the subject correctly. Human Gait Recognition (HGR) is performed without any cooperation from the individual. However, in practice, it remains a challenging task under diverse walking sequences due to the covariant factors such as normal walking and walking with wearing a coat. Researchers, over the years, have worked on successfully identifying subjects using different techniques, but there is still room for improvement in accuracy due to these covariant factors. This paper proposes an automated model-free framework for human gait recognition in this article. There are a few critical steps in the proposed method. Firstly, optical flow-based motion region estimation and dynamic coordinates-based cropping are performed. The second step involves training a fine-tuned pre-trained MobileNetV2 model on both original and optical flow cropped frames; the training has been conducted using static hyperparameters. The third step proposed a fusion technique known as normal distribution serially fusion. In the fourth step, a better optimization algorithm is applied to select the best features, which are then classified using a Bi-Layered neural network. Three publicly available datasets, CASIA A, CASIA B, and CASIA C, were used in the experimental process and obtained average accuracies of 99.6%, 91.6%, and 95.02%, respectively. The proposed framework has achieved improved accuracy compared to the other methods. [ABSTRACT FROM AUTHOR]

Background: Although handcrafted radiomics features (RF) are commonly extracted via radiomics software, employing deep features (DF) extracted from deep learning (DL) algorithms merits significant investigation. Moreover, a "tensor" radiomics paradigm where various flavours of a given feature are generated and explored can provide added value. We aimed to employ conventional and tensor DFs, and compare their outcome prediction performance to conventional and tensor RFs. Methods: 408 patients with head and neck cancer were selected from TCIA. PET images were first registered to CT, enhanced, normalized, and cropped. We employed 15 image-level fusion techniques (e.g., dual tree complex wavelet transform (DTCWT)) to combine PET and CT images. Subsequently, 215 RFs were extracted from each tumor in 17 images (or flavours) including CT only, PET only, and 15 fused PET-CT images through the standardized-SERA radiomics software. Furthermore, a 3 dimensional autoencoder was used to extract DFs. To predict the binary progression-free-survival-outcome, first, an end-to-end CNN algorithm was employed. Subsequently, we applied conventional and tensor DFs vs. RFs as extracted from each image to three sole classifiers, namely multilayer perceptron (MLP), random-forest, and logistic regression (LR), linked with dimension reduction algorithms. Results: DTCWT fusion linked with CNN resulted in accuracies of 75.6 ± 7.0% and 63.4 ± 6.7% in five-fold cross-validation and external-nested-testing, respectively. For the tensor RF-framework, polynomial transform algorithms + analysis of variance feature selector (ANOVA) + LR enabled 76.67 ± 3.3% and 70.6 ± 6.7% in the mentioned tests. For the tensor DF framework, PCA + ANOVA + MLP arrived at 87.0 ± 3.5% and 85.3 ± 5.2% in both tests. Conclusions: This study showed that tensor DF combined with proper machine learning approaches enhanced survival prediction performance compared to conventional DF, tensor and conventional RF, and end-to-end CNN frameworks. [ABSTRACT FROM AUTHOR]

Digital face manipulation has become a concern in the last few years due to its harmful impacts on society. It is especially concerning for high-profile celebrities because their identities can be easily manipulated using mobile or web applications such as FaceSwap and FaceApp. These manipulated faces are so close to real ones that it becomes really hard to detect them, even with bare eyes. Though deep learning-based models are predominantly used by researchers, they hardly check for the presence of irrelevant or redundant features produced by those models. To this end, we have proposed a hierarchical feature selection (HFS)-based method to detect deepfake images or videos. First, we have extracted both handcrafted and deep learning features from the inputs. Next, the HFS is applied to select a near optimal set of features. In each stage of it, a hybrid feature selection method is employed that integrates a population-based meta-heuristic model, called Grey Wolf Optimization, and a single solution-based meta-heuristic model, called the Vortex Search algorithm. We have evaluated our model on three publicly available datasets, namely Celeb-DF (V2), FaceForensics++, and Deepfake Detection Challenge (DFDC). The model provides 99.35%, 99.16%, and 85.67% AUC scores on the Celeb-DF (V2), FaceForensics++, and DFDC datasets, respectively, while using only 11.50%, 12.65%, and 10.22% of actual features. Besides, our model outperforms most of the state-of-the-art methods found in our literature review and evaluated on these three datasets. [ABSTRACT FROM AUTHOR]

In this paper, we propose an efficient fusion framework for brain magnetic resonance (MR) image classification using deep learning and handcrafted feature extraction methods, namely histogram of oriented gradients (HOG) and local binary patterns (LBP). The proposed framework aims to: (1) determine the optimal handcrafted features by Genetic Algorithm (GA) (2) discover the fully connected (FC) layers features using fine-tuned convolutional neural network (CNN) (3) employs the canonical correlation analysis (CCA) and the discriminant correlation analysis (DCA) methods in feature level fusion. Extensive experiments were conducted and demonstrated the classification performance on three benchmark datasets, viz., RD-DB1, TCIA-IXI-DB2 and TWB-HM-DB3. The mean accuracy of 68.69%, 90.35%, and 93.15% from CCA and 77.22%, 100.00%, and 99.40% from DCA was achieved by the Support Vector Machines (SVM) sigmoid kernel classifier on RD-DB1, TCIA-IXI-DB2, and TWB-HM-DB3 respectively. The obtained results of the proposed framework outperform when compared with other state-of-art works. [JJCIT 2022; 8(4.000): 336-344]

COVID-19 which was announced as a pandemic on 11 March 2020, is still infecting millions to date as the vaccines that have been developed do not prevent the disease but rather reduce the severity of the symptoms. Until a vaccine is developed that can prevent COVID-19 infection, the testing of individuals will be a continuous process. Medical personnel monitor and treat all health conditions; hence, the time-consuming process to monitor and test all individuals for COVID-19 becomes an impossible task, especially as COVID-19 shares similar symptoms with the common cold and pneumonia. Some off-the-counter tests have been developed and sold, but they are unreliable and add an additional burden because false-positive cases have to visit hospitals and perform specialized diagnostic tests to confirm the diagnosis. Therefore, the need for systems that can automatically detect and diagnose COVID-19 automatically without human intervention is still an urgent priority and will remain so because the same technology can be used for future pandemics and other health conditions. In this paper, we propose a modified machine learning (ML) process that integrates deep learning (DL) algorithms for feature extraction and well-known classifiers that can accurately detect and diagnose COVID-19 from chest CT scans. Publicly available datasets were made available by the China Consortium for Chest CT Image Investigation (CC-CCII). The highest average accuracy obtained was 99.9% using the modified ML process when 2000 features were extracted using GoogleNet and ResNet18 and using the support vector machine (SVM) classifier. The results obtained using the modified ML process were higher when compared to similar methods reported in the extant literature using the same datasets or different datasets of similar size; thus, this study is considered of added value to the current body of knowledge. Further research in this field is required to develop methods that can be applied in hospitals and can better equip mankind to be prepared for any future pandemics. [ABSTRACT FROM AUTHOR]

To solve the problem that a single kernel learning support vector machine (SVM) cannot consider the learning and generalization abilities and parameter optimization of the multiple kernel function, a multiple kernel learning support vector machine (MKL-SVM) algorithm based on swarm intelligence optimization was proposed. First, the impact of five single kernel functions on the classification indexes of SVM was discussed. These kernel functions include two global kernel functions — the polynomial and sigmoid kernel functions — and three local kernel functions—the radial basis function, exponential kernel function, and Laplacian kernel function. Next, an MKL-SVM algorithm with a convex combination of a polynomial kernel having global properties and a Laplacian kernel having local properties was proposed. Then, to improve particle diversity to avoid falling into local optimal solutions during the iteration, and to reduce the model’s training time, the crossover operation in the genetic algorithm was introduced into the particle swarm optimization (PSO) algorithm. This improved swarm intelligence optimization was used to optimize the parameters of the MKL-SVM. Finally, deep learning features based on the classical model VGG16 and handcrafted features according to doctors’ suggestions were used as inputs for the recognition algorithm. In this algorithm, transfer learning was used to extract deep learning features and principal component analysis was used to reduce computational complexity through dimensionality reduction. The results show that using deep learning features is better than handcrafted features. Therefore, this paper adopts the deep learning features as input for the MKL-SVM algorithm and the hybrid swarm intelligent optimization algorithm of crossover genetic and the PSO algorithm as the optimization method. To verify the generalization ability of the proposed algorithm, the public dataset LUNA16 was selected for testing. The experimental results show that the proposed algorithm is easy to jump out of the local optimal solution, improves the learning ability and generalization ability of the algorithm, and has a better classification performance.

Background: Although handcrafted radiomics features (RF) are commonly extracted via radiomics software, employing deep features (DF) extracted from deep learning (DL) algorithms merits significant investigation. Moreover, a “tensor’’ radiomics paradigm where various flavours of a given feature are generated and explored can provide added value. We aimed to employ conventional and tensor DFs, and compare their outcome prediction performance to conventional and tensor RFs. Methods: 408 patients with head and neck cancer were selected from TCIA. PET images were first registered to CT, enhanced, normalized, and cropped. We employed 15 image-level fusion techniques (e.g., dual tree complex wavelet transform (DTCWT)) to combine PET and CT images. Subsequently, 215 RFs were extracted from each tumor in 17 images (or flavours) including CT only, PET only, and 15 fused PET-CT images through the standardized-SERA radiomics software. Furthermore, a 3 dimensional autoencoder was used to extract DFs. To predict the binary progression-free-survival-outcome, first, an end-to-end CNN algorithm was employed. Subsequently, we applied conventional and tensor DFs vs. RFs as extracted from each image to three sole classifiers, namely multilayer perceptron (MLP), random-forest, and logistic regression (LR), linked with dimension reduction algorithms. Results: DTCWT fusion linked with CNN resulted in accuracies of 75.6 ± 7.0% and 63.4 ± 6.7% in five-fold cross-validation and external-nested-testing, respectively. For the tensor RF-framework, polynomial transform algorithms + analysis of variance feature selector (ANOVA) + LR enabled 76.67 ± 3.3% and 70.6 ± 6.7% in the mentioned tests. For the tensor DF framework, PCA + ANOVA + MLP arrived at 87.0 ± 3.5% and 85.3 ± 5.2% in both tests. Conclusions: This study showed that tensor DF combined with proper machine learning approaches enhanced survival prediction performance compared to conventional DF, tensor and conventional RF, and end-to-end CNN frameworks.

Environmental microorganism (EM) offers a highly efficient, harmless, and low-cost solution to environmental pollution. They are used in sanitation, monitoring, and decomposition of environmental pollutants. However, this depends on the proper identification of suitable microorganisms. In order to fasten, lower the cost, and increase consistency and accuracy of identification, we propose the novel pairwise deep learning features (PDLFs) to analyze microorganisms. The PDLFs technique combines the capability of handcrafted and deep learning features. In this technique, we leverage the Shi and Tomasi interest points by extracting deep learning features from patches which are centered at interest points' locations. Then, to increase the number of potential features that have intermediate spatial characteristics between nearby interest points, we use Delaunay triangulation theorem and straight line geometric theorem to pair the nearby deep learning features. The potential of pairwise features is justified on the classification of EMs using SVMs, Linear discriminant analysis, Logistic regression, XGBoost and Random Forest classifier. The pairwise features obtain outstanding results of 99.17%, 91.34%, 91.32%, 91.48%, and 99.56%, which are the increase of about 5.95%, 62.40%, 62.37%, 61.84%, and 3.23% in accuracy, F1-score, recall, precision, and specificity respectively, compared to non-paired deep learning features. [ABSTRACT FROM AUTHOR]

With the advancement in urbanization, it is extremely challenging to find a vacant parking space in a densely populated area. Many researchers experienced that traditional vision-based parking slot status detection approaches are more time consuming and cannot handle large video frames in practical applications, which diverted many researchers toward deep learning methods that can extract features and classify it. This study proposes an ensemble model based on deep learning to tackle weather condition effects. In most cases, Convolutional Neural Networks (CNN's) models used Red, Green, Blue frames, but it was observed that many more techniques like Hue Saturation Value, Discrete Wavelet Transform, Discrete Cosine Transform and GRADIENT also play an important role, if trained well using deep learning. Proposed study includes training of CNN's network considering five techniques followed by weighted fuzzy model. In this research, experiments have been conducted on a publicly available "PKLot" dataset, which consisted of 12,417 images which have been segmented into 695,899 images such that each segmented image represents the individual parking slot. Images have been taken from three different camera locations in two different parking lots in Brazil, under different weather conditions such as sunny, cloudy and rainy. The experimental results show significant improvement in terms of accuracy as compared with existing state-of-art methods. In the end, real-time images of outdoor parking lot of Panjab University Swami Sarvanand Giri Regional Centre, Hoshiarpur, Punjab, India have been collected for implementation of the proposed model. [ABSTRACT FROM AUTHOR]

A brain tumor is an abnormal growth of brain cells inside the head, which reduces the patient's survival chance if it is not diagnosed at an earlier stage. Brain tumors vary in size, different in type, irregular in shapes and require distinct therapies for different patients. Manual diagnosis of brain tumors is less efficient, prone to error and time-consuming. Besides, it is a strenuous task, which counts on radiologist experience and proficiency. Therefore, a modern and efficient automated computer-assisted diagnosis (CAD) system is required which may appropriately address the aforementioned problems at high accuracy is presently in need. Aiming to enhance performance and minimise human efforts, in this manuscript, the first brain MRI image is pre-processed to improve its visual quality and increase sample images to avoid over-fitting in the network. Second, the tumor proposals or locations are obtained based on the agglomerative clustering-based method. Third, image proposals and enhanced input image are transferred to backbone architecture for features extraction. Fourth, high-quality image proposals or locations are obtained based on a refinement network, and others are discarded. Next, these refined proposals are aligned to the same size, and finally, transferred to the head network to achieve the desired classification task. The proposed method is a potent tumor grading tool assessed on a publicly available brain tumor dataset. Extensive experiment results show that the proposed method outperformed the existing approaches evaluated on the same dataset and achieved an optimal performance with an overall classification accuracy of 98.04%. Besides, the model yielded the accuracy of 98.17, 98.66, 99.24%, sensitivity (recall) of 96.89, 97.82, 99.24%, and specificity of 98.55, 99.38, 99.25% for Meningioma, Glioma, and Pituitary classes, respectively.

Nowadays, Deep learning (DL) is the growing trend towards creating visual representations of human body organs for clinical analysis, medical interventions as well as to diagnose and treat diseases. This paper propose a method for neonatal and pediatric brain tumors image analysis and prerequisites a T2-weighted MR images only. The pipeline stages of the proposed work as follows: In the first stage, designed a set of specific feature vectors description for high-level classification task using Conventional and deep learning (DL) Feature Extraction methods. The second stage, select a deep features based on proposed convolutional neural network (CNN) method and conventional subset features are from Genetic Algorithm (GA). The third stage, merge the selected features by adapting fusion technique. Finally, predict the brain image is either normal or abnormal. The results demonstrated that the proposed method obtained accurate classification and revealed its robustness to difference in ages and acquisition protocols. The obtained results shows that based on combined deep learning features (DLF) and conventional features have been significantly improves the classification accuracy of the support vector machines (SVM) classifier up to 97.00%. [ABSTRACT FROM AUTHOR]

Background: Positron emission tomography (PET) is extensively employed for diagnosing and staging various tumors, including liver cancer, lung cancer, and lymphoma. Accurate subtype classification of tumors plays a crucial role in formulating effective treatment plans for patients. Notably, lymphoma comprises subtypes like diffuse large B-cell lymphoma and Hodgkin's lymphoma, while lung cancer encompasses adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. Similarly, liver cancer consists of subtypes such as cholangiocarcinoma and hepatocellular carcinoma. Consequently, the subtype classification of tumors based on PET images holds immense clinical significance. However, in clinical practice, the number of cases available for each subtype is often limited and imbalanced. Therefore, the primary challenge lies in achieving precise subtype classification using a small dataset., Method: This paper presents a novel approach for tumor subtype classification in small datasets using RA-DL (Radiomics-DeepLearning) attention. To address the limited sample size, Support Vector Machines (SVM) is employed as the classifier for tumor subtypes instead of deep learning methods. Emphasizing the importance of texture information in tumor subtype recognition, radiomics features are extracted from the tumor regions during the feature extraction stage. These features are compressed using an autoencoder to reduce redundancy. In addition to radiomics features, deep features are also extracted from the tumors to leverage the feature extraction capabilities of deep learning. In contrast to existing methods, our proposed approach utilizes the RA-DL-Attention mechanism to guide the deep network in extracting complementary deep features that enhance the expressive capacity of the final features while minimizing redundancy. To address the challenges of limited and imbalanced data, our method avoids using classification labels during deep feature extraction and instead incorporates 2D Region of Interest (ROI) segmentation and image reconstruction as auxiliary tasks. Subsequently, all lesion features of a single patient are aggregated into a feature vector using a multi-instance aggregation layer., Result: Validation experiments were conducted on three PET datasets, specifically the liver cancer dataset, lung cancer dataset, and lymphoma dataset. In the context of lung cancer, our proposed method achieved impressive performance with Area Under Curve (AUC) values of 0.82, 0.84, and 0.83 for the three-classification task. For the binary classification task of lymphoma, our method demonstrated notable results with AUC values of 0.95 and 0.75. Moreover, in the binary classification task of liver tumor, our method exhibited promising performance with AUC values of 0.84 and 0.86., Conclusion: The experimental results clearly indicate that our proposed method outperforms alternative approaches significantly. Through the extraction of complementary radiomics features and deep features, our method achieves a substantial improvement in tumor subtype classification performance using small PET datasets., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

COVID-19 which was announced as a pandemic on 11 March 2020, is still infecting millions to date as the vaccines that have been developed do not prevent the disease but rather reduce the severity of the symptoms. Until a vaccine is developed that can prevent COVID-19 infection, the testing of individuals will be a continuous process. Medical personnel monitor and treat all health conditions; hence, the time-consuming process to monitor and test all individuals for COVID-19 becomes an impossible task, especially as COVID-19 shares similar symptoms with the common cold and pneumonia. Some off-the-counter tests have been developed and sold, but they are unreliable and add an additional burden because false-positive cases have to visit hospitals and perform specialized diagnostic tests to confirm the diagnosis. Therefore, the need for systems that can automatically detect and diagnose COVID-19 automatically without human intervention is still an urgent priority and will remain so because the same technology can be used for future pandemics and other health conditions. In this paper, we propose a modified machine learning (ML) process that integrates deep learning (DL) algorithms for feature extraction and well-known classifiers that can accurately detect and diagnose COVID-19 from chest CT scans. Publicly available datasets were made available by the China Consortium for Chest CT Image Investigation (CC-CCII). The highest average accuracy obtained was 99.9% using the modified ML process when 2000 features were extracted using GoogleNet and ResNet18 and using the support vector machine (SVM) classifier. The results obtained using the modified ML process were higher when compared to similar methods reported in the extant literature using the same datasets or different datasets of similar size; thus, this study is considered of added value to the current body of knowledge. Further research in this field is required to develop methods that can be applied in hospitals and can better equip mankind to be prepared for any future pandemics.

Automating hate speech or inappropriate text detection in social media and other internet platforms is gaining a lot of interest and becoming a valuable research topic for both industry and academia in recent years. It is more important for applications to identify the disruptive contents, understand sentiment analysis, identify cyber bullying, detect flames, threats, hatred towards people or particular communities or groups etc. Text classification is a very challenging task due to the nature and complexities with languages, especially its context, micro words, emojis, typo error and sarcasm present in the text. In this paper, we have proposed a model with a novel approach for generating hybrid features for an effective feature representation to classify hate speech. We have combined features learned from deep learning methods with the semantic features like word n-grams and tweets specific syntactic features to form hybrid feature sets. We have also improvised preprocessing steps to reduce the number of missing embeddings to increase the vocabulary for efficient feature learning. We have experimented with the various neural networks for feature learning and machine learning models with hybrid features for classification. Our work delivers hybrid features and appropriate preprocessing techniques for an efficient classification of the standard dataset of 16k annotated hate speech tweets. The combination of Long Short Term Memory (LSTM) trained on Random Embeddings for deep learning features extraction and Logistic Regression (LR) as a classifier with the hybrid features is found to be the best model and it outperforms the state of the art reported in the literature.

In recent years, view-based 3D model retrieval has become one of the research focuses in the field of computer vision and machine learning. In fact, the 3D model retrieval algorithm consists of feature extraction and similarity measurement, and the robust features play a decisive role in the similarity measurement. Although deep learning has achieved comprehensive success in the field of computer vision, deep learning features are used for 3D model retrieval only in a small number of works. To the best of our knowledge, there is no benchmark to evaluate these deep learning features. To tackle this problem, in this work we systematically evaluate the performance of deep learning features in view-based 3D model retrieval on four popular datasets (ETH, NTU60, PSB, andMVRED) by different kinds of similarity measure methods. In detail, the performance of hand-crafted features and deep learning features are compared, and then the robustness of deep learning features is assessed. Finally, the difference between single-view deep learning features and multi-view deep learning features is also evaluated. By quantitatively analyzing the performances on different datasets, it is clear that these deep learning features can consistently outperform all of the hand-crafted features, and they are also more robust than the hand-crafted features when different degrees of noise are added into the image. The exploration of latent relationships among different views in multi-view deep learning network architectures shows that the performance of multi-view deep learning outperforms that of single-view deep learning features with low computational complexity. [ABSTRACT FROM AUTHOR]

In traditional single image super-resolution (SR) methods based on dictionary model, a large number of image features are needed to train the SR dictionary. In general, these features are extracted by artificial rules, such as pixel gray, gradient, and texture structure. But, the dictionary model trained by these artificial features or their combinations has exhibited poor expression especially for the images with complex and rich structures. Therefore, how to improve the dictionary expression ability and make the dictionary have more accurate description of the image features is a problem worthy of further study. In this paper, based on the advantage of dictionary training and deep learning, a new method of single image SR based on deep learning features and dictionary model is proposed. The new algorithm contains three steps. First, the features of high-resolution and low-resolution training images are extracted by a Kernel deep learning network. Second, in the sparse representation of SR framework, the dictionary model is trained by these deep learning features. Finally, an LR image SR is completed. Theoretical analysis show that the dictionary trained by deep learning features can improve in the ability to express image complex structure and texture, and it has more advantage than traditional artificial features dictionary. The experimental results indicate that the proposed algorithm can produce good SR visual results than the comparison algorithm, such as Bicubic, sparse coding super-resolution, and super-resolution convolutional neural network. And the peak signal to noise ratio and structural similarity index measurement are improved, the Computation Time is also reasonable.

Facial expression recognition plays an important role in the field involving human-computer interactions. Given the wide use of convolutional neural networks or other neural network models in automatic image classification systems, high-level features can be automatically learned by hierarchical neural networks. However, the training of CNNs requires large amounts of training data to permit adequate generalization. The traditional scale-invariant feature transform (SIFT) does not need large learning samples to obtain features. In this paper, we proposed a feature extraction method for use in the facial expressions recognition from a single image frame. The hybrid features use a combination of SIFT and deep learning features of different levels extracted from a CNN model. The combined features are adopted to classify expressions using support vector machines. The performance of proposed method is tested using the publicly available extended Cohn-Kanade (CK+) database. To evaluate the generalization ability of our method, several experiments are designed and carried out in a cross-database environment. Compared with the 76.57% accuracy obtained using SIFT-bag of features (BoF) features and the 92.87% accuracy obtained using CNN features, we achieve a FER accuracy of 94.82% using the proposed hybrid SIFT-CNN features. The results of additional cross-database experiments also demonstrate the considerable potential of combining shallow features with deep learning features, and these results are more promising than state-of-the-art models. Combining shallow and deep learning features is effective when the training data are not sufficient to obtain a deep model with considerable generalization ability. [ABSTRACT FROM AUTHOR]

Recently, scene based classification has become a new trend for very high spatial resolution remote sensing image interpretation. With the advent of deep learning, the pretrained convolutional neural networks (CNNs) have been proved effective as feature extractors for scene classification tasks in the remote sensing domain, but the potential characteristics and capabilities of such deep features have not been sufficiently analyzed and fully understood. Facing with complex remote sensing scenes with huge intra-class variations, it is still not clear about the limitation of these powerful deep features in exploring essential invariant attributes of remote sensing scenes of the same kind but, in most cases, from separate sources. Therefore, this paper makes an intensive investigation in the feature representation ability of such deep features from the aspect of inter-dataset scene classification of remote sensing images. Four well-known pretrained CNN models and three different commonly used datasets are selected and summarized. Firstly, deep features extracted from various intermediate layers of these models are compared. Then, the inter-dataset feature representation ability is evaluated using cross-classification of different datasets and discussed in terms of imaging spatial resolution, image size, model structure, and time efficiency. Finally, several instructive findings are revealed and conclusions are drawn regarding the strength and weakness of the CNN features in the application of remote sensing image scene classification. [ABSTRACT FROM AUTHOR]

• Addresses the need for rapid and accurate brain tumor detection to improve patient outcomes. • Introduces a less labor-intensive approach to diagnose brain cancers using MRI imaging. • Trains and evaluates the system using a publicly accessible brain tumor dataset, achieving an overall accuracy of approximately 98.98%. • Enhances resilience, automation, and accuracy of the diagnostic process, offering significant benefits to the healthcare industry. The accurate and rapid detection of brain tumors is crucial for expediting patient rehabilitation and saving lives. Brain tumors exhibit considerable variation in size, shape, and appearance, necessitating patient-specific treatment options. Radiologists are required to possess a high level of experience and skill to manually diagnose brain cancers. However, manual tumor detection can be inefficient, error-prone, and time-consuming. Developing a practical solution to overcome these limitations is essential. This work presents a less labor-intensive approach to diagnosing brain cancers using MRI imaging. The proposed method enhances an image's visual quality by employing a low-complexity algorithm. Prior to segmentation, morphological analysis is conducted to exclude non-tumor regions from the images. Employing segmentation and clustering techniques, high-quality tumor regions are identified. Based on their rankings, multiple deep neural networks extract features from these regions. An adaptive fusion network creates a hybrid feature vector, which is then used in conjunction with multi-class SVM to classify tumors. By supplementing the training data, overfitting is minimized. The presented system is trained and evaluated using a publicly accessible brain tumor dataset. In comparison to previous approaches, the proposed solution enhances the resilience and automation of the entire diagnostic process, enabling the healthcare industry to achieve an overall accuracy of approximately 98.98%. [ABSTRACT FROM AUTHOR]

Manual identification of brain tumors is an error-prone and tedious process for radiologists; therefore, it is crucial to adopt an automated system. The binary classification process, such as malignant or benign is relatively trivial; whereas, the multimodal brain tumors classification (T1, T2, T1CE, and Flair) is a challenging task for radiologists. Here, we present an automated multimodal classification method using deep learning for brain tumor type classification. The proposed method consists of five core steps. In the first step, the linear contrast stretching is employed using edge-based histogram equalization and discrete cosine transform (DCT). In the second step, deep learning feature extraction is performed. By utilizing transfer learning, two pre-trained convolutional neural network (CNN) models, namely VGG16 and VGG19, were used for feature extraction. In the third step, a correntropy-based joint learning approach was implemented along with the extreme learning machine (ELM) for the selection of best features. In the fourth step, the partial least square (PLS)-based robust covariant features were fused in one matrix. The combined matrix was fed to ELM for final classification. The proposed method was validated on the BraTS datasets and an accuracy of 97.8%, 96.9%, 92.5% for BraTs2015, BraTs2017, and BraTs2018, respectively, was achieved.

Facial expression recognition plays a crucial role in a wide range of applications of psychotherapy, security systems, marketing, commerce and much more. Detecting a macro-expression, which is a direct representation of an ‘emotion,’ is a relatively straight-forward task. Playing a pivotal role as macro-expressions, micro-expressions are more accurate indicators of a train of thought or even subtle, passive or involuntary thoughts. Compared to macro-expressions, identifying micro-expressions is a much more challenging research question because their time spans are narrowed down to a fraction of a second, and can only be defined using a broader classification scale. This paper is an all-inclusive survey-cum-analysis of the various micro-expression recognition techniques. We analyze the general framework for micro-expression recognition system by decomposing the pipeline into fundamental components, namely face detecting, pre-processing, facial feature detection and extraction, datasets, and classification. We discuss the role of these elements and highlight the models and new trends that are followed in their design. Moreover, we provide an extensive analysis of micro-expression recognition systems by comparing their performance. We also discuss the new deep learning features that can, in the near future, replace the hand-crafted features for facial micro-expression recognition. This survey has been developed, focusing on the methodologies applied, databases used, performance regarding recognition accuracy and comparing these to distil the gaps in the efficiencies, future scope, and research potentials. Through this survey, we intend to look into this problem and develop a comprehensive and efficient recognition scheme. This study allows us to identify open issues and to determine future directions for designing real-world micro-expression recognition systems. [ABSTRACT FROM AUTHOR]

The fields of human activity analysis have recently begun to diversify. Many researchers have taken much interest in developing action recognition or action prediction methods. The research on human action evaluation differs by aiming to design computation models and evaluation approaches for automatically assessing the quality of human actions. This line of study has become popular because of its explosively emerging real-world applications, such as physical rehabilitation, assistive living for elderly people, skill training on self-learning platforms, and sports activity scoring. This paper presents a comprehensive survey of approaches and techniques in action evaluation research, including motion detection and preprocessing using skeleton data, handcrafted feature representation methods, and deep learning-based feature representation methods. The benchmark datasets from this research field and some evaluation criteria employed to validate the algorithms’ performance are introduced. Finally, the authors present several promising future directions for further studies.

• An improved automated method for classifying brain tumors is proposed. • An effective way to enhance the visual quality of MRI images is utilized. • A system for locating objects (tumors) generates fewer but better proposals were developed. • The hybrid feature vector is generated to improve the overall classification performance. • The impact of overfitting on classification performance was explored. • Comparisons with existing methodologies demonstrated that this strategy had greater classification precision. Brain tumors can be fatal if not detected early enough. Manually diagnosing brain tumors requires the radiologist's experience and expertise, which may not always be available. Furthermore, manual processes are inefficient, prone to errors, and time-taking. Therefore, an effective solution is required to ensure an accurate diagnosis. To this end, we propose an automated technique for detecting brain tumors using magnetic resonance imaging (MRI). First, brain MRI images are pre-processed to enhance visual quality. Second, we apply two different pre-trained deep learning models to extract powerful features from images. The resulting feature vectors are then combined to form a hybrid feature vector using the partial least squares (PLS) method. Third, the top tumor locations are revealed via agglomerative clustering. Finally, these proposals are aligned to a predetermined size and then relayed to the head network for classification. Compared to existing approaches, the proposed method achieved a classification accuracy of 98.95%. [Display omitted] [ABSTRACT FROM AUTHOR]

• This research focuses on soybean cultivar identification which is known as a very fine-grained classification problem. • The first attempt on fusing deep features of triplet leaf image patterns is reported. • Two deep feature fusion methods, distance fusion and classifier fusion, are proposed. • A novel fine-tuning strategy that addresses the issue of few-shot soybean leaf image classification is designed. • The proposed method achieves an exciting accuracy of 83.55% on classifying 200 soybean cultivars. Soybean cultivar recognition plays a vital role in cultivar evaluation, selection and production. Recently, there is an increasing interest in taking leaf image patterns as clues for distinguishing soybean cultivars. However, due to the higher inter-class similarity of soybean cultivars over plant species, the cultivar classification accuracies reported by the existing methods are far lower than those published on plant species recognition which make computer vision community have a concern whether leaf image patterns can provide sufficient discriminative information for identifying soybean cultivars. In this paper, we explore fusing deep learning features of leaves from different parts of soybean plants for achieving an accurate cultivar recognition. In our method, the deep learning features of triplet leave image patterns that consists of leaves from the lower, middle, and upper parts of soybean plants are fused by two methods, distance fusion and classifier fusion. In the former, the L 1 distance measurements defined on the deep feature spaces of triplet leaf image patterns are fused prior to using 1NN classifier for classification. While in the later, the SVM classifiers trained by the deep features of triple leaf image patterns are combined by sum rule for cultivar prediction. We use the SoyCultivar200 leaf dataset which consists of 6000 samples from 200 soybean cultivars as benchmark. Our method achieves an exciting classification rate of 83.55% which demonstrates that our proposed fusion of deep features of triplet leaf image patterns can provide strong discriminative information for accurately identifying soybean cultivars. [ABSTRACT FROM AUTHOR]

Following the popularisation of media streaming, a number of video streaming services are continuously buying new video content to mine the potential profit from them. As such, the newly added content has to be handled well to be recommended to suitable users. In this paper, we address the new item cold-start problem by exploring the potential of various deep learning features to provide video recommendations. The deep learning features investigated include features that capture the visual-appearance, audio and motion information from video content. We also explore different fusion methods to evaluate how well these feature modalities can be combined to fully exploit the complementary information captured by them. Experiments on a real-world video dataset for movie recommendations show that deep learning features outperform hand-crafted features. In particular, recommendations generated with deep learning audio features and action-centric deep learning features are superior to MFCC and state-of-the-art iDT features. In addition, the combination of various deep learning features with hand-crafted features and textual metadata yields significant improvement in recommendations compared to combining only the former. • Multimodal video content is fused to alleviate the item cold-start problem. • Performance of various CNN features (object, scene, audio and action) are evaluated. • CNN features outperform hand-crafted iDT, MoSIFT and MFCC features, and metadata. • Metadata, CNN and hand-crafted features are combined to enhance recommendations. • A hybrid model is improved with matrix scaling to boost the overall performance. [ABSTRACT FROM AUTHOR]

A brain tumor is an abnormal growth of brain cells inside the head, which reduces the patient's survival chance if it is not diagnosed at an earlier stage. Brain tumors vary in size, different in type, irregular in shapes and require distinct therapies for different patients. Manual diagnosis of brain tumors is less efficient, prone to error and time-consuming. Besides, it is a strenuous task, which counts on radiologist experience and proficiency. Therefore, a modern and efficient automated computer-assisted diagnosis (CAD) system is required which may appropriately address the aforementioned problems at high accuracy is presently in need. Aiming to enhance performance and minimise human efforts, in this manuscript, the first brain MRI image is pre-processed to improve its visual quality and increase sample images to avoid over-fitting in the network. Second, the tumor proposals or locations are obtained based on the agglomerative clustering-based method. Third, image proposals and enhanced input image are transferred to backbone architecture for features extraction. Fourth, high-quality image proposals or locations are obtained based on a refinement network, and others are discarded. Next, these refined proposals are aligned to the same size, and finally, transferred to the head network to achieve the desired classification task. The proposed method is a potent tumor grading tool assessed on a publicly available brain tumor dataset. Extensive experiment results show that the proposed method outperformed the existing approaches evaluated on the same dataset and achieved an optimal performance with an overall classification accuracy of 98.04%. Besides, the model yielded the accuracy of 98.17, 98.66, 99.24%, sensitivity (recall) of 96.89, 97.82, 99.24%, and specificity of 98.55, 99.38, 99.25% for Meningioma, Glioma, and Pituitary classes, respectively.

Manual identification of brain tumors is an error-prone and tedious process for radiologists; therefore, it is crucial to adopt an automated system. The binary classification process, such as malignant or benign is relatively trivial; whereas, the multimodal brain tumors classification (T1, T2, T1CE, and Flair) is a challenging task for radiologists. Here, we present an automated multimodal classification method using deep learning for brain tumor type classification. The proposed method consists of five core steps. In the first step, the linear contrast stretching is employed using edge-based histogram equalization and discrete cosine transform (DCT). In the second step, deep learning feature extraction is performed. By utilizing transfer learning, two pre-trained convolutional neural network (CNN) models, namely VGG16 and VGG19, were used for feature extraction. In the third step, a correntropy-based joint learning approach was implemented along with the extreme learning machine (ELM) for the selection of best features. In the fourth step, the partial least square (PLS)-based robust covariant features were fused in one matrix. The combined matrix was fed to ELM for final classification. The proposed method was validated on the BraTS datasets and an accuracy of 97.8%, 96.9%, 92.5% for BraTs2015, BraTs2017, and BraTs2018, respectively, was achieved. [ABSTRACT FROM AUTHOR]

The fields of human activity analysis have recently begun to diversify. Many researchers have taken much interest in developing action recognition or action prediction methods. The research on human action evaluation differs by aiming to design computation models and evaluation approaches for automatically assessing the quality of human actions. This line of study has become popular because of its explosively emerging real-world applications, such as physical rehabilitation, assistive living for elderly people, skill training on self-learning platforms, and sports activity scoring. This paper presents a comprehensive survey of approaches and techniques in action evaluation research, including motion detection and preprocessing using skeleton data, handcrafted feature representation methods, and deep learning-based feature representation methods. The benchmark datasets from this research field and some evaluation criteria employed to validate the algorithms' performance are introduced. Finally, the authors present several promising future directions for further studies. [ABSTRACT FROM AUTHOR]