Search

Wall Shear Stress (WSS) is one of the most important parameters used in cardiovascular fluid mechanics, and it provides a lot of information like the risk level caused by any vascular occlusion. Since WSS cannot be measured directly and other available relevant methods have issues like low resolution, uncertainty and high cost, this study proposes a novel method by combining computational fluid dynamics (CFD), fluid-structure interaction (FSI), conditional generative adversarial network (cGAN) and convolutional neural network (CNN) to predict coronary artery occlusion risk using only noninvasive images accurately and rapidly. First, a cGAN model called WSSGAN was developed to predict the WSS contours on the vessel wall by training and testing the model based on the calculated WSS contours using coupling CFD-FSI simulations. Then, an 11-layer CNN was used to classify the WSS contours into three grades of occlusions, i.e. low risk, medium risk and high risk. To verify the proposed method for predicting the coronary artery occlusion risk in a real case, the patient’s Magnetic Resonance Imaging (MRI) images were converted into a 3D geometry for use in the WASSGAN model. Then, the predicted WSS contours by the WSSGAN were entered into the CNN model to classify the occlusion grade. [ABSTRACT FROM AUTHOR]

Optical character recognition is a way of converting scanned images of printed or handwritten documents into machine-encoded text, making it easier to store, browse, retrieve, and process electronic data. In this research, a Printed Amharic Characters Recognition dataset is prepared to train and test a model. Images in the dataset only contain 231 basic Amharic characters that are normalized to 32 × 32 pixels. In this work, a hybrid model of the two super classifiers is developed: the convolutional neural network (CNN) and the support vector machine (SVM). In this novel hybrid CNN-SVM model, CNN works as an automatic feature extractor from the raw images, and then, the extracted feature vectors are given as input to SVM for classification and recognition. A 99.84% accuracy was achieved on the own-prepared dataset and 95.59% accuracy on the benchmark Amharic Optical Character Recognition Database in classifying the testing dataset images. The proposed hybrid CNN-SVM model gave better results than the CNN with a fully connected layer. Moreover, the proposed model outperforms previously existing works attempted by others to recognize printed Amharic characters on the same and different datasets. [ABSTRACT FROM AUTHOR]

Lip print recognition technology originated in the field of forensic medicine, and convolutional neural network has made breakthrough achievements in the field of pattern recognition and machine vision. Convolutional neural network (CNN) algorithm is rarely used in lip pattern recognition. Further exploration and research on the network model suitable for lip pattern recognition. Lip print recognition algorithm based on depth convolution neural network aims to solve the problems of complex image preprocessing, difficult feature extraction and low recognition efficiency in traditional lip print recognition algorithms. It includes collecting lip print images to establish data sets, selecting different CNN models to conduct performance evaluation experiments on low resolution lip print data sets, and analyzing the experimental results with model evaluation indicators. [ABSTRACT FROM AUTHOR]

Currently, there is a notable prevalence of substantial traffic congestion and frequent vehicular accidents on roadways in contemporary times. Amalgamation of latest front-line technologies involving Internet of Things (IoT) and image classification has immense potential to advance the progress of a proficient traffic regulation system. To mitigate the occurrence of vehicular accidents, our research endeavors revolve around the comprehensive analysis of the prevailing road conditions. This meticulous examination allows us to effectively automate traffic routes orchestration, ensuring smooth vehicular movement across all lanes of the road network. The analysis of traffic patterns is conducted through the utilization of visual data images. The real time captured traffic images undergo processing using various object detection models named RetinaNet and the YOLO (You Only Look Once) models. A series of comparative evaluations suggests an improved traffic object identification capacity for the RetinaNet model as compared to all YOLO models. [ABSTRACT FROM AUTHOR]

Plant diseases can harm crops and reduce the amount of food that can be cultivated, which is problematic for farmers. Technology is being utilized to develop computer-based programs that can recognize plant diseases and assist farmers in making better decisions after identifying plant leaf diseases. In most of these models, machine learning algorithms are applied, to make predictions about potential plant diseases using mathematical models and neural networks. Many researchers discussed the variants of DNN and CNN algorithms to solve the discussed problems and gave better results. In this paper, the novel approach is discussed and implemented where the plant disease is identified whether the plant leaf captured image has a noisy background or not; or whether the leaf image is segmented or not. The authors developed an adaptive algorithm which gives the results in two phases: the classification of the plant disease based on the original input leaf image and secondly, the identification of plant leaf disease after applying the segmentation process. The result of this two-phase proposed model is analyzed and compared with existing popular models like AlexNet, ResNet-50, and the EffNet the results are convincing. The proposed model has 97.39% accuracy when the noiseless image is taken; while the 90.26% accuracy is there, in case of noisy background image as an input; and the results are outstanding, if the authors are applying their segmentation-based AH-CNN model on the noisy real-time image, the accuracy is 95.27%. [ABSTRACT FROM AUTHOR]

Abstract Wall Shear Stress (WSS) is one of the most important parameters used in cardiovascular fluid mechanics, and it provides a lot of information like the risk level caused by any vascular occlusion. Since WSS cannot be measured directly and other available relevant methods have issues like low resolution, uncertainty and high cost, this study proposes a novel method by combining computational fluid dynamics (CFD), fluid-structure interaction (FSI), conditional generative adversarial network (cGAN) and convolutional neural network (CNN) to predict coronary artery occlusion risk using only noninvasive images accurately and rapidly. First, a cGAN model called WSSGAN was developed to predict the WSS contours on the vessel wall by training and testing the model based on the calculated WSS contours using coupling CFD-FSI simulations. Then, an 11-layer CNN was used to classify the WSS contours into three grades of occlusions, i.e. low risk, medium risk and high risk. To verify the proposed method for predicting the coronary artery occlusion risk in a real case, the patient’s Magnetic Resonance Imaging (MRI) images were converted into a 3D geometry for use in the WASSGAN model. Then, the predicted WSS contours by the WSSGAN were entered into the CNN model to classify the occlusion grade.

Coronary artery disease (CAD) is a pathological condition that is often fatal and is the main cause of death throughout the world. Early detection of this disease is very important to avoid severe complications such as heart attacks and sudden death. This study employs artificial intelligence, specifically deep learning via Convolutional Neural Networks (CNNs), to enhance CAD detection. While CNN architectures like ResNet50V2 and MobileNetV2 exhibit satisfactory performance individually, they possess distinct strengths and weaknesses. ResNet50V2 requires significant computing resources, hindering its scalability, while MobileNetV2 struggles with extracting complex features from medical images. Therefore, this research aims to combine the EfficientNetV2B0, ResNet50V2, and MobileNetV2 using transfer learning techniques to enhance CAD detection. The methodology involves leveraging pre-trained models and fine-tuning them on a coronary artery disease dataset. Modified models, particularly EfficientNetV2B0 and MobileNetV2, achieve high accuracies of 94% and 86%, respectively, while ResNet50V2 yields 72%. However, combining the models boosts accuracy to 95%, addressing individual model limitations. The concatenated model demonstrates superior predictive capabilities, with more accurate predictions and fewer errors than individual models. [ABSTRACT FROM AUTHOR]

The fast and accurate detection of abnormal cardiac episodes is essential for quick diagnosis high-risk patients prone to irregular cardiac arrhythmias. Although various algorithms have been proposed in the literature, there is a need to develop artificial intelligence driven computer aided intelligent models with high detection rate. This specific study focuses on the application of Convolution Neural Network (CNN) approach to classify abnormal and ventricular ectopic beat (VEB) episodes from normal ECG signals. A cross-database approach comprised of open source and a proprietary database has been deployed for training–testing strategies. Single-channel ECG time series was converted into a derivative image stack before applying it to CNN. The current study has proposed extracting derivative image features using a pretrained network and classification using the SVM classifier. The proposed study was tested on 18 pretrained networks such as Resnet18, Resnet50, Resnet101, Densnet201, InceptionV3, Googlenet, Squeeznet, Vgg19, Vgg16, Alexnet, InceptionResnetV2, Shufflenet, MobilenetV2, NASnet-Mobile, Darknet-19, Draknet-53, Xception, and NASnet-Large for optimization. Among the chosen pretrained models, Mobilenetv2 has shown the highest classification accuracy of 92.73% & 99.29%, with relative performance of 1.24 & 0.95 for the normal-abnormal and normal-VEB cardiac episode classifications of opensource datasets. The highest accuracy of 73.66% (using xception) and 99.92% (using Inceptionv3) was achieved for abnormal and VEB cardiac episodes detection from proprietary datasets. In addition, the proposed method showed the highest classification accuracy of 90.14% with Inceptionv3 and 99.34% with darknet19 using the proprietary database for normal-abnormal & N-V cardiac episode classification. Further, inceptionv3 & darknet19 trained models using the proprietary database showed the highest detection rate of 81.67% & 94.66% for abnormal & ventricular ectopic beat cardiac episodes of the open-source database. The study has concluded that the abnormal and VEB cardiac episode detection using CNN-based models could help diagnose and treat cardiac patients. [ABSTRACT FROM AUTHOR]

Hand gestures (HGs) have been extensively acknowledged as a communication standard for the deaf and dumb school system in the Indian sign language (ISL). We suggested an automated Indian sign language recognition (ISLR) method for English in this study. Initially, the hand region is segmented using the Grasshopper optimization algorithm (GOA) based on the skin color model method. The segmentation’s efficacy is further assessed using three techniques: GOA-based skin color detection algorithm (SCDA), particle swarm optimization-based SCDA (PSO-SCDA), and artificial bee colony-based SCDA (ABC-SCDA). Finally, a database for the retrieved motions representing distinct English alphabets is created. For gesture recognition, the system is trained, and a template-based matching approach is used. The classification technique employs the support vector machine (SVM) and the convolution neural network (CNN). Our suggested recognition method achieved the highest accuracy of 97.85% with GOA-SCDA, compared to 89.29% and 93.96% with PSO-SCDA and ABC-SCDA, respectively. Furthermore, CNN outperformed SVM in classification, with a high accuracy of 99.2% and precision of 81.8%. [ABSTRACT FROM AUTHOR]

One of the most essential components of human life is sleep. One of the first steps in spotting abnormalities connected to sleep is classifying sleep stages. Based on the kind and frequency of signals obtained during a polysomnography test, sleep phases can be separated into groups. Accurate classification of sleep stages from electroencephalogram (EEG) signals plays a crucial role in sleep disorder diagnosis and treatment. This study proposes a novel approach that combines feature selection techniques with convolutional neural networks (CNNs) to enhance the classification performance of sleep stages using EEG signals. Firstly, a comprehensive feature selection process was employed to extract discriminative features from raw EEG data, aiming to reduce dimensionality and enhance the efficiency of subsequent classification using mutual information (MI) and analysis of variance (ANOVA) after splitting the dataset into two sets—the training set (70%) and testing set (30%)—then processing it using the standard scalar method. Subsequently, a 1D-CNN architecture was designed to automatically learn hierarchical representations of the selected features, capturing complex patterns indicative of different sleep stages. The proposed method was evaluated on a publicly available EDF-Sleep dataset, demonstrating superior performance compared to traditional approaches. The results highlight the effectiveness of integrating feature selection with CNNs in improving the accuracy and reliability of sleep stage classification from EEG signals, which reached 99.84% with MI-50. This approach not only contributes to advancing the field of sleep disorder diagnosis, but also holds promise for developing more efficient and robust clinical decision support systems. [ABSTRACT FROM AUTHOR]

Diabetic retinopathy (DR) is a complication of diabetes that damages the retina and can cause blindness if untreated due to high blood sugar levels. To accurately diagnose and grade DR, it is important to identify retinal lacerations or biomarkers. Optical coherence tomography (OCT) imaging is a commonly used tool by ophthalmologists due to its detailed visualisation of retinal lacerations, which aids in the precise treatment of retinal abnormalities. However, the number of scans obtained daily exceeds the ophthalmologist's capacity to meaningfully analyse them, given the wide range of severe OCT applications and the prevalence of visual disorders. In the past, several research studies have attempted to address this issue using OCT scans. However, none of them have tried to simultaneously perform retinal laceration segmentation and DR grading. To address this problem, we have proposed a new architecture—a cutting-edge decoupled convolutional network consisting of three distinct modules that work together to achieve accurate DR grading based on clinical standards aided by retinal laceration segmentation. Our proposed paper introduces a deep learning framework that leverages dual guidance to improve performance on two related tasks. It was extensively tested using 26,841 multi-vendor scans, four publicly available datasets, and a real-time dataset containing 307 OCT scans from various patients. The results confirmed the effectiveness of our design, with a mean Dice score of 0.88 (4.76% improvement) in retinal laceration segmentation and 98.93% accuracy in DR grading, with an actual positive rate of about 98.46% and a true negative rate of 99.37%. [ABSTRACT FROM AUTHOR]

The use of cloud computing service models is rapidly increasing, but inefficient resource usage in cloud data centers can lead to great energy consumption and costs. To address this issue, plans have been made to allocate resources more efficiently by utilizing live migration of virtual machines (VMs) and consolidating them into a smaller number of physical machines (PMs). Although, selecting a suitable VM for migration is still a significant challenge. One solution is to classify VMs into Delay-Sensitive (Interactive) or Delay-Insensitive classes based on user request patterns, and then select suitable VMs for migration. This selection process can be enabled through workload prediction, which involves predicting and analyzing the workload of VMs as a pre-migration process. A hybrid model based on Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) combination is introduced in this paper for classifying VMs in the Microsoft Azure cloud service. Microsoft Azure Dataset is a labeled dataset and virtual machine workload in this dataset is categorized as Delay-Insensitive or Delay-sensitive (Interactive). However, the samples distribution in this dataset is unbalanced and majority samples are classified as Delay-Insensitive. To address the challenge, Synthetic Minority Oversampling Technique (SMOTE) method is leveraged in this paper. The empirical results showed that their proposed model achieved an accuracy of 95.18%, indicating that it outperformed other existing models. [ABSTRACT FROM AUTHOR]

This study focuses on gender recognition from speech, and the gender classification performances of various machine learning methods such as support vector machines (SVM), k-nearest neighbor (KNN), convolutional neural network (CNN), and random forest (RF) are investigated. In the study, the Turkish subsection of the Common Voice dataset are used. From this data set, 3000 speech data are randomly selected with a homogeneous distribution according to age and gender, and 75% of the data is used in training the models and the remaining 25% is used in testing. The first 40 Mel Frequency Cepstral Coefficients (MFCC) obtained from each speech are averaged on the time axis and then applied as input to the classifiers. In the study, after parameter optimization is performed on the training data set for all models with the grid search method, performance evaluations are made on the test set using the optimum parameters. Experimental results show that in gender classification from speech, CNN is the most prominent classifier with 98.67% accuracy, followed by SVM with 97.33%, kNN with 96.8% and RF with 95.73%. [ABSTRACT FROM AUTHOR]

Deep learning (DL) plays a significant role in several tasks, especially classification and prediction. Classification tasks can be efficiently achieved via convolutional neural networks (CNN) with a huge dataset, while recurrent neural networks (RNN) can perform prediction tasks due to their ability to remember time series data. In this paper, three models have been proposed to certify the evaluation track for classification and prediction tasks associated with four datasets (two for each task). These models are CNN and RNN, which include two models (Long Short Term Memory (LSTM)) and GRU (Gated Recurrent Unit). Each model is employed to work consequently over the two mentioned tasks to draw a road map of deep learning models for a variety of tasks under the control of a unified architecture for each proposed model. [ABSTRACT FROM AUTHOR]

Face recognition has become indispensable in our daily lives as a quick and painless technique of confirming our identities since in the era of wearing face masks the traditional face recognition system may not effectively recognize the concerned person as an important part of the face (mouth, nose, and chin) which makes a substantial contribution to the face recognition process are occluded and partially hidden. The objective of our research is to tackle the challenges of the partially occluded face with a mask by training our custom dataset using those powerful pretrained deep learning models like YOLO and Mask R-CNN which have not been used before for this purpose and to compare which one is outperforming the better results. To this end, models like (YOLOv5, YOLOv7, YOLOv8, and Mask R-CNN) have been employed and trained on the created dataset to check the accuracy and robustness of the occluded face recognition process. In addition, an online dataset such as (mfr2) which contains celebrities, and politicians masked and unmasked faces after expansion with more images has been used. The experimental findings demonstrate that the proposed algorithms give an accurate result, we achieve an accuracy of 97.5% using YOLOv8s, an accuracy of 89.7% using YOLOv7, and an accuracy of 89% using YOLOv5x, while an accuracy of 94.5% using Mask R-CNN. The study concludes that YOLOv8s outperforms the other models in masked face recognition. [ABSTRACT FROM AUTHOR]

Plants are integral to the agriculture industry, profoundly impacting a nation’s economy and environmental stability, with a significant portion of certain countries’ economies reliant on crop production. Much like human health, plants face susceptibility to diseases induced by viruses and bacteria, necessitating careful attention to plant care and disease identification. This study introduces an AI (Artificial Intelligence) model that detects and explains plant diseases through image analysis. The proposed system, distinct from existing detectors, identifies numerous diseases in vegetables and fruits by employing our proposed ensemble learning classifier involving four deep learning models: VGG16, VGG19, ResNet101 V2, and Inception V3, achieving an accuracy exceeding 90%. The reason for using ensemble learning is to obtain accurate predictions. Furthermore, the system sets itself apart by providing explanations for predictions using LIME (Local Interpretable Model-Agnostic Explanations), applied to interpret the predictions of deep learning models. The visualizations generated from multiple methods point to specific pixels’ influence on accurate and incorrect predictions, clearly illustrating the model’s decision-making process. This technique shows areas of the image that contributed positively to the model’s decision, like key regions where the object of interest was most prominent, and areas that added negative values, where irrelevant or misleading features were present. By exploring these features, we gained insights into how the model interprets and prioritizes different aspects of the image during prediction. The study aims to address existing limitations in plant disease detection, offering a comprehensive solution to enhance agricultural practices, foster economic growth, and contribute to environmental sustainability.

The advent of cloud computing and advanced processing technologies has elevated deep learning (DL) as a leading method for hyperspectral imaging (HSI) classification. Classifying crops accurately is vital for generating precise agricultural data to support informed decision-making. This study introduces a DL framework, called HypsLiDNet, tailored for remote sensing activities. This model processes HSI in conjunction with innovative, comprehensive light detection and ranging (LiDAR) data from Hungary to conduct thorough examinations of the Earth's surface. Integrating LiDAR attributes with HSI is anticipated to enhance classification accuracy beyond HSI-only techniques. LiDAR integration provides a significant advantage by adding structural details to spectral data, aiding in the correct identification of objects with similar spectral characteristics but different shapes.The HypsLiDNet method utilizes morphological operations on LiDAR data to extract features indicative of the land's shape and texture. These features are then combined with HSI data through an attention mechanism that selectively highlights key features from both data types, improving the model's accuracy in predictions. This is particularly beneficial for complex environmental assessments, such as distinguishing between plant species. The attention mechanism also refines the feature selection process, prioritizing relevant information, which boosts computational efficiency and reduces the use of resources. Moreover, this method requires a smaller number of training samples. HypsLiDNet showcases its ability to classify with precision by harnessing the combined power of HSI and LiDAR data. Experimental results show a significant improvement in classification outcomes, outperforming traditional machine learning approaches by more than 14% and recent DL techniques by approximately 1%–3%.

Cast-resin transformers are affected by deterioration due to manufacturing defects and continuous load. Studying PD, which is capable of detecting defects or degradation in advance, is important. With the rapid advancement of AI technologies, research on PD classification using CNN models is being actively conducted. However, due to the black box problem, it is impossible to explain the reasoning behind the learning outcomes. Therefore, relying solely on predictive outcomes of learning for PD classification raises issues of reliability. Recent studies in various fields are progressing with the application of XAI to address the black box issue of CNNs, aiming to identify the criteria used for making predictions. However, research on applying XAI in AI-based PD classification is currently insufficient. Therefore, further study on the implementation of XAI is necessary. In this paper, an excellent CNN model was applied to image classification for PD classification of cast-resin transformers, and the grad-cam model was used for XAI. This approach proposes a method for humans to comprehend the rationale behind the learning outcomes. The training data includes artificial defects created in laboratory settings and noise captured in cast-resin transformers using UHF sensors. Our research demonstrated that PD and noise due to defects can be identified with an accuracy of approximately 97%. The reasons for successful and failed results were analyzed through XAI. Consequently, it was observed that the application of XAI to CNN models leads to the construction of a more reliable model.

Infrared small target detection is a challenging task in which many researchers have made lots of achievements. While the performance of single-frame (SF) detection is still limited due to the lack of usage of multiframe (MF) continuous information, many spatial-temporal detection methods have been developed. However, most algorithms need to register the image or feed a set of images as the input. Inputting a batch of group images usually leads to large computations, which heavily affects their real-time processing capability in resource-limited machines. To tackle the problem, we propose a nonlocal multiframe network (NLMF-Net) with only a few additional computations (no more than 0.01 GFLOPs) compared to the SF baseline while achieving significant performance improvements. The NLMF-Net correlates features from grid cells with high confidence between current and past frames. While most background grid cells are removed after the SF processing, the MF feature fusion only focuses on a few potential target grid cells, resulting in high computation efficiency. The proposed vector length similarity module enlarges the difference between different grid cells and the non max similarity suppression further suppresses the backgrounds during the fusion, promoting the MF performance. The NLMF-Net can be readily deployed on Jetson Nano at a speed of 20 FPS for 288 × 384 image processing or Mi Pad 2 with a speed over 35 FPS for 128 × 128 part image processing. Extensive experiments show that our proposed method achieves state-of-the-art performance on three datasets while maintaining high efficiency in a real-time processing manner.

Arrhythmia detection in electrocardiogram (ECG) signals is a vital aspect of cardiovascular health monitoring. Current automated methods for arrhythmia classification often struggle to attain satisfactory performance in the detection of various heart conditions, particularly when dealing with imbalanced datasets. This study introduces a novel deep learning approach for the detection and classification of ECG arrhythmia plot images. Our methodology features a Lightweight Custom Convolutional Neural Network model(LC-CNN), comprising just three convolutional layers and a transfer learning model with MobileNet-V2 architecture that leverages pre-trained features to enhance arrhythmia classification. Data preprocessing of the ECG signals involving noise reduction with a Butterworth filter and precise beat segmentation via R-peak detection, ensure high-quality input for our model. Furthermore, a notable contribution for ECG data augmentation, adopting the implementation of an Auxiliary Classifier Generative Adversarial Network (ACGAN), specifically addressing class imbalance in the benchmark MIT-BIH dataset to classify four types of ECG heartbeats. This approach enriches the dataset, enhancing the models’ ability to detect underrepresented arrhythmia classes. The proposed system demonstrates an impressive average classification accuracy achieving 99.22% using the LC-CNN model, closely followed by the fine-tuned MobileNet-V2 model with 98.69% accuracy, outperforming other methods and underscoring its effectiveness when faced with diverse irregular heartbeats and arrhythmia.

Multispectral (MS) image pansharpening is a significant technology for remote sensing image analysis, aiming to restore a high-resolution MS image by merging a high-resolution panchromatic image with a low-resolution MS image. While the convolutional neural networks (CNNs) have garnered considerable attention for their exceptional fusion capabilities in recent years, the existing CNN-based methods cannot effectively integrate spectral–spatial information. In this article, we introduce a novel MS pansharpening framework that integrates spectral and spatial networks in a bilateral pyramid form, allowing for the extraction of hierarchical spectral–spatial information. The proposed reduced residual dense (RRD) module serves as the fundamental building block of the spatial network. The RRD module gradually reduces the dimension of feature maps and employs the concatenation of RRD with global residual learning for comprehensive feature representation. In the spectral network, we present a cooperative attention fusion module to further enhance the correlation between spectral and spatial features. Through extensive experiments conducted on benchmarking simulated and real datasets, our proposed framework consistently outperforms state-of-the-art methods, demonstrating its effectiveness in MS image pansharpening applications.

In document forensics, identifying ink mismatches is crucial for detecting forgeries and determining document authenticity. However, identifying and separating specific inks from paper can be challenging. Hyperspectral imaging can capture unique spectral patterns exhibited by inks composed of different materials, even when their colors appear identical. Hyperspectral document analysis (HSDI) can be employed to authenticate documents by analyzing the ink. Earlier studies had insufficiently accurate black ink detection, so this paper proposes a novel approach using deep learning and pre-processing methods to improve black ink detection accuracy. The proposed approach uses supervised deep learning to identify ink mismatches in hyperspectral document images. The study evaluated the performance of the proposed model using a hyperspectral image dataset consisting of UWA (University of Western Australia) writing ink in both blue and black colors and different types of artificially identical color inks mixed in various ratios to find ink mismatches. The results showed that the proposed system performed better than other systems reported in the literature, improving the average accuracy by up to 0.18% for blue and 0.36% for black. The proposed method accurately detects ink mismatches and identifies various inks based on their unique spectral response, rendering it highly beneficial for applications in document forensics.

Skin cancer is the most common form of cancer. It is predicted that the total number of cases of cancer will double in the next fifty years. It is an expensive procedure to discover skin cancer types in the early stages. Additionally, the survival rate reduces as cancer progresses. The current study proposes an aseptic approach toward skin lesion detection, classification, and segmentation using deep learning and Harris Hawks Optimization Algorithm (HHO). The current study utilizes the manual and automatic segmentation approaches. The manual segmentation is used when the dataset has no masks to use while the automatic segmentation approach is used, using U-Net models, to build an adaptive segmentation model. Additionally, the meta-heuristic HHO optimizer is utilized to achieve the optimization of the hyperparameters of 5 pre-trained CNN models, namely VGG16, VGG19, DenseNet169, DenseNet201, and MobileNet. Two datasets are used, namely "Melanoma Skin Cancer Dataset of 10000 Images" and "Skin Cancer ISIC" dataset from two publicly available sources for variety purpose. For the segmentation, the best-reported scores are 0.15908, 91.95%, 0.08864, 0.04313, 0.02072, 0.20767 in terms of loss, accuracy, Mean Absolute Error, Mean Squared Error, Mean Squared Logarithmic Error, and Root Mean Squared Error, respectively. For the "Melanoma Skin Cancer Dataset of 10000 Images" dataset, from the applied experiments, the best reported scores are 97.08%, 98.50%, 95.38%, 98.65%, 96.92% in terms of overall accuracy, precision, sensitivity, specificity, and F1-score, respectively by the DenseNet169 pre-trained model. For the "Skin Cancer ISIC" dataset, the best reported scores are 96.06%, 83.05%, 81.05%, 97.93%, 82.03% in terms of overall accuracy, precision, sensitivity, specificity, and F1-score, respectively by the MobileNet pre-trained model. After computing the results, the suggested approach is compared with 9 related studies. The results of comparison proves the efficiency of the proposed framework. [ABSTRACT FROM AUTHOR]