With increased digitization comes an increase in the speed at which threats to the data are emerging. Although it can be challenging to identify, fake image creation doesn't require any particular memory, computational equipment, or hardware. Consequently, this study uses deep learning to achieve accurate detection. In order to improve detection performance, the study strengthened the line separating the background from the object. It also used the adaptive 2D Wiener filter for preprocessing in order to attenuate noise that was unintentionally reinforced throughout the process of improving the image. This essay suggests an Efficient Skip Connections-based Residual Network (ESkip-ResNet) by utilizing skip connections with the Residual Network (ResNet). The ESkip-ResNet architecture also has a number of stages and progressively more leftover blocks to enhance the classification process. ESkip-ResNet uses the remaining blocks of identity mapping through skip connections in the ResNet architecture. Additionally, ESkip-ResNet has effective techniques for downsampling and stable batch normalization layers, which both improve its stable and dependable performance. The Coot Bird Optimization (CBO) method is used to fine-tune the hyper-parameters of the proposed classifier. The suggested model, ESkip-ResNet, was proposed to be more sensible and to offer better performance. The ESkip-ResNet architecture also has a number of stages and progressively more leftover blocks to enhance the classification process. The proposed model achieved 98.9% and 98.8% accuracy and precision, respectively. Comprehensive test results demonstrate that CBO-based ESkip-ResNet outperforms other approaches in fake detection. The proposed research also took into account every kind of facial alteration, improving the model's robustness, lightweight nature, and generalizability. It was able to identify every type of facial alteration found in images taken from the Deepfake Detection Challenge dataset. Received: 28 March 2024 | Revised: 10 May 2024 | Accepted: 28 May 2024 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The DFDC data that support the findings of this study are openly available at https://doi.org/10.48550/arXiv.2102.11126, reference number [20]. The Deepfake Detection Challenge datasets that support the findings of this study are openly available at https:// www.kaggle.com/c/deepfake-detection-challenge. Author Contribution Statement V. Gokula Krishnan: Conceptualization, Methodology, Writing – original draft, Writing – review & editing, Project administration. R. Vadivel: Software, Visualization. K. Sankar: Validation, Resources. K. Sathyamoorthy: Formal analysis, Data curation. B. Prathusha Laxmi: Investigation, Supervision.

Artificial intelligence (AI) and machine learning (ML) are transforming medical diagnostics, but human nail, an easily accessible and rich biological substrate, is still not fully exploited in the digital health field. Nail pathologies are easily diagnosed, non-invasive disease biomarkers, including systemic diseases such as anemia, diabetes, psoriasis, melanoma, and fungal diseases. This review presents the first big synthesis of image analysis for nail lesions incorporating AI/ML for diagnostic purposes. Where dermatological reviews to date have been more wide-ranging in scope, our review will focus specifically on diagnosis and screening related to nails. The various technological modalities involved (smartphone imaging, dermoscopy, Optical Coherence Tomography) will be presented, together with the different processing techniques for images (color corrections, segmentation, cropping of regions of interest), and models that range from classical methods to deep learning, with annotated descriptions of each. There will also be additional descriptions of AI applications related to some diseases, together with analytical discussions regarding real-world impediments to clinical application, including scarcity of data, variations in skin type, annotation errors, and other laws of clinical adoption. Some emerging solutions will also be emphasized: explainable AI (XAI), federated learning, and platform diagnostics allied with smartphones. Bridging the gap between clinical dermatology, artificial intelligence and mobile health, this review consolidates our existing knowledge and charts a path through yet others to scalable, equitable, and trustworthy nail based medically diagnostic techniques. Our findings advocate for interdisciplinary innovation to bring AI-enabled nail analysis from lab prototypes to routine healthcare and global screening initiatives.

Deep learning-based automated classification of dermatoscopic skin lesions has exhibited promising potential in diagnostics. However, two prominent issues need to be addressed before achieving high-quality diagnostic tools: inconsistent performance in the case of imbalanced classes and poor clinical interpretability of models. Even though some studies have attempted to leverage both deep and shallow learning by combining pretrained convolutional neural networks (CNNs)-based feature extraction with classical machine learning (ML) models, very few of them systematically explore several model combinations based on various clinically important metrics, such as F1-score, precision, recall, accuracy, etc., and utilize decision threshold calibration techniques. In this research, we present an evaluation of a systematic framework with threshold calibration for the comparison of several hybrid models on seven-class skin lesion classification (multi-class) on the HAM10000 dataset. In particular, we used deep features extracted from three pretrained CNN architectures, i.e., DenseNet201, InceptionV3 and EfficientNet-B4. These deep features were used as inputs for six different classical classifiers. As a result, we obtained 18 comparable hybrid models that were then systematically compared by multiple clinically relevant metrics: accuracy, macro-precision, macro-recall, macro-F1, ROC-AUC, Precision-Recall-AUC, and log loss. Also, fold-wise optimization of decision thresholds was performed, which was based on the maximization of the macro-F1 score. Finally, we found out that DenseNet201 with an SVM-RBF classifier yielded the highest performance among all 18 tested models, showing 90.88% accuracy, 90.7% macro-precision, and 0.921 ROC-AUC. To analyze the clinical plausibility, top-performing models were further explained with explainable artificial intelligence (XAI) techniques: Grad-CAM, LIME and Occlusion Sensitivity. Results show that the most successful models concentrated mostly on lesion-specific areas. Overall, this study contributes a reproducible hybrid-XAI model-selection framework rather than a single black-box classifier, supporting more transparent and clinically meaningful skin lesion diagnosis.

Effective trash management has become a top environmental priority, especially in urban areas with significant population growth where garbage output is on the rise. As cities work to manage garbage properly, innovative waste management programmes have the potential to increase effectiveness, cut costs, and improve the aesthetic appeal of public places. This article introduces SCM-RIAN, a powerful "Smart City Management and Classification System" built on the Internet of Things (IoT) and deep learning (DL) technologies. Convolutional neural networks are used in the garbage classification model that is implemented within this smart city management and classification framework. This system for classifying waste is intended to categorise rubbish into several classes at waste collection sites, encouraging recycling. The Rotation-Invariant Attention Network (RIAN) is a unique approach presented for the categorization process to address a prevalent problem in smart city management (SCM). A Centre Spectral Attention (CSpeA) module built within RIAN isolates spectral bands from other categories of pixels' influence, reducing redundancy. As an alternative to the conventional 3 3 convolution, to obtain rotation-invariant spectral-spatial data contained in SCM patches, the Rectified Spatial Attention (RSpaA) module is also introduced. The suggested RIAN for SCM classification is built on the integration of the CSpeA, 11 convolution, and RSpaA modules. The Ladybird Beetle Optimisation Algorithm (LBBOA) is used to optimise hyperparameters. With improved results compared to other current models, this suggested SCM-RIAN achieved 98.12% accuracy (ACC) with high sensitivity (SEN), specificity (SPEC), and kappa index (KI) using the garbage classification dataset.

Early detection of melanoma skin cancer (MSC) is critical in order to prevent deaths from fatal skin cancer. Even though the modern research methods are effective in identifying and detecting skin cancer, it is a challenging task due to a higher level of color similarity between melanoma non-affected areas and affected areas, and a lower contrast between the skin portions and melanoma moles. For highlighting the aforementioned problems, an efficient automated system is proposed for an early diagnosis of MSC. Firstly, dermoscopic images are collected from two benchmark datasets namely, international skin imaging collaboration (ISIC)-2017 and PH2. Next, skin lesions are segmented from dermoscopic images by implementing a fuzzy based SegNet model which is a combination of both deep fuzzy clustering algorithm and the SegNet model. Then, hybrid feature extraction (ResNet-50 model and local tri-directional pattern (LTriDP) descriptor) is performed to capture the features from segmented skin lesions. These features are given into the normalized stacked long short-term memory (LSTM) network to categorize the classes of skin lesions. The empirical evaluation reveals that the proposed normalized stacked LSTM network achieves 98.98% and 98.97% of accuracy respectively on the ISIC2017 and PH2 datasets, and these outcomes are more impressive than those of the conventional detection models.