Recent Advances in Computer Science and Communications - Volume 19, Issue 1, 2026

Basal Cell Carcinoma (BCC) is the most prevalent type of skin cancer, accounting for three-quarters of all cancer cases. It is often confused with other benign lesions, such as Acne Vulgaris (AV). In this paper, we propose a classification approach to discriminate BCC from AV in dermoscopic images, using an algorithm based on deep learning techniques.

A two-branch Convolutional Neural Network (CNN) is employed to construct the model. The first branch consists of CNN structures that process an RGB dermoscopic image, while the second branch leverages the pre-trained ResNet18 network with an HSV dermoscopic image input. Both branches use Principal Component Analysis (PCA), normalization, and image resizing. This concatenated architecture enables the system to exploit features extracted from both color spaces as well as CNN networks, enhancing the overall performance of the model.

The proposed architecture is assessed using two public datasets. The first one is dedicated to the binary classification of Basal Cell Carcinoma (BCC) versus Acne Vulgaris (AV), achieving an accuracy of 99.06%, a sensitivity of 98.73%, and a specificity of 99.37%. The second dataset addresses multiclass classification along melanoma, BCC, and AV, and achieves an accuracy of 97.17%, a sensitivity of 97.13%, and a specificity of 98.57%. The results highlight the effectiveness of the proposed model.

The dual-input hybrid algorithm, based on convolutional neural networks and incorporating principal component analysis, demonstrates promising results in distinguishing BCC from AV.

Nowadays, Artificial intelligence and machine learning have emerged as a powerful tool for the analysis of medical images such as MRI scans. This technology holds significant potential to improve diagnostic services and accelerate medical advances by facilitating clinical decision-making.

In this work, we developed a Convolutional Neural Network (CNN) model specifically designed for the classification of medical images. Using a selected database, the model achieved a classification accuracy of 92%. To further improve the performance, we leveraged the pre-trained VGG16 model, which increased the classification accuracy to 100%. Additionally, we preprocessed the MRI images using the Roboflow platform and then developed YOLOv5 models for the detection of tumors, infections, and cancerous lesions.

The results demonstrate a localization accuracy of 50.41% for these medical conditions.

This research highlights the value of AI-driven approaches in enhancing medical image analysis and their potential to support more accurate diagnoses and accelerate advancements in healthcare.

The growing demand for cloud computing services necessitates innovative strategies to enhance cloud deployment efficiency. Existing cloud load balancing models often grapple with suboptimal resource allocation, leading to increased task makespan, lowered virtual machine (VM) efficiency, and failure to meet task deadlines effectively.

Addressing these limitations, this study aims to introduce a novel model that significantly improves cloud deployment efficiency through a strategic blend of pre-emptive load analysis and resource pre-allocation operations.

The proposed model hinges on the innovative use of Grey Wolf-based TOPSIS (GWTOPSIS) operations for the initial segregation of VMs. This approach considers VM capacity, current load, and availability levels, dynamically modifying internal weights to adapt VM categories based on fluctuating demand. The novelty of GWTOPSIS lies in its extension of traditional TOPSIS methods, allowing for more responsive and demand-aligned VM categorization. Furthermore, the clustering of new tasks based on makespan, deadline, and resource utilization levels-with a higher preference for makespan-addresses critical task scheduling challenges. A pivotal aspect of our model is the integration of the Bacterial Foraging Firefly Optimizer (BFFO) for task assignment to VMs. This optimizer synergizes the exploratory prowess of bacterial foraging with the efficient search capabilities of fireflies, leading to a more effective task-to-VM assignment process.

Empirical evaluations of our model reveal substantial improvements over existing models: a reduction in task makespan by 8.5%, a 3.9% boost in VM computation efficiency, a 1.9% enhancement in deadline hit ratio, a 3.5% increase in task diversity, a 2.4% improvement in execution efficiency, and a 3.5% reduction in decision delay.

These improvements not only demonstrate the efficacy of our model in real-time cloud deployment scenarios but also underscore its potential to revolutionize cloud resource management.

Deformable image registration is an essential task in medical image analysis. The UNet model, or the model with the U-shaped structure, has been popularly proposed in deep learning-based registration methods. However, they easily lose the important similarity information in the up-sampling stage, and these methods usually ignore the inherent inverse consistency of the transformation between a pair of images. Furthermore, the traditional smoothing constraints used in the existing methods can only partially ensure the folding of the deformation field.

An inverse consistent deformable medical image registration network (ICSANet) based on the inverse consistency constraint and the similarity-based local attention is developed. A new UNet network is constructed by introducing similarity-based local attention to focus on the spatial correspondence in the high-similarity space. A novel inverse consistency constraint is proposed, and the objective function of the new form is presented with the combination of the traditional constraint conditions.

The performance of the proposed method is compared with the typical registration models, such as the VoxelMorph, PVT, nnFormer, and TransMorph-diff model, on the brain IXI and OASIS datasets.

Experimental results on the brain MRI datasets show that the images can be deformed symmetrically until two distorted images are well matched. The quantitative comparison and visual analysis indicate that the proposed method performs better, and the Dice index can be improved by at least 12% with only 10% parameters.

This paper presents a new medical image registration network, ICSANet. By introducing a similarity attention gate, it accurately captures high-similarity spatial correspondences between source and target images, resulting in better registration performance.