Authors:
Vijilius Helena Raj, B. Manjunatha, M. Arunadevi Thirumalraj, Thirumalraj Karthikeyan, Sheila Agnes Vidot
Addresses:
Department of Mathematics, New Horizon College of Engineering, Bengaluru, Karnataka, India. Department of Computer Science and Engineering, Karunya Institute of Technology and Science, Coimbatore, Tamil Nadu, India. Department of Computer Science and Business Management, Saranathan College of Engineering, Tiruchirappalli, Tamil Nadu, India. Department of Artificial Intelligence, Trichy Research Labs, Quest Technologies, Tiruchirappalli, Tamil Nadu, India. Department of Medical Consultant, IPS Health, Mahé, Victoria, Seychelles.
Abstract:
Skin cancer, often known as SC, is one of the most common types of cancer. Visual inspections account for most examinations, with clinical examination, histology, biopsy, and dermatological phases optional. The purpose of this study is to investigate the successful diagnosis of multiple SC types using modern pipelines on publicly available data. An important stage in the preprocessing process that lays the groundwork for improved image quality is histogram equalisation. A fully connected pyramid network (FC-PRNet) is used for subsequent segmentation, improving the accuracy of local detection. To adjust hyperparameters during segmentation, the new architecture uses a subtraction-average-based optimiser, also known as SABO. For classification, an architecture known as ECA-Net-VGG16 has been introduced. This architecture is an efficient channel attention (ECA)-infused visual geometry group (VGG16). A robust design that delivers strong SC detection performance and promises improvements in classification accuracy results from the synergy of these qualities, ultimately culminating in the design. With a Jaccard Index (JI) of 99.12% and a Dice Similarity Coefficient (DSC) of 99.25%, the proposed SABO-based FC-PRNet segmentation model achieves an amazing accuracy of 99.45%.
Keywords: Fully Convolutional Pyramidal Networks; Histogram Equalisation; Subtraction-Average-Based Optimiser; Visual Geometry Group; Efficient Channel Attention; Dice Similarity Coefficient.
Received: 30/09/2024, Revised: 17/01/2025, Accepted: 25/03/2025, Published: 07/09/2025
DOI: 10.64091/ATIHL.2025.000171
AVE Trends in Intelligent Health Letters, 2025 Vol. 2 No. 3 , Pages: 117-132