Browsing by Author "Syeda Tehzeeb Zahra and Fahad Zubair"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    SKIN CANCER MULTI-CLASS CLASSIFICATION USING TRANSFER LEARNING TECHNIQUES
    (UMT, Lahore, 2023) Syeda Tehzeeb Zahra and Fahad Zubair
    Skin cancer is a significant global health concern, and early detection is crucial for successful treatment. This research presents a comprehensive study on the application of deep learning techniques for the automated classification of skin lesions using demo scopic images. The study leverages state-of-the-art convolutional neural networks (CNNs) and explores various methodologies for skin lesion segmentation and classification. The HAM10000 skin cancer dataset, a valuable resource in dermatology, is employed as the primary dataset for experimentation. The research addresses class imbalance issues through data augmentation techniques, self-attention-based generative adversarial networks (GANs), and multi-label deep neural networks. Transfer learning is also explored, with models like InceptionResNetV2 and Efficient Nets, to harness the knowledge of pretrained networks. The study encompasses several faces of skin cancer identification, including multi-class classification, lesion segmentation, and feature extraction. GradCAM visualization is applied to elucidate regions contributing to classification decisions. Ethical considerations concerning medical image data and patient privacy are meticulously observed. The outcomes demonstrate a deep learning model with an impressive accuracy of 89% in classifying various skin conditions, signifying its potential for early skin cancer detection. Precision, recall, F1-score, and support metrics are analyzed for individual classes. The study not only contributes to the growing body of research in dermatological image analysis but also provides valuable insights and methodologies for advancing the field of skin cancer diagnosis using deep learning. This research serves as a foundation for further exploration and development of computer-aided diagnosis systems for skin cancer, offering a significant stride towards more accessible and accurate early detection methods