Skin Cancer Detection Using Deep Learning

EasyChair Preprint 13262

Abstract

Skin cancer is one of the most common illnesses worldwide, and early identification is essential for a successful course of therapy and well-tolerated results. This theory of- fers an analysis centred on optimising a robotized framework through the application of cutting-edge artificial intelligence tools to discover skin growths that are cancerous. The suggested framework breaks down dermatoscopic images of skin lesions to assist dermatologists in precise and timely investigation. It highlights the challenges and setbacks associated in diagnosing skin conditions. Biomedical imaging has become a powerful tool in the early diagnosis and detection of skin malignant develop- ment because it provides precise and safe methods for evaluating skin lesions. Many techniques were employed, including optical soundness tomography, dermoscopy, confocal microscopy, and multi-ghostly imaging. However, human mistake or subjective picture interpretation may result in inaccurate diagnosis; hence, medical experts’ experience is essential when analysing biological images. In order to provide a comprehensive examination of the status of profound learning techniques in biomedical imaging for the identification of skin cancers, this extensive assessment intends to be presented. Deep learning algorithms can differen- tiate between frequent changes in skin lesions that may suggest different types of illnesses since they are built to extract discrete levels of highlighting. With pre-prepared models, motion learning is applied to enhance the model’s presentation, particularly in situations when there is minimal marked information. To adapt the model to the specific characteristics of the skin disease dataset, systems of adjustment are used.

Keyphrases: CNN, deep learning, image processing

@booklet{EasyChair:13262,
  author    = {P Siva Prasad and Sk Sham Riyaz and Ch Abhishek Vardhan and K Praveen Kumar},
  title     = {Skin Cancer Detection Using Deep Learning},
  howpublished = {EasyChair Preprint 13262},
  year      = {EasyChair, 2024}}