International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 03 | Mar 2025
p-ISSN: 2395-0072
www.irjet.net
CNN Based Skin Cancer Detection with 3D Data Visualization for Enhanced Diagnosis D. Vigneshwar Reddy1, Vairagade Swapnil 2, Peetla Shiva Krishna3 123 Graduate Student, Department Of Computer Science and Engineering, Vardhaman College Of Engineering
Hyderabad, India. ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This project focuses on enhancing skin cancer
structural complexities of skin lesions. This project combines CNN-based analysis with techniques for 3D data visualization to address this issue.
detection by combining Convolutional Neural Networks (CNNs) with 3D data visualization techniques. The primary objective is to develop a CNN model to accurately classify skin cancer images from a publicly available dataset. We incorporate 3D visualization to analyze and understand the dataset, focusing on features such as age, localization, and diagnosis. By integrating these visual insights with the CNN's performance metrics, the project aims to provide a comprehensive evaluation of how different factors influence model accuracy and effectiveness. The outcome will include a functional CNN model with high classification accuracy, an interactive 3D visualization of the dataset, and a detailed performance analysis report, contributing valuable insights into skin cancer detection and the underlying data characteristics.
Clinicians can get a better idea of the depth, shape, and texture of skin lesions by drawing them in three dimensions. This lets them see patterns that might not show up in 2D images. This improves diagnosis and makes CNN predictions easier to understand. The combination of CNNs and 3D visualization has two advantages: it improves decisionmaking support and improves diagnostic accuracy through deep learning. misdiagnosis can be reduced, unnecessary biopsies can be avoided, and the progression of a disease over time can be better tracked with this strategy. In addition, utilizing 3D visualization makes it simpler to monitor minute shifts in the morphology of the lesion, which can be essential for prompt intervention and early detection. In short, the goal of this project is to create a sophisticated AI-driven diagnostic tool that uses CNN-based classification and 3D visualization to help detect skin cancer. This system can assist dermatologists in making more precise diagnoses by providing a more detailed and comprehensible representation of skin lesions. This will ultimately lead to improved patient outcomes and advancements in the field of medical imaging.
Key Words: 3D Visualization, CNN(Convolutional Neural Networks), Deep Learning, Skin Cancer Detection, Lesion Classification, Medical Image Processing.
I.
INTRODUCTION
Skin cancer is one of the most prevalent and life-threatening diseases globally, with millions of cases diagnosed each year. For effective treatment and increased survival rates, it is essential to detect the problem quickly and accurately. Visual inspections and 2D dermatoscopic imaging are two examples of traditional diagnostic techniques that frequently lack the ability to distinguish between benign and malignant lesions. This has led to an increasing need for advanced computational techniques that can enhance diagnostic accuracy and provide a more detailed understanding of skin lesions.
II. LITERATURE REVIEW Skin cancer detection has gained significant attention in recent years, with deep learning models playing a crucial role in automating diagnostic processes. Various approaches have been explored to enhance the accuracy and efficiency of skin cancer detection, including convolutional neural networks (CNNs), support vector machines (SVMs), and transfer learning techniques. Additionally, 3D visualization has emerged as a promising method for making diagnostic results easier to understand. M. A. The use of the YOLOv8n model for real-time skin lesions detection and classification was investigated by Riyadi et al. [1], demonstrating its potential for accurately distinguishing skin conditions that aren't cancerous from those that are. Similarly, M. I. H. The use of the ImageNet-trained Xception model for binary skin cancer image classification by Abir et al. [2] exemplifies the efficacy of transfer learning in enhancing diagnostic accuracy. Additionally, K. A comparative analysis of deep learning and machine learning models, including CNN and SVM, for
By automating the precise classification of skin lesions, deep learning, particularly Convolutional Neural Networks (CNNs), has revolutionized medical image analysis. CNNs are ideal for dermatological applications because they are made to analyze and learn patterns from image data. This project intends to use CNNs to classify dermatoscopic images into various categories, enabling dermatologists to make more informed decisions, thereby increasing the accuracy of skin cancer detection. However, despite CNNs' effectiveness, conventional 2D imaging methods may not be able to adequately reveal the
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 465