Skin Cancer Classification using Hierarchical Neural Networks and Fuzzy Systems


Skin cancers of all types comprise the largest group of cancers in the United States. This group accounts for 1% of all cancer deaths. Fortunately, even the deadliest form of skin cancer can be treated successfully if detected in its early stage. Therefore, automated detection and diagnosis of skin cancer has become an important issue in recent years. In this study, we present a diagnostic-tree-based hierarchical neural network system, which is integrated with a fuzzy system for the classification of four classes of skin tumors. These classes are: malignant melanoma, atypical mole, basal cell carcinoma or actinic keratosis, and intradermal nevus or seborrheic keratosis. The hierarchical neural network system is the integration of three distinct neural networks, which are trained separately, using the backpropagation learning algorithm. In this approach, for a given tumor image, the hierarchical neural network system produces a crisp outcome, suggesting a category to which the tumor belongs; whereas the fuzzy system generates four fuzzy outcomes showing the likelihood of belonging to each of the four tumor classes. In our study, we also compare the performance of the hierarchical neural network to the results obtained from a single neural network, which attempts to classify all four categories at once.


Computer Science

Second Department

Nuclear Engineering and Radiation Science

Third Department


Fourth Department

Electrical and Computer Engineering

International Standard Serial Number (ISSN)


Document Type

Article - Journal

Document Version


File Type





© 2024 The Authors, All rights reserved.

Publication Date

01 Dec 1998

This document is currently not available here.